| model | size | params | backend | ngl | n_batch | n_ubatch | type_k | type_v | sm | ts | test | t/s | | ------------------------------ | ---------: | ---------: | ---------- | --: | ------: | -------: | -----: | -----: | ----: | ------------ | ------------: | ---------------: | Device 0: 48.875 MiB Device 1: 0 MiB Device 2: 0 MiB Device 3: 0 MiB GET_ROWS(inp_embd): 43 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 3932 us MUL_MAT_ID(ffn_moe_down-25): 384 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 685 us MUL_MAT_ID(ffn_moe_down-26): 347 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 689 us MUL_MAT_ID(ffn_moe_down-27): 344 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 667 us MUL_MAT_ID(ffn_moe_down-28): 368 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 605 us MUL_MAT_ID(ffn_moe_down-29): 397 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 644 us MUL_MAT_ID(ffn_moe_down-30): 384 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 692 us MUL_MAT_ID(ffn_moe_down-31): 342 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 695 us MUL_MAT_ID(ffn_moe_down-32): 331 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 634 us MUL_MAT_ID(ffn_moe_down-33): 370 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 633 us MUL_MAT_ID(ffn_moe_down-34): 304 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 699 us MUL_MAT_ID(ffn_moe_down-35): 399 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 640 us MUL_MAT_ID(ffn_moe_down-36): 377 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 614 us MUL_MAT_ID(ffn_moe_down-37): 563 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 668 us MUL_MAT_ID(ffn_moe_down-38): 347 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 714 us MUL_MAT_ID(ffn_moe_down-39): 422 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 715 us MUL_MAT_ID(ffn_moe_down-40): 370 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 693 us MUL_MAT_ID(ffn_moe_down-41): 390 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 664 us MUL_MAT_ID(ffn_moe_down-42): 362 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 648 us MUL_MAT_ID(ffn_moe_down-43): 352 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 604 us MUL_MAT_ID(ffn_moe_down-44): 377 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 697 us MUL_MAT_ID(ffn_moe_down-45): 332 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 678 us MUL_MAT_ID(ffn_moe_down-46): 330 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 649 us MUL_MAT_ID(ffn_moe_down-47): 351 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 639 us MUL_MAT_ID(ffn_moe_down-48): 351 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 652 us MUL_MAT_ID(ffn_moe_down-49): 286 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 640 us MUL_MAT_ID(ffn_moe_down-50): 339 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 705 us MUL_MAT_ID(ffn_moe_down-51): 332 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 697 us MUL_MAT_ID(ffn_moe_down-52): 367 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 700 us MUL_MAT_ID(ffn_moe_down-53): 290 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 704 us MUL_MAT_ID(ffn_moe_down-54): 312 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 727 us MUL_MAT_ID(ffn_moe_down-55): 353 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 662 us MUL_MAT_ID(ffn_moe_down-56): 394 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 631 us MUL_MAT_ID(ffn_moe_down-57): 406 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 603 us MUL_MAT_ID(ffn_moe_down-58): 342 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 670 us MUL_MAT_ID(ffn_moe_down-59): 393 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 641 us MUL_MAT_ID(ffn_moe_down-60): 409 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 600 us MUL_MAT_ID(ffn_moe_down-61): 388 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 674 us MUL_MAT_ID(ffn_moe_down-62): 425 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 700 us MUL_MAT_ID(ffn_moe_down-63): 408 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 640 us MUL_MAT_ID(ffn_moe_down-64): 391 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 611 us MUL_MAT_ID(ffn_moe_down-65): 397 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 588 us MUL_MAT_ID(ffn_moe_down-66): 386 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 678 us MUL_MAT_ID(ffn_moe_down-67): 376 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 708 us MUL_MAT_ID(ffn_moe_down-68): 336 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 652 us MUL_MAT_ID(ffn_moe_down-69): 341 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 696 us MUL_MAT_ID(ffn_moe_down-70): 408 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 656 us MUL_MAT_ID(ffn_moe_down-71): 406 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 645 us MUL_MAT_ID(ffn_moe_down-72): 382 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 775 us MUL_MAT_ID(ffn_moe_down-73): 355 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 622 us MUL_MAT_ID(ffn_moe_down-74): 396 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 687 us MUL_MAT_ID(ffn_moe_down-75): 346 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 575 us MUL_MAT_ID(ffn_moe_down-76): 361 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 728 us MUL_MAT_ID(ffn_moe_down-77): 311 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 645 us MUL_MAT_ID(ffn_moe_down-78): 417 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 681 us MUL_MAT_ID(ffn_moe_down-79): 303 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 703 us MUL_MAT_ID(ffn_moe_down-80): 314 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 640 us MUL_MAT_ID(ffn_moe_down-81): 403 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 779 us MUL_MAT_ID(ffn_moe_down-82): 356 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 673 us MUL_MAT_ID(ffn_moe_down-83): 297 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 650 us MUL_MAT_ID(ffn_moe_down-84): 378 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 685 us MUL_MAT_ID(ffn_moe_down-85): 373 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 671 us MUL_MAT_ID(ffn_moe_down-86): 330 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 652 us MUL_MAT_ID(ffn_moe_down-87): 334 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 672 us MUL_MAT_ID(ffn_moe_down-88): 326 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 590 us MUL_MAT_ID(ffn_moe_down-89): 379 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 606 us MUL_MAT_ID(ffn_moe_down-90): 413 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 664 us MUL_MAT_ID(ffn_moe_down-91): 437 us ggml_barrier(...): 73 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 576 us MUL_MAT_ID(ffn_moe_down-25): 367 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 680 us MUL_MAT_ID(ffn_moe_down-26): 421 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 593 us MUL_MAT_ID(ffn_moe_down-27): 380 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 642 us MUL_MAT_ID(ffn_moe_down-28): 343 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 693 us MUL_MAT_ID(ffn_moe_down-29): 307 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 674 us MUL_MAT_ID(ffn_moe_down-30): 337 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 662 us MUL_MAT_ID(ffn_moe_down-31): 396 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 635 us MUL_MAT_ID(ffn_moe_down-32): 366 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 601 us MUL_MAT_ID(ffn_moe_down-33): 402 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 679 us MUL_MAT_ID(ffn_moe_down-34): 427 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 705 us MUL_MAT_ID(ffn_moe_down-35): 339 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 586 us MUL_MAT_ID(ffn_moe_down-36): 431 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 670 us MUL_MAT_ID(ffn_moe_down-37): 418 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 662 us MUL_MAT_ID(ffn_moe_down-38): 401 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 629 us MUL_MAT_ID(ffn_moe_down-39): 363 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 645 us MUL_MAT_ID(ffn_moe_down-40): 401 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 704 us MUL_MAT_ID(ffn_moe_down-41): 319 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 716 us MUL_MAT_ID(ffn_moe_down-42): 297 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 713 us MUL_MAT_ID(ffn_moe_down-43): 332 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 702 us MUL_MAT_ID(ffn_moe_down-44): 316 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 602 us MUL_MAT_ID(ffn_moe_down-45): 400 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 678 us MUL_MAT_ID(ffn_moe_down-46): 353 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 709 us MUL_MAT_ID(ffn_moe_down-47): 344 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 733 us MUL_MAT_ID(ffn_moe_down-48): 292 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 638 us MUL_MAT_ID(ffn_moe_down-49): 326 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 584 us MUL_MAT_ID(ffn_moe_down-50): 388 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 567 us MUL_MAT_ID(ffn_moe_down-51): 370 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 660 us MUL_MAT_ID(ffn_moe_down-52): 367 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 716 us MUL_MAT_ID(ffn_moe_down-53): 343 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 623 us MUL_MAT_ID(ffn_moe_down-54): 322 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 683 us MUL_MAT_ID(ffn_moe_down-55): 327 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 648 us MUL_MAT_ID(ffn_moe_down-56): 413 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 559 us MUL_MAT_ID(ffn_moe_down-57): 377 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 621 us MUL_MAT_ID(ffn_moe_down-58): 382 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 649 us MUL_MAT_ID(ffn_moe_down-59): 293 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 737 us MUL_MAT_ID(ffn_moe_down-60): 338 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 645 us MUL_MAT_ID(ffn_moe_down-61): 402 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 667 us MUL_MAT_ID(ffn_moe_down-62): 390 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 618 us MUL_MAT_ID(ffn_moe_down-63): 380 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 701 us MUL_MAT_ID(ffn_moe_down-64): 308 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 611 us MUL_MAT_ID(ffn_moe_down-65): 324 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 699 us MUL_MAT_ID(ffn_moe_down-66): 310 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 683 us MUL_MAT_ID(ffn_moe_down-67): 344 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 658 us MUL_MAT_ID(ffn_moe_down-68): 387 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 711 us MUL_MAT_ID(ffn_moe_down-69): 448 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 586 us MUL_MAT_ID(ffn_moe_down-70): 369 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 623 us MUL_MAT_ID(ffn_moe_down-71): 330 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 620 us MUL_MAT_ID(ffn_moe_down-72): 360 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 685 us MUL_MAT_ID(ffn_moe_down-73): 317 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 655 us MUL_MAT_ID(ffn_moe_down-74): 374 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 591 us MUL_MAT_ID(ffn_moe_down-75): 384 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 656 us MUL_MAT_ID(ffn_moe_down-76): 317 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 636 us MUL_MAT_ID(ffn_moe_down-77): 407 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 614 us MUL_MAT_ID(ffn_moe_down-78): 345 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 620 us MUL_MAT_ID(ffn_moe_down-79): 302 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 663 us MUL_MAT_ID(ffn_moe_down-80): 304 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 643 us MUL_MAT_ID(ffn_moe_down-81): 311 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 706 us MUL_MAT_ID(ffn_moe_down-82): 328 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 597 us MUL_MAT_ID(ffn_moe_down-83): 352 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 661 us MUL_MAT_ID(ffn_moe_down-84): 419 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 605 us MUL_MAT_ID(ffn_moe_down-85): 372 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 661 us MUL_MAT_ID(ffn_moe_down-86): 276 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 692 us MUL_MAT_ID(ffn_moe_down-87): 329 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 655 us MUL_MAT_ID(ffn_moe_down-88): 386 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 565 us MUL_MAT_ID(ffn_moe_down-89): 358 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 700 us MUL_MAT_ID(ffn_moe_down-90): 345 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 634 us MUL_MAT_ID(ffn_moe_down-91): 342 us ggml_barrier(...): 159 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 608 us MUL_MAT_ID(ffn_moe_down-25): 362 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 553 us MUL_MAT_ID(ffn_moe_down-26): 300 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 559 us MUL_MAT_ID(ffn_moe_down-27): 364 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 662 us MUL_MAT_ID(ffn_moe_down-28): 294 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 580 us MUL_MAT_ID(ffn_moe_down-29): 270 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 536 us MUL_MAT_ID(ffn_moe_down-30): 292 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 519 us MUL_MAT_ID(ffn_moe_down-31): 310 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 573 us MUL_MAT_ID(ffn_moe_down-32): 334 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 617 us MUL_MAT_ID(ffn_moe_down-33): 263 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 649 us MUL_MAT_ID(ffn_moe_down-34): 330 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 576 us MUL_MAT_ID(ffn_moe_down-35): 298 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 601 us MUL_MAT_ID(ffn_moe_down-36): 282 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 607 us MUL_MAT_ID(ffn_moe_down-37): 406 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 554 us MUL_MAT_ID(ffn_moe_down-38): 583 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 528 us MUL_MAT_ID(ffn_moe_down-39): 374 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 602 us MUL_MAT_ID(ffn_moe_down-40): 374 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 641 us MUL_MAT_ID(ffn_moe_down-41): 304 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 694 us MUL_MAT_ID(ffn_moe_down-42): 274 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 656 us MUL_MAT_ID(ffn_moe_down-43): 295 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 580 us MUL_MAT_ID(ffn_moe_down-44): 254 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 566 us MUL_MAT_ID(ffn_moe_down-45): 307 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 630 us MUL_MAT_ID(ffn_moe_down-46): 316 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 650 us MUL_MAT_ID(ffn_moe_down-47): 259 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 660 us MUL_MAT_ID(ffn_moe_down-48): 293 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 543 us MUL_MAT_ID(ffn_moe_down-49): 263 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 600 us MUL_MAT_ID(ffn_moe_down-50): 322 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 597 us MUL_MAT_ID(ffn_moe_down-51): 347 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 677 us MUL_MAT_ID(ffn_moe_down-52): 363 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 687 us MUL_MAT_ID(ffn_moe_down-53): 299 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 659 us MUL_MAT_ID(ffn_moe_down-54): 303 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 672 us MUL_MAT_ID(ffn_moe_down-55): 304 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 604 us MUL_MAT_ID(ffn_moe_down-56): 334 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 626 us MUL_MAT_ID(ffn_moe_down-57): 340 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 735 us MUL_MAT_ID(ffn_moe_down-58): 284 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 575 us MUL_MAT_ID(ffn_moe_down-59): 410 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 636 us MUL_MAT_ID(ffn_moe_down-60): 291 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 679 us MUL_MAT_ID(ffn_moe_down-61): 426 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 593 us MUL_MAT_ID(ffn_moe_down-62): 295 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 571 us MUL_MAT_ID(ffn_moe_down-63): 387 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 665 us MUL_MAT_ID(ffn_moe_down-64): 373 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 600 us MUL_MAT_ID(ffn_moe_down-65): 319 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 619 us MUL_MAT_ID(ffn_moe_down-66): 828 us ggml_barrier(...): 401 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 572 us MUL_MAT_ID(ffn_moe_down-67): 323 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 647 us MUL_MAT_ID(ffn_moe_down-68): 389 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 638 us MUL_MAT_ID(ffn_moe_down-69): 290 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 657 us MUL_MAT_ID(ffn_moe_down-70): 331 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 647 us MUL_MAT_ID(ffn_moe_down-71): 286 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 695 us MUL_MAT_ID(ffn_moe_down-72): 297 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 598 us MUL_MAT_ID(ffn_moe_down-73): 336 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 524 us MUL_MAT_ID(ffn_moe_down-74): 345 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 639 us MUL_MAT_ID(ffn_moe_down-75): 264 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 602 us MUL_MAT_ID(ffn_moe_down-76): 284 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 641 us MUL_MAT_ID(ffn_moe_down-77): 269 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 596 us MUL_MAT_ID(ffn_moe_down-78): 315 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 607 us MUL_MAT_ID(ffn_moe_down-79): 325 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 636 us MUL_MAT_ID(ffn_moe_down-80): 344 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 683 us MUL_MAT_ID(ffn_moe_down-81): 300 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 548 us MUL_MAT_ID(ffn_moe_down-82): 402 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 577 us MUL_MAT_ID(ffn_moe_down-83): 265 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 566 us MUL_MAT_ID(ffn_moe_down-84): 332 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 552 us MUL_MAT_ID(ffn_moe_down-85): 364 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 675 us MUL_MAT_ID(ffn_moe_down-86): 346 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 522 us MUL_MAT_ID(ffn_moe_down-87): 320 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 548 us MUL_MAT_ID(ffn_moe_down-88): 291 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 615 us MUL_MAT_ID(ffn_moe_down-89): 427 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 577 us MUL_MAT_ID(ffn_moe_down-90): 297 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 565 us MUL_MAT_ID(ffn_moe_down-91): 330 us ggml_barrier(...): 162 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 575 us MUL_MAT_ID(ffn_moe_down-25): 374 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 543 us MUL_MAT_ID(ffn_moe_down-26): 312 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 639 us MUL_MAT_ID(ffn_moe_down-27): 279 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 508 us MUL_MAT_ID(ffn_moe_down-28): 359 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 532 us MUL_MAT_ID(ffn_moe_down-29): 342 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 637 us MUL_MAT_ID(ffn_moe_down-30): 292 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 596 us MUL_MAT_ID(ffn_moe_down-31): 292 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 613 us MUL_MAT_ID(ffn_moe_down-32): 261 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 594 us MUL_MAT_ID(ffn_moe_down-33): 303 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 564 us MUL_MAT_ID(ffn_moe_down-34): 357 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 542 us MUL_MAT_ID(ffn_moe_down-35): 325 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 584 us MUL_MAT_ID(ffn_moe_down-36): 313 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 577 us MUL_MAT_ID(ffn_moe_down-37): 285 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 537 us MUL_MAT_ID(ffn_moe_down-38): 279 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 576 us MUL_MAT_ID(ffn_moe_down-39): 269 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 559 us MUL_MAT_ID(ffn_moe_down-40): 292 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 577 us MUL_MAT_ID(ffn_moe_down-41): 274 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 569 us MUL_MAT_ID(ffn_moe_down-42): 279 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 614 us MUL_MAT_ID(ffn_moe_down-43): 268 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 665 us MUL_MAT_ID(ffn_moe_down-44): 367 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 527 us MUL_MAT_ID(ffn_moe_down-45): 324 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 579 us MUL_MAT_ID(ffn_moe_down-46): 344 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 567 us MUL_MAT_ID(ffn_moe_down-47): 257 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 670 us MUL_MAT_ID(ffn_moe_down-48): 358 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 589 us MUL_MAT_ID(ffn_moe_down-49): 311 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 638 us MUL_MAT_ID(ffn_moe_down-50): 287 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 506 us MUL_MAT_ID(ffn_moe_down-51): 307 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 592 us MUL_MAT_ID(ffn_moe_down-52): 357 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 562 us MUL_MAT_ID(ffn_moe_down-53): 356 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 566 us MUL_MAT_ID(ffn_moe_down-54): 366 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 558 us MUL_MAT_ID(ffn_moe_down-55): 345 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 478 us MUL_MAT_ID(ffn_moe_down-56): 334 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 441 us MUL_MAT_ID(ffn_moe_down-57): 209 us ggml_barrier(...): 365 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 506 us MUL_MAT_ID(ffn_moe_down-58): 266 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 626 us MUL_MAT_ID(ffn_moe_down-59): 334 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 1363 us MUL_MAT_ID(ffn_moe_down-60): 462 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 872 us MUL_MAT_ID(ffn_moe_down-61): 489 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 737 us MUL_MAT_ID(ffn_moe_down-62): 431 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 764 us MUL_MAT_ID(ffn_moe_down-63): 491 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 640 us MUL_MAT_ID(ffn_moe_down-64): 412 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 495 us MUL_MAT_ID(ffn_moe_down-65): 237 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 476 us MUL_MAT_ID(ffn_moe_down-66): 319 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 518 us MUL_MAT_ID(ffn_moe_down-67): 232 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 575 us MUL_MAT_ID(ffn_moe_down-68): 420 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 564 us MUL_MAT_ID(ffn_moe_down-69): 387 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 745 us MUL_MAT_ID(ffn_moe_down-70): 255 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 789 us MUL_MAT_ID(ffn_moe_down-71): 513 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 773 us MUL_MAT_ID(ffn_moe_down-72): 495 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 788 us MUL_MAT_ID(ffn_moe_down-73): 276 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 762 us MUL_MAT_ID(ffn_moe_down-74): 663 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 935 us MUL_MAT_ID(ffn_moe_down-75): 573 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 767 us MUL_MAT_ID(ffn_moe_down-76): 440 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 487 us MUL_MAT_ID(ffn_moe_down-77): 296 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 526 us MUL_MAT_ID(ffn_moe_down-78): 240 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 540 us MUL_MAT_ID(ffn_moe_down-79): 420 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 487 us MUL_MAT_ID(ffn_moe_down-80): 213 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 514 us MUL_MAT_ID(ffn_moe_down-81): 343 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 508 us MUL_MAT_ID(ffn_moe_down-82): 223 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 592 us MUL_MAT_ID(ffn_moe_down-83): 230 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 741 us MUL_MAT_ID(ffn_moe_down-84): 355 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 562 us MUL_MAT_ID(ffn_moe_down-85): 345 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 560 us MUL_MAT_ID(ffn_moe_down-86): 295 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 467 us MUL_MAT_ID(ffn_moe_down-87): 330 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 465 us MUL_MAT_ID(ffn_moe_down-88): 287 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 476 us MUL_MAT_ID(ffn_moe_down-89): 222 us ggml_barrier(...): 361 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 531 us MUL_MAT_ID(ffn_moe_down-90): 296 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 489 us MUL_MAT_ID(ffn_moe_down-91): 457 us ggml_barrier(...): 337 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 384 us MUL_MAT_ID(ffn_moe_down-25): 226 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 463 us MUL_MAT_ID(ffn_moe_down-26): 278 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 469 us MUL_MAT_ID(ffn_moe_down-27): 308 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 369 us MUL_MAT_ID(ffn_moe_down-28): 206 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 463 us MUL_MAT_ID(ffn_moe_down-29): 176 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 440 us MUL_MAT_ID(ffn_moe_down-30): 240 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 449 us MUL_MAT_ID(ffn_moe_down-31): 238 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 409 us MUL_MAT_ID(ffn_moe_down-32): 300 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 443 us MUL_MAT_ID(ffn_moe_down-33): 310 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 470 us MUL_MAT_ID(ffn_moe_down-34): 482 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 596 us MUL_MAT_ID(ffn_moe_down-35): 202 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 428 us MUL_MAT_ID(ffn_moe_down-36): 228 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 516 us MUL_MAT_ID(ffn_moe_down-37): 337 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 407 us MUL_MAT_ID(ffn_moe_down-38): 219 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 498 us MUL_MAT_ID(ffn_moe_down-39): 285 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 453 us MUL_MAT_ID(ffn_moe_down-40): 238 us ggml_barrier(...): 340 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 505 us MUL_MAT_ID(ffn_moe_down-41): 229 us ggml_barrier(...): 339 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 533 us MUL_MAT_ID(ffn_moe_down-42): 264 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 457 us MUL_MAT_ID(ffn_moe_down-43): 270 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 474 us MUL_MAT_ID(ffn_moe_down-44): 294 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 399 us MUL_MAT_ID(ffn_moe_down-45): 214 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 488 us MUL_MAT_ID(ffn_moe_down-46): 259 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 505 us MUL_MAT_ID(ffn_moe_down-47): 233 us ggml_barrier(...): 342 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 501 us MUL_MAT_ID(ffn_moe_down-48): 243 us ggml_barrier(...): 352 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 535 us MUL_MAT_ID(ffn_moe_down-49): 232 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 530 us MUL_MAT_ID(ffn_moe_down-50): 211 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 517 us MUL_MAT_ID(ffn_moe_down-51): 287 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 412 us MUL_MAT_ID(ffn_moe_down-52): 212 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 449 us MUL_MAT_ID(ffn_moe_down-53): 293 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 456 us MUL_MAT_ID(ffn_moe_down-54): 322 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 530 us MUL_MAT_ID(ffn_moe_down-55): 287 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 536 us MUL_MAT_ID(ffn_moe_down-56): 287 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 563 us MUL_MAT_ID(ffn_moe_down-57): 309 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 468 us MUL_MAT_ID(ffn_moe_down-58): 245 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 465 us MUL_MAT_ID(ffn_moe_down-59): 289 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 499 us MUL_MAT_ID(ffn_moe_down-60): 262 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 517 us MUL_MAT_ID(ffn_moe_down-61): 220 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 537 us MUL_MAT_ID(ffn_moe_down-62): 216 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 485 us MUL_MAT_ID(ffn_moe_down-63): 249 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 613 us MUL_MAT_ID(ffn_moe_down-64): 229 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 464 us MUL_MAT_ID(ffn_moe_down-65): 224 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 409 us MUL_MAT_ID(ffn_moe_down-66): 270 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 486 us MUL_MAT_ID(ffn_moe_down-67): 219 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 549 us MUL_MAT_ID(ffn_moe_down-68): 249 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 428 us MUL_MAT_ID(ffn_moe_down-69): 200 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 540 us MUL_MAT_ID(ffn_moe_down-70): 430 us ggml_barrier(...): 389 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 497 us MUL_MAT_ID(ffn_moe_down-71): 357 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 479 us MUL_MAT_ID(ffn_moe_down-72): 271 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 510 us MUL_MAT_ID(ffn_moe_down-73): 322 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 485 us MUL_MAT_ID(ffn_moe_down-74): 209 us ggml_barrier(...): 420 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 585 us MUL_MAT_ID(ffn_moe_down-75): 264 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 864 us MUL_MAT_ID(ffn_moe_down-76): 669 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 494 us MUL_MAT_ID(ffn_moe_down-77): 189 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 416 us MUL_MAT_ID(ffn_moe_down-78): 272 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 523 us MUL_MAT_ID(ffn_moe_down-79): 235 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 516 us MUL_MAT_ID(ffn_moe_down-80): 349 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 456 us MUL_MAT_ID(ffn_moe_down-81): 319 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 434 us MUL_MAT_ID(ffn_moe_down-82): 332 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 494 us MUL_MAT_ID(ffn_moe_down-83): 382 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 501 us MUL_MAT_ID(ffn_moe_down-84): 256 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 484 us MUL_MAT_ID(ffn_moe_down-85): 266 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 453 us MUL_MAT_ID(ffn_moe_down-86): 277 us ggml_barrier(...): 416 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 434 us MUL_MAT_ID(ffn_moe_down-87): 261 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 488 us MUL_MAT_ID(ffn_moe_down-88): 204 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 494 us MUL_MAT_ID(ffn_moe_down-89): 333 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 506 us MUL_MAT_ID(ffn_moe_down-90): 296 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 472 us MUL_MAT_ID(ffn_moe_down-91): 309 us ggml_barrier(...): 285 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 510 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 443 us MUL_MAT_ID(ffn_moe_down-26): 251 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 521 us MUL_MAT_ID(ffn_moe_down-27): 210 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 388 us MUL_MAT_ID(ffn_moe_down-28): 230 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 416 us MUL_MAT_ID(ffn_moe_down-29): 202 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 475 us MUL_MAT_ID(ffn_moe_down-30): 291 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 527 us MUL_MAT_ID(ffn_moe_down-31): 203 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 442 us MUL_MAT_ID(ffn_moe_down-32): 240 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 433 us MUL_MAT_ID(ffn_moe_down-33): 189 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 463 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 421 us MUL_MAT_ID(ffn_moe_down-35): 251 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 429 us MUL_MAT_ID(ffn_moe_down-36): 221 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 508 us MUL_MAT_ID(ffn_moe_down-37): 341 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 448 us MUL_MAT_ID(ffn_moe_down-38): 282 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 431 us MUL_MAT_ID(ffn_moe_down-39): 223 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 391 us MUL_MAT_ID(ffn_moe_down-40): 533 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 456 us MUL_MAT_ID(ffn_moe_down-41): 213 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 414 us MUL_MAT_ID(ffn_moe_down-42): 270 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 660 us MUL_MAT_ID(ffn_moe_down-43): 291 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 387 us MUL_MAT_ID(ffn_moe_down-44): 236 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 449 us MUL_MAT_ID(ffn_moe_down-45): 690 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 444 us MUL_MAT_ID(ffn_moe_down-46): 288 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 406 us MUL_MAT_ID(ffn_moe_down-47): 226 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 419 us MUL_MAT_ID(ffn_moe_down-48): 213 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 517 us MUL_MAT_ID(ffn_moe_down-49): 302 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 440 us MUL_MAT_ID(ffn_moe_down-50): 199 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 487 us MUL_MAT_ID(ffn_moe_down-51): 206 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 388 us MUL_MAT_ID(ffn_moe_down-52): 528 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 454 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 428 us MUL_MAT_ID(ffn_moe_down-54): 208 us ggml_barrier(...): 351 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 459 us MUL_MAT_ID(ffn_moe_down-55): 215 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 477 us MUL_MAT_ID(ffn_moe_down-56): 300 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 512 us MUL_MAT_ID(ffn_moe_down-57): 298 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 467 us MUL_MAT_ID(ffn_moe_down-58): 304 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 447 us MUL_MAT_ID(ffn_moe_down-59): 216 us ggml_barrier(...): 342 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 465 us MUL_MAT_ID(ffn_moe_down-60): 205 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 439 us MUL_MAT_ID(ffn_moe_down-61): 203 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 442 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 383 us MUL_MAT_ID(ffn_moe_down-63): 200 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 457 us MUL_MAT_ID(ffn_moe_down-64): 281 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 466 us MUL_MAT_ID(ffn_moe_down-65): 229 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 550 us MUL_MAT_ID(ffn_moe_down-66): 217 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 464 us MUL_MAT_ID(ffn_moe_down-67): 233 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 473 us MUL_MAT_ID(ffn_moe_down-68): 202 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 544 us MUL_MAT_ID(ffn_moe_down-69): 212 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 541 us MUL_MAT_ID(ffn_moe_down-70): 262 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 551 us MUL_MAT_ID(ffn_moe_down-71): 229 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 617 us MUL_MAT_ID(ffn_moe_down-72): 276 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 830 us MUL_MAT_ID(ffn_moe_down-73): 396 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 737 us MUL_MAT_ID(ffn_moe_down-74): 319 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 432 us MUL_MAT_ID(ffn_moe_down-75): 325 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 403 us MUL_MAT_ID(ffn_moe_down-76): 220 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 410 us MUL_MAT_ID(ffn_moe_down-77): 294 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 440 us MUL_MAT_ID(ffn_moe_down-78): 188 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 381 us MUL_MAT_ID(ffn_moe_down-79): 216 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 448 us MUL_MAT_ID(ffn_moe_down-80): 409 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 413 us MUL_MAT_ID(ffn_moe_down-81): 213 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 675 us MUL_MAT_ID(ffn_moe_down-82): 351 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 397 us MUL_MAT_ID(ffn_moe_down-83): 279 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 403 us MUL_MAT_ID(ffn_moe_down-84): 244 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 472 us MUL_MAT_ID(ffn_moe_down-85): 248 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 559 us MUL_MAT_ID(ffn_moe_down-86): 324 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 477 us MUL_MAT_ID(ffn_moe_down-87): 209 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 408 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 475 us MUL_MAT_ID(ffn_moe_down-89): 287 us ggml_barrier(...): 347 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 408 us MUL_MAT_ID(ffn_moe_down-90): 337 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 516 us MUL_MAT_ID(ffn_moe_down-91): 227 us ggml_barrier(...): 235 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 390 us MUL_MAT_ID(ffn_moe_down-25): 260 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 435 us MUL_MAT_ID(ffn_moe_down-26): 285 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 434 us MUL_MAT_ID(ffn_moe_down-27): 333 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 426 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 463 us MUL_MAT_ID(ffn_moe_down-29): 244 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 449 us MUL_MAT_ID(ffn_moe_down-30): 218 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 502 us MUL_MAT_ID(ffn_moe_down-31): 327 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 452 us MUL_MAT_ID(ffn_moe_down-32): 225 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 436 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 454 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 452 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 435 us MUL_MAT_ID(ffn_moe_down-36): 315 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 528 us MUL_MAT_ID(ffn_moe_down-37): 261 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 438 us MUL_MAT_ID(ffn_moe_down-38): 556 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 434 us MUL_MAT_ID(ffn_moe_down-39): 319 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 455 us MUL_MAT_ID(ffn_moe_down-40): 223 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 564 us MUL_MAT_ID(ffn_moe_down-41): 211 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 501 us MUL_MAT_ID(ffn_moe_down-42): 195 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 544 us MUL_MAT_ID(ffn_moe_down-43): 246 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 437 us MUL_MAT_ID(ffn_moe_down-44): 280 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 481 us MUL_MAT_ID(ffn_moe_down-45): 264 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 503 us MUL_MAT_ID(ffn_moe_down-46): 239 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 470 us MUL_MAT_ID(ffn_moe_down-47): 209 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 426 us MUL_MAT_ID(ffn_moe_down-48): 263 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 343 us MUL_MAT_ID(ffn_moe_down-49): 246 us ggml_barrier(...): 370 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 518 us MUL_MAT_ID(ffn_moe_down-50): 509 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 545 us MUL_MAT_ID(ffn_moe_down-51): 246 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 390 us MUL_MAT_ID(ffn_moe_down-52): 216 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 492 us MUL_MAT_ID(ffn_moe_down-53): 354 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 637 us MUL_MAT_ID(ffn_moe_down-54): 320 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 480 us MUL_MAT_ID(ffn_moe_down-55): 268 us ggml_barrier(...): 341 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 519 us MUL_MAT_ID(ffn_moe_down-56): 234 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 470 us MUL_MAT_ID(ffn_moe_down-57): 225 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 432 us MUL_MAT_ID(ffn_moe_down-58): 215 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 431 us MUL_MAT_ID(ffn_moe_down-59): 261 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 517 us MUL_MAT_ID(ffn_moe_down-60): 209 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 474 us MUL_MAT_ID(ffn_moe_down-61): 249 us ggml_barrier(...): 427 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 533 us MUL_MAT_ID(ffn_moe_down-62): 235 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 450 us MUL_MAT_ID(ffn_moe_down-63): 292 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 516 us MUL_MAT_ID(ffn_moe_down-64): 203 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 516 us MUL_MAT_ID(ffn_moe_down-65): 217 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 498 us MUL_MAT_ID(ffn_moe_down-66): 210 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 485 us MUL_MAT_ID(ffn_moe_down-67): 218 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 369 us MUL_MAT_ID(ffn_moe_down-68): 298 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 556 us MUL_MAT_ID(ffn_moe_down-69): 337 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 664 us MUL_MAT_ID(ffn_moe_down-70): 419 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 419 us MUL_MAT_ID(ffn_moe_down-71): 623 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 526 us MUL_MAT_ID(ffn_moe_down-72): 232 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 511 us MUL_MAT_ID(ffn_moe_down-73): 274 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 452 us MUL_MAT_ID(ffn_moe_down-74): 240 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 422 us MUL_MAT_ID(ffn_moe_down-75): 268 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 472 us MUL_MAT_ID(ffn_moe_down-76): 268 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 458 us MUL_MAT_ID(ffn_moe_down-77): 306 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 558 us MUL_MAT_ID(ffn_moe_down-78): 244 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 464 us MUL_MAT_ID(ffn_moe_down-79): 294 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 519 us MUL_MAT_ID(ffn_moe_down-80): 596 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 417 us MUL_MAT_ID(ffn_moe_down-81): 207 us ggml_barrier(...): 364 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 408 us MUL_MAT_ID(ffn_moe_down-82): 222 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 532 us MUL_MAT_ID(ffn_moe_down-83): 480 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 574 us MUL_MAT_ID(ffn_moe_down-84): 329 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 552 us MUL_MAT_ID(ffn_moe_down-85): 287 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 523 us MUL_MAT_ID(ffn_moe_down-86): 301 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 454 us MUL_MAT_ID(ffn_moe_down-87): 242 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 523 us MUL_MAT_ID(ffn_moe_down-88): 372 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 389 us MUL_MAT_ID(ffn_moe_down-89): 183 us ggml_barrier(...): 353 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 423 us MUL_MAT_ID(ffn_moe_down-90): 236 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 495 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 325 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 444 us MUL_MAT_ID(ffn_moe_down-25): 233 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 486 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 450 us MUL_MAT_ID(ffn_moe_down-27): 212 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 490 us MUL_MAT_ID(ffn_moe_down-28): 516 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 490 us MUL_MAT_ID(ffn_moe_down-29): 214 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 506 us MUL_MAT_ID(ffn_moe_down-30): 203 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 427 us MUL_MAT_ID(ffn_moe_down-31): 305 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 458 us MUL_MAT_ID(ffn_moe_down-32): 239 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 702 us MUL_MAT_ID(ffn_moe_down-33): 269 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 455 us MUL_MAT_ID(ffn_moe_down-34): 187 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 443 us MUL_MAT_ID(ffn_moe_down-35): 264 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 447 us MUL_MAT_ID(ffn_moe_down-36): 208 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 463 us MUL_MAT_ID(ffn_moe_down-37): 298 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 384 us MUL_MAT_ID(ffn_moe_down-38): 218 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 519 us MUL_MAT_ID(ffn_moe_down-39): 280 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 531 us MUL_MAT_ID(ffn_moe_down-40): 278 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 475 us MUL_MAT_ID(ffn_moe_down-41): 291 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 474 us MUL_MAT_ID(ffn_moe_down-42): 212 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 474 us MUL_MAT_ID(ffn_moe_down-43): 347 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 524 us MUL_MAT_ID(ffn_moe_down-44): 230 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 441 us MUL_MAT_ID(ffn_moe_down-45): 250 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 475 us MUL_MAT_ID(ffn_moe_down-46): 236 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 416 us MUL_MAT_ID(ffn_moe_down-47): 305 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 478 us MUL_MAT_ID(ffn_moe_down-48): 727 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 418 us MUL_MAT_ID(ffn_moe_down-49): 213 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 417 us MUL_MAT_ID(ffn_moe_down-50): 362 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 451 us MUL_MAT_ID(ffn_moe_down-51): 356 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 530 us MUL_MAT_ID(ffn_moe_down-52): 266 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 509 us MUL_MAT_ID(ffn_moe_down-53): 206 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 520 us MUL_MAT_ID(ffn_moe_down-54): 370 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 550 us MUL_MAT_ID(ffn_moe_down-55): 205 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 608 us MUL_MAT_ID(ffn_moe_down-56): 309 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 515 us MUL_MAT_ID(ffn_moe_down-57): 211 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 434 us MUL_MAT_ID(ffn_moe_down-58): 236 us ggml_barrier(...): 377 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 497 us MUL_MAT_ID(ffn_moe_down-59): 266 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 396 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 432 us MUL_MAT_ID(ffn_moe_down-61): 281 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 482 us MUL_MAT_ID(ffn_moe_down-62): 324 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 485 us MUL_MAT_ID(ffn_moe_down-63): 240 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 428 us MUL_MAT_ID(ffn_moe_down-64): 301 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 410 us MUL_MAT_ID(ffn_moe_down-65): 279 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 536 us MUL_MAT_ID(ffn_moe_down-66): 298 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 821 us MUL_MAT_ID(ffn_moe_down-67): 476 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 649 us MUL_MAT_ID(ffn_moe_down-68): 278 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 490 us MUL_MAT_ID(ffn_moe_down-69): 262 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 407 us MUL_MAT_ID(ffn_moe_down-70): 257 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 458 us MUL_MAT_ID(ffn_moe_down-71): 255 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 460 us MUL_MAT_ID(ffn_moe_down-72): 264 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 472 us MUL_MAT_ID(ffn_moe_down-73): 348 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 490 us MUL_MAT_ID(ffn_moe_down-74): 300 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 491 us MUL_MAT_ID(ffn_moe_down-75): 294 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 727 us MUL_MAT_ID(ffn_moe_down-76): 241 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 508 us MUL_MAT_ID(ffn_moe_down-77): 288 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 486 us MUL_MAT_ID(ffn_moe_down-78): 332 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 460 us MUL_MAT_ID(ffn_moe_down-79): 207 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 505 us MUL_MAT_ID(ffn_moe_down-80): 277 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 433 us MUL_MAT_ID(ffn_moe_down-81): 213 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 454 us MUL_MAT_ID(ffn_moe_down-82): 272 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 390 us MUL_MAT_ID(ffn_moe_down-83): 222 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 521 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 448 us MUL_MAT_ID(ffn_moe_down-85): 215 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 548 us MUL_MAT_ID(ffn_moe_down-86): 250 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 472 us MUL_MAT_ID(ffn_moe_down-87): 404 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 500 us MUL_MAT_ID(ffn_moe_down-88): 223 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 433 us MUL_MAT_ID(ffn_moe_down-89): 319 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 455 us MUL_MAT_ID(ffn_moe_down-90): 277 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 536 us MUL_MAT_ID(ffn_moe_down-91): 266 us ggml_barrier(...): 307 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 466 us MUL_MAT_ID(ffn_moe_down-25): 221 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 393 us MUL_MAT_ID(ffn_moe_down-26): 188 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 508 us MUL_MAT_ID(ffn_moe_down-27): 233 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 531 us MUL_MAT_ID(ffn_moe_down-28): 540 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 372 us MUL_MAT_ID(ffn_moe_down-29): 247 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 526 us MUL_MAT_ID(ffn_moe_down-30): 182 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 460 us MUL_MAT_ID(ffn_moe_down-31): 259 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 446 us MUL_MAT_ID(ffn_moe_down-32): 302 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 428 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 439 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 494 us MUL_MAT_ID(ffn_moe_down-35): 195 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 414 us MUL_MAT_ID(ffn_moe_down-36): 188 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 385 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 407 us MUL_MAT_ID(ffn_moe_down-38): 217 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 509 us MUL_MAT_ID(ffn_moe_down-39): 288 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 496 us MUL_MAT_ID(ffn_moe_down-40): 351 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 386 us MUL_MAT_ID(ffn_moe_down-41): 183 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 452 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 464 us MUL_MAT_ID(ffn_moe_down-43): 217 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 385 us MUL_MAT_ID(ffn_moe_down-44): 184 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 433 us MUL_MAT_ID(ffn_moe_down-45): 279 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 406 us MUL_MAT_ID(ffn_moe_down-46): 188 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 483 us MUL_MAT_ID(ffn_moe_down-47): 197 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 521 us MUL_MAT_ID(ffn_moe_down-48): 207 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 465 us MUL_MAT_ID(ffn_moe_down-49): 239 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 385 us MUL_MAT_ID(ffn_moe_down-50): 263 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 438 us MUL_MAT_ID(ffn_moe_down-51): 539 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 387 us MUL_MAT_ID(ffn_moe_down-52): 312 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 422 us MUL_MAT_ID(ffn_moe_down-53): 477 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 408 us MUL_MAT_ID(ffn_moe_down-54): 302 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 497 us MUL_MAT_ID(ffn_moe_down-55): 206 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 484 us MUL_MAT_ID(ffn_moe_down-56): 213 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 433 us MUL_MAT_ID(ffn_moe_down-57): 203 us ggml_barrier(...): 347 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 471 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 462 us MUL_MAT_ID(ffn_moe_down-59): 261 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 568 us MUL_MAT_ID(ffn_moe_down-60): 411 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 456 us MUL_MAT_ID(ffn_moe_down-61): 305 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 478 us MUL_MAT_ID(ffn_moe_down-62): 224 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 391 us MUL_MAT_ID(ffn_moe_down-63): 246 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 420 us MUL_MAT_ID(ffn_moe_down-64): 258 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 528 us MUL_MAT_ID(ffn_moe_down-65): 514 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 549 us MUL_MAT_ID(ffn_moe_down-66): 385 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 550 us MUL_MAT_ID(ffn_moe_down-67): 225 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 424 us MUL_MAT_ID(ffn_moe_down-68): 322 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 419 us MUL_MAT_ID(ffn_moe_down-69): 184 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 431 us MUL_MAT_ID(ffn_moe_down-70): 258 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 454 us MUL_MAT_ID(ffn_moe_down-71): 289 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 548 us MUL_MAT_ID(ffn_moe_down-72): 292 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 234 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 418 us MUL_MAT_ID(ffn_moe_down-74): 222 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 774 us MUL_MAT_ID(ffn_moe_down-75): 267 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 478 us MUL_MAT_ID(ffn_moe_down-76): 221 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 398 us MUL_MAT_ID(ffn_moe_down-77): 167 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 430 us MUL_MAT_ID(ffn_moe_down-78): 292 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 375 us MUL_MAT_ID(ffn_moe_down-79): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 398 us MUL_MAT_ID(ffn_moe_down-80): 191 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 459 us MUL_MAT_ID(ffn_moe_down-81): 217 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 385 us MUL_MAT_ID(ffn_moe_down-82): 201 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 537 us MUL_MAT_ID(ffn_moe_down-83): 251 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 445 us MUL_MAT_ID(ffn_moe_down-84): 295 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 396 us MUL_MAT_ID(ffn_moe_down-85): 246 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 463 us MUL_MAT_ID(ffn_moe_down-86): 219 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 412 us MUL_MAT_ID(ffn_moe_down-87): 249 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 426 us MUL_MAT_ID(ffn_moe_down-88): 251 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 473 us MUL_MAT_ID(ffn_moe_down-89): 288 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 517 us MUL_MAT_ID(ffn_moe_down-90): 303 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 443 us MUL_MAT_ID(ffn_moe_down-91): 297 us ggml_barrier(...): 214 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 493 us MUL_MAT_ID(ffn_moe_down-25): 442 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 374 us MUL_MAT_ID(ffn_moe_down-26): 188 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 446 us MUL_MAT_ID(ffn_moe_down-27): 303 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 383 us MUL_MAT_ID(ffn_moe_down-28): 196 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 443 us MUL_MAT_ID(ffn_moe_down-29): 237 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 441 us MUL_MAT_ID(ffn_moe_down-30): 364 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 347 us MUL_MAT_ID(ffn_moe_down-31): 241 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 423 us MUL_MAT_ID(ffn_moe_down-32): 262 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 400 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 388 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 498 us MUL_MAT_ID(ffn_moe_down-35): 210 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 427 us MUL_MAT_ID(ffn_moe_down-36): 267 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 396 us MUL_MAT_ID(ffn_moe_down-37): 243 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 477 us MUL_MAT_ID(ffn_moe_down-38): 252 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 537 us MUL_MAT_ID(ffn_moe_down-39): 273 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 439 us MUL_MAT_ID(ffn_moe_down-40): 320 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 404 us MUL_MAT_ID(ffn_moe_down-41): 281 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 444 us MUL_MAT_ID(ffn_moe_down-42): 279 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 405 us MUL_MAT_ID(ffn_moe_down-43): 218 us ggml_barrier(...): 344 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 446 us MUL_MAT_ID(ffn_moe_down-44): 228 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 468 us MUL_MAT_ID(ffn_moe_down-45): 546 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 374 us MUL_MAT_ID(ffn_moe_down-46): 268 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 625 us MUL_MAT_ID(ffn_moe_down-47): 360 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 365 us MUL_MAT_ID(ffn_moe_down-48): 269 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 387 us MUL_MAT_ID(ffn_moe_down-49): 202 us ggml_barrier(...): 334 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 469 us MUL_MAT_ID(ffn_moe_down-50): 199 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 384 us MUL_MAT_ID(ffn_moe_down-51): 248 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 413 us MUL_MAT_ID(ffn_moe_down-52): 223 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 426 us MUL_MAT_ID(ffn_moe_down-53): 200 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 429 us MUL_MAT_ID(ffn_moe_down-54): 266 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 338 us ggml_barrier(...): 352 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 385 us MUL_MAT_ID(ffn_moe_down-56): 269 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 463 us MUL_MAT_ID(ffn_moe_down-57): 429 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 527 us MUL_MAT_ID(ffn_moe_down-58): 270 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 408 us MUL_MAT_ID(ffn_moe_down-59): 178 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 537 us MUL_MAT_ID(ffn_moe_down-60): 205 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 395 us MUL_MAT_ID(ffn_moe_down-61): 268 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 415 us MUL_MAT_ID(ffn_moe_down-62): 227 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 397 us MUL_MAT_ID(ffn_moe_down-63): 204 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 519 us MUL_MAT_ID(ffn_moe_down-64): 345 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 510 us MUL_MAT_ID(ffn_moe_down-65): 252 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 615 us MUL_MAT_ID(ffn_moe_down-66): 213 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 465 us MUL_MAT_ID(ffn_moe_down-67): 209 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 483 us MUL_MAT_ID(ffn_moe_down-68): 248 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 435 us MUL_MAT_ID(ffn_moe_down-69): 237 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 424 us MUL_MAT_ID(ffn_moe_down-70): 173 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 413 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 506 us MUL_MAT_ID(ffn_moe_down-72): 420 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 355 us MUL_MAT_ID(ffn_moe_down-73): 199 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 406 us MUL_MAT_ID(ffn_moe_down-74): 269 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 365 us MUL_MAT_ID(ffn_moe_down-75): 229 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 535 us MUL_MAT_ID(ffn_moe_down-76): 344 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 367 us MUL_MAT_ID(ffn_moe_down-77): 295 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 402 us MUL_MAT_ID(ffn_moe_down-78): 494 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 418 us MUL_MAT_ID(ffn_moe_down-79): 263 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 437 us MUL_MAT_ID(ffn_moe_down-80): 215 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 402 us MUL_MAT_ID(ffn_moe_down-81): 236 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 460 us MUL_MAT_ID(ffn_moe_down-82): 252 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 489 us MUL_MAT_ID(ffn_moe_down-83): 208 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 432 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 404 us MUL_MAT_ID(ffn_moe_down-85): 591 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 433 us MUL_MAT_ID(ffn_moe_down-86): 206 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 458 us MUL_MAT_ID(ffn_moe_down-87): 274 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 388 us MUL_MAT_ID(ffn_moe_down-88): 208 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 408 us MUL_MAT_ID(ffn_moe_down-89): 223 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 468 us MUL_MAT_ID(ffn_moe_down-90): 270 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 507 us MUL_MAT_ID(ffn_moe_down-91): 214 us ggml_barrier(...): 138 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 455 us MUL_MAT_ID(ffn_moe_down-25): 182 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 464 us MUL_MAT_ID(ffn_moe_down-26): 267 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 657 us MUL_MAT_ID(ffn_moe_down-27): 218 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 376 us MUL_MAT_ID(ffn_moe_down-28): 237 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 447 us MUL_MAT_ID(ffn_moe_down-29): 239 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 466 us MUL_MAT_ID(ffn_moe_down-30): 208 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 492 us MUL_MAT_ID(ffn_moe_down-31): 268 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 371 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 398 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 442 us MUL_MAT_ID(ffn_moe_down-34): 178 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 408 us MUL_MAT_ID(ffn_moe_down-35): 554 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 453 us MUL_MAT_ID(ffn_moe_down-36): 186 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 455 us MUL_MAT_ID(ffn_moe_down-37): 246 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 417 us MUL_MAT_ID(ffn_moe_down-38): 226 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 405 us MUL_MAT_ID(ffn_moe_down-39): 227 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 442 us MUL_MAT_ID(ffn_moe_down-40): 252 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 463 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 526 us MUL_MAT_ID(ffn_moe_down-42): 469 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 397 us MUL_MAT_ID(ffn_moe_down-43): 186 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 761 us MUL_MAT_ID(ffn_moe_down-44): 259 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 495 us MUL_MAT_ID(ffn_moe_down-45): 340 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 428 us MUL_MAT_ID(ffn_moe_down-46): 209 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 561 us MUL_MAT_ID(ffn_moe_down-47): 212 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 443 us MUL_MAT_ID(ffn_moe_down-48): 205 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 410 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 465 us MUL_MAT_ID(ffn_moe_down-50): 300 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 426 us MUL_MAT_ID(ffn_moe_down-51): 183 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 482 us MUL_MAT_ID(ffn_moe_down-52): 253 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 491 us MUL_MAT_ID(ffn_moe_down-53): 186 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 552 us MUL_MAT_ID(ffn_moe_down-54): 280 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 399 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 480 us MUL_MAT_ID(ffn_moe_down-56): 333 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 399 us MUL_MAT_ID(ffn_moe_down-57): 404 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 486 us MUL_MAT_ID(ffn_moe_down-58): 231 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 375 us MUL_MAT_ID(ffn_moe_down-59): 209 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 468 us MUL_MAT_ID(ffn_moe_down-60): 273 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 425 us MUL_MAT_ID(ffn_moe_down-61): 268 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 431 us MUL_MAT_ID(ffn_moe_down-62): 256 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 567 us MUL_MAT_ID(ffn_moe_down-63): 231 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 499 us MUL_MAT_ID(ffn_moe_down-64): 330 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 792 us MUL_MAT_ID(ffn_moe_down-65): 429 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 459 us MUL_MAT_ID(ffn_moe_down-66): 308 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 445 us MUL_MAT_ID(ffn_moe_down-67): 390 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 548 us MUL_MAT_ID(ffn_moe_down-68): 339 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 487 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 469 us MUL_MAT_ID(ffn_moe_down-70): 354 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 421 us MUL_MAT_ID(ffn_moe_down-71): 225 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 418 us MUL_MAT_ID(ffn_moe_down-72): 253 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 535 us MUL_MAT_ID(ffn_moe_down-73): 236 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 521 us MUL_MAT_ID(ffn_moe_down-74): 382 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 357 us MUL_MAT_ID(ffn_moe_down-75): 266 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 389 us MUL_MAT_ID(ffn_moe_down-76): 333 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 444 us MUL_MAT_ID(ffn_moe_down-77): 237 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 394 us MUL_MAT_ID(ffn_moe_down-78): 309 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 405 us MUL_MAT_ID(ffn_moe_down-79): 210 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 500 us MUL_MAT_ID(ffn_moe_down-80): 224 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 530 us MUL_MAT_ID(ffn_moe_down-81): 199 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 418 us MUL_MAT_ID(ffn_moe_down-82): 220 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 417 us MUL_MAT_ID(ffn_moe_down-83): 280 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 437 us MUL_MAT_ID(ffn_moe_down-84): 262 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 417 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 456 us MUL_MAT_ID(ffn_moe_down-86): 463 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 421 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 503 us MUL_MAT_ID(ffn_moe_down-88): 185 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 399 us MUL_MAT_ID(ffn_moe_down-89): 207 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 459 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 370 us MUL_MAT_ID(ffn_moe_down-91): 199 us ggml_barrier(...): 313 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 532 us MUL_MAT_ID(ffn_moe_down-25): 224 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 428 us MUL_MAT_ID(ffn_moe_down-26): 404 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 378 us MUL_MAT_ID(ffn_moe_down-27): 204 us ggml_barrier(...): 389 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 403 us MUL_MAT_ID(ffn_moe_down-28): 250 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 383 us MUL_MAT_ID(ffn_moe_down-29): 238 us ggml_barrier(...): 423 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 524 us MUL_MAT_ID(ffn_moe_down-30): 231 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 502 us MUL_MAT_ID(ffn_moe_down-31): 574 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 388 us MUL_MAT_ID(ffn_moe_down-32): 368 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 408 us MUL_MAT_ID(ffn_moe_down-33): 182 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 620 us MUL_MAT_ID(ffn_moe_down-34): 295 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 376 us MUL_MAT_ID(ffn_moe_down-35): 259 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 437 us MUL_MAT_ID(ffn_moe_down-36): 230 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 566 us MUL_MAT_ID(ffn_moe_down-37): 243 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 393 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 449 us MUL_MAT_ID(ffn_moe_down-39): 259 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 404 us MUL_MAT_ID(ffn_moe_down-40): 402 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 431 us MUL_MAT_ID(ffn_moe_down-41): 250 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 469 us MUL_MAT_ID(ffn_moe_down-42): 238 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 521 us MUL_MAT_ID(ffn_moe_down-43): 211 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 376 us MUL_MAT_ID(ffn_moe_down-44): 233 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 426 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 364 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 399 us MUL_MAT_ID(ffn_moe_down-47): 320 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 394 us MUL_MAT_ID(ffn_moe_down-48): 324 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 384 us MUL_MAT_ID(ffn_moe_down-49): 212 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 396 us MUL_MAT_ID(ffn_moe_down-50): 215 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 642 us MUL_MAT_ID(ffn_moe_down-51): 309 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 460 us MUL_MAT_ID(ffn_moe_down-52): 249 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 415 us MUL_MAT_ID(ffn_moe_down-53): 256 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 480 us MUL_MAT_ID(ffn_moe_down-54): 211 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 437 us MUL_MAT_ID(ffn_moe_down-55): 449 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 415 us MUL_MAT_ID(ffn_moe_down-56): 275 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 539 us MUL_MAT_ID(ffn_moe_down-57): 225 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 459 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 430 us MUL_MAT_ID(ffn_moe_down-59): 244 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 373 us MUL_MAT_ID(ffn_moe_down-60): 307 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 411 us MUL_MAT_ID(ffn_moe_down-61): 187 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 452 us MUL_MAT_ID(ffn_moe_down-62): 266 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 506 us MUL_MAT_ID(ffn_moe_down-63): 231 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 492 us MUL_MAT_ID(ffn_moe_down-64): 258 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 799 us MUL_MAT_ID(ffn_moe_down-65): 416 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 368 us MUL_MAT_ID(ffn_moe_down-66): 257 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 386 us MUL_MAT_ID(ffn_moe_down-67): 271 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 406 us MUL_MAT_ID(ffn_moe_down-68): 245 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 457 us MUL_MAT_ID(ffn_moe_down-69): 233 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 529 us MUL_MAT_ID(ffn_moe_down-70): 205 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 431 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 437 us MUL_MAT_ID(ffn_moe_down-72): 313 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 449 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 504 us MUL_MAT_ID(ffn_moe_down-74): 210 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 403 us MUL_MAT_ID(ffn_moe_down-75): 255 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 363 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 420 us MUL_MAT_ID(ffn_moe_down-77): 261 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 511 us MUL_MAT_ID(ffn_moe_down-78): 211 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 423 us MUL_MAT_ID(ffn_moe_down-79): 244 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 396 us MUL_MAT_ID(ffn_moe_down-80): 389 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 474 us MUL_MAT_ID(ffn_moe_down-81): 222 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 502 us MUL_MAT_ID(ffn_moe_down-82): 256 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 476 us MUL_MAT_ID(ffn_moe_down-83): 216 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 478 us MUL_MAT_ID(ffn_moe_down-84): 378 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 451 us MUL_MAT_ID(ffn_moe_down-85): 461 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 436 us MUL_MAT_ID(ffn_moe_down-86): 357 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 446 us MUL_MAT_ID(ffn_moe_down-87): 231 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 413 us MUL_MAT_ID(ffn_moe_down-88): 233 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 416 us MUL_MAT_ID(ffn_moe_down-89): 196 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 427 us MUL_MAT_ID(ffn_moe_down-90): 224 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 470 us MUL_MAT_ID(ffn_moe_down-91): 339 us ggml_barrier(...): 224 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 471 us MUL_MAT_ID(ffn_moe_down-25): 183 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 422 us MUL_MAT_ID(ffn_moe_down-26): 209 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 383 us MUL_MAT_ID(ffn_moe_down-27): 175 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 401 us MUL_MAT_ID(ffn_moe_down-28): 227 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 431 us MUL_MAT_ID(ffn_moe_down-29): 218 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 395 us MUL_MAT_ID(ffn_moe_down-30): 364 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 439 us MUL_MAT_ID(ffn_moe_down-31): 190 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 393 us MUL_MAT_ID(ffn_moe_down-32): 262 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 571 us MUL_MAT_ID(ffn_moe_down-33): 386 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 374 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 426 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 541 us MUL_MAT_ID(ffn_moe_down-36): 243 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 452 us MUL_MAT_ID(ffn_moe_down-37): 308 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 552 us MUL_MAT_ID(ffn_moe_down-38): 274 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 493 us MUL_MAT_ID(ffn_moe_down-39): 225 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 480 us MUL_MAT_ID(ffn_moe_down-40): 209 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 458 us MUL_MAT_ID(ffn_moe_down-41): 207 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 454 us MUL_MAT_ID(ffn_moe_down-42): 202 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 441 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 389 us MUL_MAT_ID(ffn_moe_down-44): 217 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 453 us MUL_MAT_ID(ffn_moe_down-45): 242 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 441 us MUL_MAT_ID(ffn_moe_down-46): 242 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 405 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 454 us MUL_MAT_ID(ffn_moe_down-48): 202 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 387 us MUL_MAT_ID(ffn_moe_down-49): 205 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 393 us MUL_MAT_ID(ffn_moe_down-50): 221 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 412 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 435 us MUL_MAT_ID(ffn_moe_down-52): 218 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 403 us MUL_MAT_ID(ffn_moe_down-53): 204 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 426 us MUL_MAT_ID(ffn_moe_down-54): 257 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 444 us MUL_MAT_ID(ffn_moe_down-55): 399 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 468 us MUL_MAT_ID(ffn_moe_down-56): 307 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 410 us MUL_MAT_ID(ffn_moe_down-57): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 456 us MUL_MAT_ID(ffn_moe_down-58): 209 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 435 us MUL_MAT_ID(ffn_moe_down-59): 181 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 456 us MUL_MAT_ID(ffn_moe_down-60): 280 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 425 us MUL_MAT_ID(ffn_moe_down-61): 199 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 477 us MUL_MAT_ID(ffn_moe_down-62): 223 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 405 us MUL_MAT_ID(ffn_moe_down-63): 187 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 489 us MUL_MAT_ID(ffn_moe_down-64): 314 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 462 us MUL_MAT_ID(ffn_moe_down-65): 168 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 506 us MUL_MAT_ID(ffn_moe_down-66): 267 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 755 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 363 us MUL_MAT_ID(ffn_moe_down-68): 184 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 409 us MUL_MAT_ID(ffn_moe_down-69): 334 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 461 us MUL_MAT_ID(ffn_moe_down-70): 230 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 430 us MUL_MAT_ID(ffn_moe_down-71): 205 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 414 us MUL_MAT_ID(ffn_moe_down-72): 192 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 403 us MUL_MAT_ID(ffn_moe_down-73): 204 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 513 us MUL_MAT_ID(ffn_moe_down-74): 428 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 387 us MUL_MAT_ID(ffn_moe_down-75): 179 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 409 us MUL_MAT_ID(ffn_moe_down-76): 434 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 410 us MUL_MAT_ID(ffn_moe_down-77): 282 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 480 us MUL_MAT_ID(ffn_moe_down-78): 314 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 403 us MUL_MAT_ID(ffn_moe_down-79): 215 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 430 us MUL_MAT_ID(ffn_moe_down-80): 290 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 416 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 454 us MUL_MAT_ID(ffn_moe_down-82): 611 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 410 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 455 us MUL_MAT_ID(ffn_moe_down-84): 246 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 421 us MUL_MAT_ID(ffn_moe_down-85): 240 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 396 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 444 us MUL_MAT_ID(ffn_moe_down-87): 210 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 468 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 428 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 407 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 354 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 64 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 394 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 428 us MUL_MAT_ID(ffn_moe_down-26): 240 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 430 us MUL_MAT_ID(ffn_moe_down-27): 259 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 464 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 621 us MUL_MAT_ID(ffn_moe_down-29): 339 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 434 us MUL_MAT_ID(ffn_moe_down-30): 275 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 484 us MUL_MAT_ID(ffn_moe_down-31): 332 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 427 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 410 us MUL_MAT_ID(ffn_moe_down-33): 244 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 428 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 440 us MUL_MAT_ID(ffn_moe_down-35): 216 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 499 us MUL_MAT_ID(ffn_moe_down-36): 218 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 404 us MUL_MAT_ID(ffn_moe_down-37): 296 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 425 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 385 us MUL_MAT_ID(ffn_moe_down-39): 291 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 427 us MUL_MAT_ID(ffn_moe_down-40): 237 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 510 us MUL_MAT_ID(ffn_moe_down-41): 209 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 510 us MUL_MAT_ID(ffn_moe_down-42): 222 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 721 us MUL_MAT_ID(ffn_moe_down-43): 288 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 412 us MUL_MAT_ID(ffn_moe_down-44): 184 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 514 us MUL_MAT_ID(ffn_moe_down-45): 238 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 516 us MUL_MAT_ID(ffn_moe_down-46): 208 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 462 us MUL_MAT_ID(ffn_moe_down-47): 239 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 468 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 366 us MUL_MAT_ID(ffn_moe_down-49): 170 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 427 us MUL_MAT_ID(ffn_moe_down-50): 242 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 418 us MUL_MAT_ID(ffn_moe_down-51): 231 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 457 us MUL_MAT_ID(ffn_moe_down-52): 210 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 388 us MUL_MAT_ID(ffn_moe_down-53): 221 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 372 us MUL_MAT_ID(ffn_moe_down-54): 562 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 363 us MUL_MAT_ID(ffn_moe_down-55): 438 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 460 us MUL_MAT_ID(ffn_moe_down-56): 263 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 417 us MUL_MAT_ID(ffn_moe_down-57): 538 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 551 us MUL_MAT_ID(ffn_moe_down-58): 202 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 474 us MUL_MAT_ID(ffn_moe_down-59): 238 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 466 us MUL_MAT_ID(ffn_moe_down-60): 438 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 487 us MUL_MAT_ID(ffn_moe_down-61): 194 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 388 us MUL_MAT_ID(ffn_moe_down-62): 259 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 399 us MUL_MAT_ID(ffn_moe_down-63): 178 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 436 us MUL_MAT_ID(ffn_moe_down-64): 268 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 403 us MUL_MAT_ID(ffn_moe_down-65): 253 us ggml_barrier(...): 334 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 387 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 474 us MUL_MAT_ID(ffn_moe_down-67): 214 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 540 us MUL_MAT_ID(ffn_moe_down-68): 260 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 603 us MUL_MAT_ID(ffn_moe_down-69): 443 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 430 us MUL_MAT_ID(ffn_moe_down-70): 219 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 425 us MUL_MAT_ID(ffn_moe_down-71): 238 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 496 us MUL_MAT_ID(ffn_moe_down-72): 449 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 526 us MUL_MAT_ID(ffn_moe_down-73): 297 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 411 us MUL_MAT_ID(ffn_moe_down-74): 331 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 473 us MUL_MAT_ID(ffn_moe_down-75): 209 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 490 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 472 us MUL_MAT_ID(ffn_moe_down-77): 247 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 410 us MUL_MAT_ID(ffn_moe_down-78): 302 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 471 us MUL_MAT_ID(ffn_moe_down-79): 236 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 369 us MUL_MAT_ID(ffn_moe_down-80): 216 us ggml_barrier(...): 322 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 538 us MUL_MAT_ID(ffn_moe_down-81): 211 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 488 us MUL_MAT_ID(ffn_moe_down-82): 311 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 384 us MUL_MAT_ID(ffn_moe_down-83): 268 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 414 us MUL_MAT_ID(ffn_moe_down-84): 224 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 482 us MUL_MAT_ID(ffn_moe_down-85): 251 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 444 us MUL_MAT_ID(ffn_moe_down-86): 208 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 405 us MUL_MAT_ID(ffn_moe_down-87): 215 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 549 us MUL_MAT_ID(ffn_moe_down-88): 200 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 392 us MUL_MAT_ID(ffn_moe_down-89): 206 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 428 us MUL_MAT_ID(ffn_moe_down-90): 210 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 382 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 52 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 385 us MUL_MAT_ID(ffn_moe_down-25): 232 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 373 us MUL_MAT_ID(ffn_moe_down-26): 248 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 462 us MUL_MAT_ID(ffn_moe_down-27): 190 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 552 us MUL_MAT_ID(ffn_moe_down-28): 223 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 445 us MUL_MAT_ID(ffn_moe_down-29): 311 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 433 us MUL_MAT_ID(ffn_moe_down-30): 225 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 518 us MUL_MAT_ID(ffn_moe_down-31): 286 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 480 us MUL_MAT_ID(ffn_moe_down-32): 223 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 423 us MUL_MAT_ID(ffn_moe_down-33): 706 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 434 us MUL_MAT_ID(ffn_moe_down-34): 231 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 466 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 569 us MUL_MAT_ID(ffn_moe_down-36): 269 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 382 us MUL_MAT_ID(ffn_moe_down-37): 207 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 413 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 439 us MUL_MAT_ID(ffn_moe_down-39): 226 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 459 us MUL_MAT_ID(ffn_moe_down-40): 456 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 466 us MUL_MAT_ID(ffn_moe_down-41): 224 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 513 us MUL_MAT_ID(ffn_moe_down-42): 224 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 372 us MUL_MAT_ID(ffn_moe_down-43): 161 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 473 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 580 us MUL_MAT_ID(ffn_moe_down-45): 194 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 469 us MUL_MAT_ID(ffn_moe_down-46): 375 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 428 us MUL_MAT_ID(ffn_moe_down-47): 535 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 379 us MUL_MAT_ID(ffn_moe_down-48): 334 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 372 us MUL_MAT_ID(ffn_moe_down-49): 286 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 419 us MUL_MAT_ID(ffn_moe_down-50): 216 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 533 us MUL_MAT_ID(ffn_moe_down-51): 177 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 376 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 461 us MUL_MAT_ID(ffn_moe_down-53): 283 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 406 us MUL_MAT_ID(ffn_moe_down-54): 197 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 564 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 510 us MUL_MAT_ID(ffn_moe_down-56): 273 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 438 us MUL_MAT_ID(ffn_moe_down-57): 229 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 397 us MUL_MAT_ID(ffn_moe_down-58): 186 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 402 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 393 us MUL_MAT_ID(ffn_moe_down-60): 242 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 467 us MUL_MAT_ID(ffn_moe_down-61): 223 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 474 us MUL_MAT_ID(ffn_moe_down-62): 209 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 412 us MUL_MAT_ID(ffn_moe_down-63): 299 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 474 us MUL_MAT_ID(ffn_moe_down-64): 313 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 419 us MUL_MAT_ID(ffn_moe_down-65): 198 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 396 us MUL_MAT_ID(ffn_moe_down-66): 251 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 471 us MUL_MAT_ID(ffn_moe_down-67): 322 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 459 us MUL_MAT_ID(ffn_moe_down-68): 277 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 423 us MUL_MAT_ID(ffn_moe_down-69): 283 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 807 us MUL_MAT_ID(ffn_moe_down-70): 393 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 466 us MUL_MAT_ID(ffn_moe_down-71): 235 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 418 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 417 us MUL_MAT_ID(ffn_moe_down-73): 318 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 413 us MUL_MAT_ID(ffn_moe_down-74): 205 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 373 us MUL_MAT_ID(ffn_moe_down-75): 186 us ggml_barrier(...): 371 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 485 us MUL_MAT_ID(ffn_moe_down-76): 319 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 407 us MUL_MAT_ID(ffn_moe_down-77): 232 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 401 us MUL_MAT_ID(ffn_moe_down-78): 295 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 482 us MUL_MAT_ID(ffn_moe_down-79): 278 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 442 us MUL_MAT_ID(ffn_moe_down-80): 243 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 518 us MUL_MAT_ID(ffn_moe_down-81): 269 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 403 us MUL_MAT_ID(ffn_moe_down-82): 307 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 540 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 479 us MUL_MAT_ID(ffn_moe_down-84): 265 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 401 us MUL_MAT_ID(ffn_moe_down-85): 521 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 418 us MUL_MAT_ID(ffn_moe_down-86): 301 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 429 us MUL_MAT_ID(ffn_moe_down-87): 277 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 392 us MUL_MAT_ID(ffn_moe_down-88): 211 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 454 us MUL_MAT_ID(ffn_moe_down-89): 210 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 406 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 400 us MUL_MAT_ID(ffn_moe_down-91): 229 us ggml_barrier(...): 25 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 435 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 408 us MUL_MAT_ID(ffn_moe_down-26): 242 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 421 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 423 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 396 us MUL_MAT_ID(ffn_moe_down-29): 549 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 475 us MUL_MAT_ID(ffn_moe_down-30): 234 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 410 us MUL_MAT_ID(ffn_moe_down-31): 211 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 437 us MUL_MAT_ID(ffn_moe_down-32): 205 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 471 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 452 us MUL_MAT_ID(ffn_moe_down-34): 283 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 472 us MUL_MAT_ID(ffn_moe_down-35): 212 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 429 us MUL_MAT_ID(ffn_moe_down-36): 417 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 413 us MUL_MAT_ID(ffn_moe_down-37): 317 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 435 us MUL_MAT_ID(ffn_moe_down-38): 237 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 479 us MUL_MAT_ID(ffn_moe_down-39): 242 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 407 us MUL_MAT_ID(ffn_moe_down-40): 715 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 342 us MUL_MAT_ID(ffn_moe_down-41): 296 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 398 us MUL_MAT_ID(ffn_moe_down-42): 181 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 338 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 382 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 353 us MUL_MAT_ID(ffn_moe_down-45): 286 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 386 us MUL_MAT_ID(ffn_moe_down-46): 490 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 448 us MUL_MAT_ID(ffn_moe_down-47): 244 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 424 us MUL_MAT_ID(ffn_moe_down-48): 235 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 385 us MUL_MAT_ID(ffn_moe_down-49): 388 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 453 us MUL_MAT_ID(ffn_moe_down-50): 293 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 418 us MUL_MAT_ID(ffn_moe_down-51): 212 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 435 us MUL_MAT_ID(ffn_moe_down-52): 188 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 412 us MUL_MAT_ID(ffn_moe_down-53): 209 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 500 us MUL_MAT_ID(ffn_moe_down-54): 227 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 694 us MUL_MAT_ID(ffn_moe_down-55): 387 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 436 us MUL_MAT_ID(ffn_moe_down-56): 253 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 415 us MUL_MAT_ID(ffn_moe_down-57): 181 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 411 us MUL_MAT_ID(ffn_moe_down-58): 517 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 380 us MUL_MAT_ID(ffn_moe_down-59): 463 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 397 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 441 us MUL_MAT_ID(ffn_moe_down-61): 199 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 376 us MUL_MAT_ID(ffn_moe_down-62): 206 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 516 us MUL_MAT_ID(ffn_moe_down-63): 404 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 473 us MUL_MAT_ID(ffn_moe_down-64): 476 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 426 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 441 us MUL_MAT_ID(ffn_moe_down-66): 218 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 417 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 499 us MUL_MAT_ID(ffn_moe_down-68): 203 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 395 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 375 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 420 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 353 us MUL_MAT_ID(ffn_moe_down-72): 240 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 446 us MUL_MAT_ID(ffn_moe_down-73): 719 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 407 us MUL_MAT_ID(ffn_moe_down-74): 217 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 655 us MUL_MAT_ID(ffn_moe_down-75): 218 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 393 us MUL_MAT_ID(ffn_moe_down-76): 187 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 425 us MUL_MAT_ID(ffn_moe_down-77): 190 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 480 us MUL_MAT_ID(ffn_moe_down-78): 295 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 441 us MUL_MAT_ID(ffn_moe_down-79): 520 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 399 us MUL_MAT_ID(ffn_moe_down-80): 203 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 356 us MUL_MAT_ID(ffn_moe_down-81): 213 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 421 us MUL_MAT_ID(ffn_moe_down-82): 206 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 403 us MUL_MAT_ID(ffn_moe_down-83): 236 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 409 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 379 us MUL_MAT_ID(ffn_moe_down-85): 739 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 489 us MUL_MAT_ID(ffn_moe_down-86): 252 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 526 us MUL_MAT_ID(ffn_moe_down-87): 331 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 361 us MUL_MAT_ID(ffn_moe_down-88): 212 us ggml_barrier(...): 347 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 402 us MUL_MAT_ID(ffn_moe_down-89): 223 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 423 us MUL_MAT_ID(ffn_moe_down-90): 242 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 397 us MUL_MAT_ID(ffn_moe_down-91): 251 us ggml_barrier(...): 260 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 395 us MUL_MAT_ID(ffn_moe_down-25): 204 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 384 us MUL_MAT_ID(ffn_moe_down-26): 200 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 374 us MUL_MAT_ID(ffn_moe_down-27): 180 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 511 us MUL_MAT_ID(ffn_moe_down-28): 190 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 376 us MUL_MAT_ID(ffn_moe_down-29): 225 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 406 us MUL_MAT_ID(ffn_moe_down-30): 257 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 394 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 426 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 496 us MUL_MAT_ID(ffn_moe_down-33): 191 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 459 us MUL_MAT_ID(ffn_moe_down-34): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 389 us MUL_MAT_ID(ffn_moe_down-35): 223 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 456 us MUL_MAT_ID(ffn_moe_down-36): 210 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 378 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 414 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 438 us MUL_MAT_ID(ffn_moe_down-39): 247 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 387 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 568 us MUL_MAT_ID(ffn_moe_down-41): 336 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 366 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 404 us MUL_MAT_ID(ffn_moe_down-43): 189 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 396 us MUL_MAT_ID(ffn_moe_down-44): 433 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 526 us MUL_MAT_ID(ffn_moe_down-45): 179 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 366 us MUL_MAT_ID(ffn_moe_down-46): 209 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 394 us MUL_MAT_ID(ffn_moe_down-47): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 389 us MUL_MAT_ID(ffn_moe_down-48): 232 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 420 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 376 us MUL_MAT_ID(ffn_moe_down-50): 223 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 394 us MUL_MAT_ID(ffn_moe_down-51): 202 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 347 us MUL_MAT_ID(ffn_moe_down-52): 206 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 399 us MUL_MAT_ID(ffn_moe_down-53): 423 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 485 us MUL_MAT_ID(ffn_moe_down-54): 225 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 496 us MUL_MAT_ID(ffn_moe_down-55): 185 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 408 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 358 us MUL_MAT_ID(ffn_moe_down-57): 434 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 509 us MUL_MAT_ID(ffn_moe_down-58): 166 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 504 us MUL_MAT_ID(ffn_moe_down-59): 247 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 402 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 340 us MUL_MAT_ID(ffn_moe_down-61): 199 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 444 us MUL_MAT_ID(ffn_moe_down-62): 192 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 437 us MUL_MAT_ID(ffn_moe_down-63): 493 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 410 us MUL_MAT_ID(ffn_moe_down-64): 227 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 415 us MUL_MAT_ID(ffn_moe_down-65): 186 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 401 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 360 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 413 us MUL_MAT_ID(ffn_moe_down-68): 226 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 397 us MUL_MAT_ID(ffn_moe_down-69): 219 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 441 us MUL_MAT_ID(ffn_moe_down-70): 172 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 419 us MUL_MAT_ID(ffn_moe_down-71): 205 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 467 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 405 us MUL_MAT_ID(ffn_moe_down-73): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 518 us MUL_MAT_ID(ffn_moe_down-74): 183 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 376 us MUL_MAT_ID(ffn_moe_down-75): 207 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 563 us MUL_MAT_ID(ffn_moe_down-76): 332 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 371 us MUL_MAT_ID(ffn_moe_down-77): 189 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 417 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 393 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 369 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 440 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 429 us MUL_MAT_ID(ffn_moe_down-82): 218 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 775 us MUL_MAT_ID(ffn_moe_down-83): 378 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 391 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 502 us MUL_MAT_ID(ffn_moe_down-85): 254 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 376 us MUL_MAT_ID(ffn_moe_down-86): 183 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 412 us MUL_MAT_ID(ffn_moe_down-87): 235 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 404 us MUL_MAT_ID(ffn_moe_down-88): 429 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 429 us MUL_MAT_ID(ffn_moe_down-89): 186 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 424 us MUL_MAT_ID(ffn_moe_down-90): 257 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 433 us MUL_MAT_ID(ffn_moe_down-91): 254 us ggml_barrier(...): 213 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 384 us MUL_MAT_ID(ffn_moe_down-25): 250 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 384 us MUL_MAT_ID(ffn_moe_down-26): 257 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 401 us MUL_MAT_ID(ffn_moe_down-27): 238 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 631 us MUL_MAT_ID(ffn_moe_down-28): 708 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 490 us MUL_MAT_ID(ffn_moe_down-29): 227 us ggml_barrier(...): 352 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 416 us MUL_MAT_ID(ffn_moe_down-30): 312 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 460 us MUL_MAT_ID(ffn_moe_down-31): 253 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 414 us MUL_MAT_ID(ffn_moe_down-32): 186 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 370 us MUL_MAT_ID(ffn_moe_down-33): 230 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 407 us MUL_MAT_ID(ffn_moe_down-34): 224 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 354 us MUL_MAT_ID(ffn_moe_down-35): 184 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 404 us MUL_MAT_ID(ffn_moe_down-36): 225 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 398 us MUL_MAT_ID(ffn_moe_down-37): 259 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 467 us MUL_MAT_ID(ffn_moe_down-38): 506 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 472 us MUL_MAT_ID(ffn_moe_down-39): 663 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 378 us MUL_MAT_ID(ffn_moe_down-40): 235 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 417 us MUL_MAT_ID(ffn_moe_down-41): 203 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 407 us MUL_MAT_ID(ffn_moe_down-42): 232 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 384 us MUL_MAT_ID(ffn_moe_down-43): 209 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 521 us MUL_MAT_ID(ffn_moe_down-44): 414 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 484 us MUL_MAT_ID(ffn_moe_down-45): 198 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 510 us MUL_MAT_ID(ffn_moe_down-46): 217 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 391 us MUL_MAT_ID(ffn_moe_down-47): 213 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 455 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 395 us MUL_MAT_ID(ffn_moe_down-49): 249 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 416 us MUL_MAT_ID(ffn_moe_down-50): 312 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 419 us MUL_MAT_ID(ffn_moe_down-51): 211 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 387 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 384 us MUL_MAT_ID(ffn_moe_down-53): 178 us ggml_barrier(...): 372 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 394 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 425 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 508 us MUL_MAT_ID(ffn_moe_down-56): 209 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 439 us MUL_MAT_ID(ffn_moe_down-57): 328 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 380 us MUL_MAT_ID(ffn_moe_down-58): 176 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 429 us MUL_MAT_ID(ffn_moe_down-59): 356 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 493 us MUL_MAT_ID(ffn_moe_down-60): 252 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 376 us MUL_MAT_ID(ffn_moe_down-61): 236 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 479 us MUL_MAT_ID(ffn_moe_down-62): 157 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 422 us MUL_MAT_ID(ffn_moe_down-63): 210 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 446 us MUL_MAT_ID(ffn_moe_down-64): 286 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 439 us MUL_MAT_ID(ffn_moe_down-65): 250 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 409 us MUL_MAT_ID(ffn_moe_down-66): 218 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 387 us MUL_MAT_ID(ffn_moe_down-67): 286 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 420 us MUL_MAT_ID(ffn_moe_down-68): 227 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 404 us MUL_MAT_ID(ffn_moe_down-69): 189 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 433 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 429 us MUL_MAT_ID(ffn_moe_down-71): 385 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 371 us MUL_MAT_ID(ffn_moe_down-72): 186 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 520 us MUL_MAT_ID(ffn_moe_down-73): 255 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 381 us MUL_MAT_ID(ffn_moe_down-74): 314 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 428 us MUL_MAT_ID(ffn_moe_down-75): 208 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 377 us MUL_MAT_ID(ffn_moe_down-76): 263 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 435 us MUL_MAT_ID(ffn_moe_down-77): 304 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 412 us MUL_MAT_ID(ffn_moe_down-78): 263 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 388 us MUL_MAT_ID(ffn_moe_down-79): 409 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 461 us MUL_MAT_ID(ffn_moe_down-80): 355 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 410 us MUL_MAT_ID(ffn_moe_down-81): 255 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 455 us MUL_MAT_ID(ffn_moe_down-82): 288 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 482 us MUL_MAT_ID(ffn_moe_down-83): 238 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 519 us MUL_MAT_ID(ffn_moe_down-84): 517 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 517 us MUL_MAT_ID(ffn_moe_down-85): 313 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 366 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 402 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 416 us MUL_MAT_ID(ffn_moe_down-88): 378 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 395 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 382 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 780 us MUL_MAT_ID(ffn_moe_down-91): 258 us ggml_barrier(...): 154 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 440 us MUL_MAT_ID(ffn_moe_down-25): 317 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 565 us MUL_MAT_ID(ffn_moe_down-26): 219 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 444 us MUL_MAT_ID(ffn_moe_down-27): 183 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 470 us MUL_MAT_ID(ffn_moe_down-28): 208 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 407 us MUL_MAT_ID(ffn_moe_down-29): 169 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 383 us MUL_MAT_ID(ffn_moe_down-30): 393 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 460 us MUL_MAT_ID(ffn_moe_down-31): 223 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 419 us MUL_MAT_ID(ffn_moe_down-32): 309 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 450 us MUL_MAT_ID(ffn_moe_down-33): 227 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 374 us MUL_MAT_ID(ffn_moe_down-34): 203 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 479 us MUL_MAT_ID(ffn_moe_down-35): 244 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 448 us MUL_MAT_ID(ffn_moe_down-36): 312 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 398 us MUL_MAT_ID(ffn_moe_down-37): 224 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 459 us MUL_MAT_ID(ffn_moe_down-38): 251 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 409 us MUL_MAT_ID(ffn_moe_down-39): 254 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 410 us MUL_MAT_ID(ffn_moe_down-40): 245 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 404 us MUL_MAT_ID(ffn_moe_down-41): 377 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 420 us MUL_MAT_ID(ffn_moe_down-42): 367 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 408 us MUL_MAT_ID(ffn_moe_down-43): 291 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 478 us MUL_MAT_ID(ffn_moe_down-44): 254 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 392 us MUL_MAT_ID(ffn_moe_down-45): 240 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 571 us MUL_MAT_ID(ffn_moe_down-46): 202 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 436 us MUL_MAT_ID(ffn_moe_down-47): 520 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 370 us MUL_MAT_ID(ffn_moe_down-48): 238 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 444 us MUL_MAT_ID(ffn_moe_down-49): 258 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 485 us MUL_MAT_ID(ffn_moe_down-50): 186 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 499 us MUL_MAT_ID(ffn_moe_down-51): 332 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 498 us MUL_MAT_ID(ffn_moe_down-52): 289 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 528 us MUL_MAT_ID(ffn_moe_down-53): 229 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 419 us MUL_MAT_ID(ffn_moe_down-54): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 454 us MUL_MAT_ID(ffn_moe_down-55): 269 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 458 us MUL_MAT_ID(ffn_moe_down-56): 179 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 458 us MUL_MAT_ID(ffn_moe_down-57): 305 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 438 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 440 us MUL_MAT_ID(ffn_moe_down-59): 210 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 402 us MUL_MAT_ID(ffn_moe_down-60): 187 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 465 us MUL_MAT_ID(ffn_moe_down-61): 176 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 424 us MUL_MAT_ID(ffn_moe_down-62): 228 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 403 us MUL_MAT_ID(ffn_moe_down-63): 253 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 417 us MUL_MAT_ID(ffn_moe_down-64): 345 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 419 us MUL_MAT_ID(ffn_moe_down-65): 343 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 468 us MUL_MAT_ID(ffn_moe_down-66): 197 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 506 us MUL_MAT_ID(ffn_moe_down-67): 221 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 483 us MUL_MAT_ID(ffn_moe_down-68): 433 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 419 us MUL_MAT_ID(ffn_moe_down-69): 264 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 439 us MUL_MAT_ID(ffn_moe_down-70): 220 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 453 us MUL_MAT_ID(ffn_moe_down-71): 236 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 520 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 465 us MUL_MAT_ID(ffn_moe_down-73): 207 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 434 us MUL_MAT_ID(ffn_moe_down-74): 256 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 413 us MUL_MAT_ID(ffn_moe_down-75): 262 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 484 us MUL_MAT_ID(ffn_moe_down-76): 198 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 453 us MUL_MAT_ID(ffn_moe_down-77): 242 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 447 us MUL_MAT_ID(ffn_moe_down-78): 307 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 382 us MUL_MAT_ID(ffn_moe_down-79): 183 us ggml_barrier(...): 362 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 407 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 424 us MUL_MAT_ID(ffn_moe_down-81): 223 us ggml_barrier(...): 409 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 389 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 554 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 494 us MUL_MAT_ID(ffn_moe_down-84): 354 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 504 us MUL_MAT_ID(ffn_moe_down-85): 604 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 462 us MUL_MAT_ID(ffn_moe_down-86): 246 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 482 us MUL_MAT_ID(ffn_moe_down-87): 264 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 432 us MUL_MAT_ID(ffn_moe_down-88): 388 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 351 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 502 us MUL_MAT_ID(ffn_moe_down-90): 247 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 420 us MUL_MAT_ID(ffn_moe_down-91): 413 us ggml_barrier(...): 265 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 427 us MUL_MAT_ID(ffn_moe_down-25): 243 us ggml_barrier(...): 368 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 417 us MUL_MAT_ID(ffn_moe_down-26): 230 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 394 us MUL_MAT_ID(ffn_moe_down-27): 219 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 425 us MUL_MAT_ID(ffn_moe_down-28): 325 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 419 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 337 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 544 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 473 us MUL_MAT_ID(ffn_moe_down-31): 468 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 441 us MUL_MAT_ID(ffn_moe_down-32): 313 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 400 us MUL_MAT_ID(ffn_moe_down-33): 242 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 514 us MUL_MAT_ID(ffn_moe_down-34): 188 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 398 us MUL_MAT_ID(ffn_moe_down-35): 292 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 410 us MUL_MAT_ID(ffn_moe_down-36): 218 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 434 us MUL_MAT_ID(ffn_moe_down-37): 300 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 486 us MUL_MAT_ID(ffn_moe_down-38): 220 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 395 us MUL_MAT_ID(ffn_moe_down-39): 221 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 523 us MUL_MAT_ID(ffn_moe_down-40): 262 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 546 us MUL_MAT_ID(ffn_moe_down-41): 206 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 373 us MUL_MAT_ID(ffn_moe_down-42): 259 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 421 us MUL_MAT_ID(ffn_moe_down-43): 239 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 490 us MUL_MAT_ID(ffn_moe_down-44): 222 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 430 us MUL_MAT_ID(ffn_moe_down-45): 320 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 473 us MUL_MAT_ID(ffn_moe_down-46): 431 us ggml_barrier(...): 361 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 531 us MUL_MAT_ID(ffn_moe_down-47): 260 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 419 us MUL_MAT_ID(ffn_moe_down-48): 181 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 433 us MUL_MAT_ID(ffn_moe_down-49): 297 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 410 us MUL_MAT_ID(ffn_moe_down-50): 214 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 420 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 433 us MUL_MAT_ID(ffn_moe_down-52): 352 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 443 us MUL_MAT_ID(ffn_moe_down-53): 272 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 425 us MUL_MAT_ID(ffn_moe_down-54): 202 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 422 us MUL_MAT_ID(ffn_moe_down-55): 232 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 474 us MUL_MAT_ID(ffn_moe_down-56): 461 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 448 us MUL_MAT_ID(ffn_moe_down-57): 272 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 374 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 419 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 526 us MUL_MAT_ID(ffn_moe_down-60): 201 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 422 us MUL_MAT_ID(ffn_moe_down-61): 218 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 405 us MUL_MAT_ID(ffn_moe_down-62): 243 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 484 us MUL_MAT_ID(ffn_moe_down-63): 323 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 413 us MUL_MAT_ID(ffn_moe_down-64): 293 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 505 us MUL_MAT_ID(ffn_moe_down-65): 214 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 377 us MUL_MAT_ID(ffn_moe_down-66): 268 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 426 us MUL_MAT_ID(ffn_moe_down-67): 263 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 431 us MUL_MAT_ID(ffn_moe_down-68): 281 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 438 us MUL_MAT_ID(ffn_moe_down-69): 241 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 410 us MUL_MAT_ID(ffn_moe_down-70): 205 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 533 us MUL_MAT_ID(ffn_moe_down-71): 241 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 396 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 400 us MUL_MAT_ID(ffn_moe_down-73): 199 us ggml_barrier(...): 353 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 457 us MUL_MAT_ID(ffn_moe_down-74): 620 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 430 us MUL_MAT_ID(ffn_moe_down-75): 238 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 432 us MUL_MAT_ID(ffn_moe_down-76): 201 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 460 us MUL_MAT_ID(ffn_moe_down-77): 227 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 435 us MUL_MAT_ID(ffn_moe_down-78): 242 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 425 us MUL_MAT_ID(ffn_moe_down-79): 251 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 481 us MUL_MAT_ID(ffn_moe_down-80): 181 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 401 us MUL_MAT_ID(ffn_moe_down-81): 257 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 429 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 530 us MUL_MAT_ID(ffn_moe_down-83): 329 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 491 us MUL_MAT_ID(ffn_moe_down-84): 283 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 625 us MUL_MAT_ID(ffn_moe_down-85): 270 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 528 us MUL_MAT_ID(ffn_moe_down-86): 205 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 428 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 506 us MUL_MAT_ID(ffn_moe_down-88): 425 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 411 us MUL_MAT_ID(ffn_moe_down-89): 185 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 398 us MUL_MAT_ID(ffn_moe_down-90): 279 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 403 us MUL_MAT_ID(ffn_moe_down-91): 274 us ggml_barrier(...): 228 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 459 us MUL_MAT_ID(ffn_moe_down-25): 454 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 387 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 361 us MUL_MAT_ID(ffn_moe_down-27): 215 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 554 us MUL_MAT_ID(ffn_moe_down-28): 263 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 426 us MUL_MAT_ID(ffn_moe_down-29): 272 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 447 us MUL_MAT_ID(ffn_moe_down-30): 596 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 357 us MUL_MAT_ID(ffn_moe_down-31): 205 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 519 us MUL_MAT_ID(ffn_moe_down-32): 177 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 433 us MUL_MAT_ID(ffn_moe_down-33): 184 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 500 us MUL_MAT_ID(ffn_moe_down-34): 427 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 409 us MUL_MAT_ID(ffn_moe_down-35): 209 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 399 us MUL_MAT_ID(ffn_moe_down-36): 212 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 376 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 423 us MUL_MAT_ID(ffn_moe_down-38): 235 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 377 us MUL_MAT_ID(ffn_moe_down-39): 386 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 486 us MUL_MAT_ID(ffn_moe_down-40): 251 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 423 us MUL_MAT_ID(ffn_moe_down-41): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 514 us MUL_MAT_ID(ffn_moe_down-42): 177 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 455 us MUL_MAT_ID(ffn_moe_down-43): 201 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 521 us MUL_MAT_ID(ffn_moe_down-44): 285 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 388 us MUL_MAT_ID(ffn_moe_down-45): 188 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 375 us MUL_MAT_ID(ffn_moe_down-46): 223 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 435 us MUL_MAT_ID(ffn_moe_down-47): 238 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 376 us MUL_MAT_ID(ffn_moe_down-48): 290 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 442 us MUL_MAT_ID(ffn_moe_down-49): 571 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 214 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 547 us MUL_MAT_ID(ffn_moe_down-51): 234 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 499 us MUL_MAT_ID(ffn_moe_down-52): 257 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 412 us MUL_MAT_ID(ffn_moe_down-53): 190 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 396 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 387 us MUL_MAT_ID(ffn_moe_down-55): 613 us ggml_barrier(...): 353 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 593 us MUL_MAT_ID(ffn_moe_down-56): 234 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 440 us MUL_MAT_ID(ffn_moe_down-57): 739 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 353 us MUL_MAT_ID(ffn_moe_down-58): 330 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 388 us MUL_MAT_ID(ffn_moe_down-59): 409 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 354 us MUL_MAT_ID(ffn_moe_down-60): 209 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 444 us MUL_MAT_ID(ffn_moe_down-61): 269 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 447 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 416 us MUL_MAT_ID(ffn_moe_down-63): 254 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 433 us MUL_MAT_ID(ffn_moe_down-64): 200 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 420 us MUL_MAT_ID(ffn_moe_down-65): 296 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 420 us MUL_MAT_ID(ffn_moe_down-66): 226 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 484 us MUL_MAT_ID(ffn_moe_down-67): 204 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 428 us MUL_MAT_ID(ffn_moe_down-68): 187 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 420 us MUL_MAT_ID(ffn_moe_down-69): 206 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 391 us MUL_MAT_ID(ffn_moe_down-70): 207 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 389 us MUL_MAT_ID(ffn_moe_down-71): 311 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 415 us MUL_MAT_ID(ffn_moe_down-72): 223 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 404 us MUL_MAT_ID(ffn_moe_down-73): 325 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 471 us MUL_MAT_ID(ffn_moe_down-74): 208 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 449 us MUL_MAT_ID(ffn_moe_down-75): 393 us ggml_barrier(...): 362 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 421 us MUL_MAT_ID(ffn_moe_down-76): 219 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 447 us MUL_MAT_ID(ffn_moe_down-77): 226 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 408 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 440 us MUL_MAT_ID(ffn_moe_down-79): 252 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 396 us MUL_MAT_ID(ffn_moe_down-80): 175 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 380 us MUL_MAT_ID(ffn_moe_down-81): 211 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 457 us MUL_MAT_ID(ffn_moe_down-82): 180 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 471 us MUL_MAT_ID(ffn_moe_down-83): 276 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 361 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 373 us MUL_MAT_ID(ffn_moe_down-85): 281 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 517 us MUL_MAT_ID(ffn_moe_down-86): 287 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 695 us MUL_MAT_ID(ffn_moe_down-87): 256 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 395 us MUL_MAT_ID(ffn_moe_down-88): 216 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 377 us MUL_MAT_ID(ffn_moe_down-89): 195 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 467 us MUL_MAT_ID(ffn_moe_down-90): 304 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 475 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 133 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 372 us MUL_MAT_ID(ffn_moe_down-25): 196 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 400 us MUL_MAT_ID(ffn_moe_down-26): 204 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 364 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 414 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 414 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 406 us MUL_MAT_ID(ffn_moe_down-30): 185 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 376 us MUL_MAT_ID(ffn_moe_down-31): 199 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 501 us MUL_MAT_ID(ffn_moe_down-32): 227 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 377 us MUL_MAT_ID(ffn_moe_down-33): 319 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 475 us MUL_MAT_ID(ffn_moe_down-34): 207 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 396 us MUL_MAT_ID(ffn_moe_down-35): 230 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 380 us MUL_MAT_ID(ffn_moe_down-36): 195 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 438 us MUL_MAT_ID(ffn_moe_down-37): 251 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 380 us MUL_MAT_ID(ffn_moe_down-38): 199 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 487 us MUL_MAT_ID(ffn_moe_down-39): 292 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 410 us MUL_MAT_ID(ffn_moe_down-40): 238 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 462 us MUL_MAT_ID(ffn_moe_down-41): 438 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 391 us MUL_MAT_ID(ffn_moe_down-42): 417 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 406 us MUL_MAT_ID(ffn_moe_down-43): 186 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 381 us MUL_MAT_ID(ffn_moe_down-44): 266 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 353 us MUL_MAT_ID(ffn_moe_down-45): 566 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 428 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 402 us MUL_MAT_ID(ffn_moe_down-47): 415 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 485 us MUL_MAT_ID(ffn_moe_down-48): 218 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 381 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 374 us MUL_MAT_ID(ffn_moe_down-50): 210 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 395 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 433 us MUL_MAT_ID(ffn_moe_down-52): 226 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 442 us MUL_MAT_ID(ffn_moe_down-53): 225 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 422 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 415 us MUL_MAT_ID(ffn_moe_down-55): 405 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 459 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 393 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 487 us MUL_MAT_ID(ffn_moe_down-58): 220 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 430 us MUL_MAT_ID(ffn_moe_down-59): 304 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 397 us MUL_MAT_ID(ffn_moe_down-60): 441 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 434 us MUL_MAT_ID(ffn_moe_down-61): 453 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 401 us MUL_MAT_ID(ffn_moe_down-62): 276 us ggml_barrier(...): 385 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 423 us MUL_MAT_ID(ffn_moe_down-63): 223 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 419 us MUL_MAT_ID(ffn_moe_down-64): 217 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 393 us MUL_MAT_ID(ffn_moe_down-65): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 458 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 417 us MUL_MAT_ID(ffn_moe_down-67): 209 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 435 us MUL_MAT_ID(ffn_moe_down-68): 279 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 433 us MUL_MAT_ID(ffn_moe_down-69): 227 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 431 us MUL_MAT_ID(ffn_moe_down-70): 272 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 487 us MUL_MAT_ID(ffn_moe_down-71): 182 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 441 us MUL_MAT_ID(ffn_moe_down-72): 254 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 446 us MUL_MAT_ID(ffn_moe_down-73): 237 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 530 us MUL_MAT_ID(ffn_moe_down-74): 244 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 397 us MUL_MAT_ID(ffn_moe_down-75): 187 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 459 us MUL_MAT_ID(ffn_moe_down-76): 215 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 423 us MUL_MAT_ID(ffn_moe_down-77): 249 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 410 us MUL_MAT_ID(ffn_moe_down-78): 286 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 408 us MUL_MAT_ID(ffn_moe_down-79): 207 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 452 us MUL_MAT_ID(ffn_moe_down-80): 239 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 430 us MUL_MAT_ID(ffn_moe_down-81): 217 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 440 us MUL_MAT_ID(ffn_moe_down-82): 255 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 391 us MUL_MAT_ID(ffn_moe_down-83): 261 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 495 us MUL_MAT_ID(ffn_moe_down-84): 271 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 503 us MUL_MAT_ID(ffn_moe_down-85): 253 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 411 us MUL_MAT_ID(ffn_moe_down-86): 209 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 489 us MUL_MAT_ID(ffn_moe_down-87): 344 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 386 us MUL_MAT_ID(ffn_moe_down-88): 220 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 493 us MUL_MAT_ID(ffn_moe_down-89): 265 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 733 us MUL_MAT_ID(ffn_moe_down-90): 410 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 425 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 241 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 427 us MUL_MAT_ID(ffn_moe_down-25): 431 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 460 us MUL_MAT_ID(ffn_moe_down-26): 175 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 422 us MUL_MAT_ID(ffn_moe_down-27): 243 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 438 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 426 us MUL_MAT_ID(ffn_moe_down-29): 540 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 435 us MUL_MAT_ID(ffn_moe_down-30): 213 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 412 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 625 us MUL_MAT_ID(ffn_moe_down-32): 316 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 350 us MUL_MAT_ID(ffn_moe_down-33): 408 us ggml_barrier(...): 322 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 396 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 429 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 404 us MUL_MAT_ID(ffn_moe_down-36): 203 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 443 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 270 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 415 us MUL_MAT_ID(ffn_moe_down-39): 316 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 444 us MUL_MAT_ID(ffn_moe_down-40): 241 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 431 us MUL_MAT_ID(ffn_moe_down-41): 209 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 374 us MUL_MAT_ID(ffn_moe_down-42): 183 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 407 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 592 us MUL_MAT_ID(ffn_moe_down-44): 205 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 383 us MUL_MAT_ID(ffn_moe_down-45): 229 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 355 us MUL_MAT_ID(ffn_moe_down-46): 173 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 381 us MUL_MAT_ID(ffn_moe_down-47): 237 us ggml_barrier(...): 340 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 418 us MUL_MAT_ID(ffn_moe_down-48): 312 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 460 us MUL_MAT_ID(ffn_moe_down-49): 219 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 401 us MUL_MAT_ID(ffn_moe_down-50): 574 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 408 us MUL_MAT_ID(ffn_moe_down-51): 431 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 380 us MUL_MAT_ID(ffn_moe_down-52): 242 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 463 us MUL_MAT_ID(ffn_moe_down-53): 294 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 479 us MUL_MAT_ID(ffn_moe_down-54): 189 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 413 us MUL_MAT_ID(ffn_moe_down-55): 207 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 483 us MUL_MAT_ID(ffn_moe_down-56): 179 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 461 us MUL_MAT_ID(ffn_moe_down-57): 161 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 485 us MUL_MAT_ID(ffn_moe_down-58): 173 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 477 us MUL_MAT_ID(ffn_moe_down-59): 221 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 618 us MUL_MAT_ID(ffn_moe_down-60): 227 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 432 us MUL_MAT_ID(ffn_moe_down-61): 299 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 566 us MUL_MAT_ID(ffn_moe_down-62): 177 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 403 us MUL_MAT_ID(ffn_moe_down-63): 204 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 517 us MUL_MAT_ID(ffn_moe_down-64): 239 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 439 us MUL_MAT_ID(ffn_moe_down-65): 169 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 501 us MUL_MAT_ID(ffn_moe_down-66): 185 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 420 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 357 us MUL_MAT_ID(ffn_moe_down-68): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 412 us MUL_MAT_ID(ffn_moe_down-69): 224 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 406 us MUL_MAT_ID(ffn_moe_down-70): 214 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 425 us MUL_MAT_ID(ffn_moe_down-71): 238 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 471 us MUL_MAT_ID(ffn_moe_down-72): 211 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 378 us MUL_MAT_ID(ffn_moe_down-73): 650 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 491 us MUL_MAT_ID(ffn_moe_down-74): 219 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 381 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 554 us MUL_MAT_ID(ffn_moe_down-76): 294 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 442 us MUL_MAT_ID(ffn_moe_down-77): 212 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 400 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 458 us MUL_MAT_ID(ffn_moe_down-79): 188 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 455 us MUL_MAT_ID(ffn_moe_down-80): 203 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 391 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 408 us MUL_MAT_ID(ffn_moe_down-82): 236 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 409 us MUL_MAT_ID(ffn_moe_down-83): 278 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 417 us MUL_MAT_ID(ffn_moe_down-84): 656 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 419 us MUL_MAT_ID(ffn_moe_down-85): 230 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 458 us MUL_MAT_ID(ffn_moe_down-86): 444 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 537 us MUL_MAT_ID(ffn_moe_down-87): 222 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 412 us MUL_MAT_ID(ffn_moe_down-88): 186 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 357 us MUL_MAT_ID(ffn_moe_down-89): 203 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 455 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 460 us MUL_MAT_ID(ffn_moe_down-91): 303 us ggml_barrier(...): 259 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 380 us MUL_MAT_ID(ffn_moe_down-25): 228 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 398 us MUL_MAT_ID(ffn_moe_down-26): 212 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 391 us MUL_MAT_ID(ffn_moe_down-27): 250 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 414 us MUL_MAT_ID(ffn_moe_down-28): 236 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 444 us MUL_MAT_ID(ffn_moe_down-29): 209 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 406 us MUL_MAT_ID(ffn_moe_down-30): 205 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 436 us MUL_MAT_ID(ffn_moe_down-31): 208 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 414 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 384 us MUL_MAT_ID(ffn_moe_down-33): 593 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 414 us MUL_MAT_ID(ffn_moe_down-34): 215 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 336 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 370 us MUL_MAT_ID(ffn_moe_down-36): 201 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 321 us MUL_MAT_ID(ffn_moe_down-37): 196 us ggml_barrier(...): 401 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 393 us MUL_MAT_ID(ffn_moe_down-38): 212 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 365 us MUL_MAT_ID(ffn_moe_down-39): 244 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 402 us MUL_MAT_ID(ffn_moe_down-40): 346 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 544 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 492 us MUL_MAT_ID(ffn_moe_down-42): 230 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 405 us MUL_MAT_ID(ffn_moe_down-43): 706 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 442 us MUL_MAT_ID(ffn_moe_down-44): 204 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 635 us MUL_MAT_ID(ffn_moe_down-45): 218 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 377 us MUL_MAT_ID(ffn_moe_down-46): 172 us ggml_barrier(...): 357 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 413 us MUL_MAT_ID(ffn_moe_down-47): 209 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 433 us MUL_MAT_ID(ffn_moe_down-48): 291 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 404 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 360 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 401 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 414 us MUL_MAT_ID(ffn_moe_down-52): 216 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 422 us MUL_MAT_ID(ffn_moe_down-53): 412 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 434 us MUL_MAT_ID(ffn_moe_down-54): 341 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 400 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 459 us MUL_MAT_ID(ffn_moe_down-56): 261 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 457 us MUL_MAT_ID(ffn_moe_down-57): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 403 us MUL_MAT_ID(ffn_moe_down-58): 198 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 446 us MUL_MAT_ID(ffn_moe_down-59): 238 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 413 us MUL_MAT_ID(ffn_moe_down-60): 185 us ggml_barrier(...): 327 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 426 us MUL_MAT_ID(ffn_moe_down-61): 203 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 459 us MUL_MAT_ID(ffn_moe_down-62): 172 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 401 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 373 us MUL_MAT_ID(ffn_moe_down-64): 613 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 360 us MUL_MAT_ID(ffn_moe_down-65): 211 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 374 us MUL_MAT_ID(ffn_moe_down-66): 415 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 405 us MUL_MAT_ID(ffn_moe_down-67): 407 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 407 us MUL_MAT_ID(ffn_moe_down-68): 190 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 431 us MUL_MAT_ID(ffn_moe_down-69): 286 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 403 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 439 us MUL_MAT_ID(ffn_moe_down-71): 182 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 323 us MUL_MAT_ID(ffn_moe_down-72): 216 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 475 us MUL_MAT_ID(ffn_moe_down-73): 204 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 529 us MUL_MAT_ID(ffn_moe_down-74): 204 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 439 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 518 us MUL_MAT_ID(ffn_moe_down-76): 208 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 377 us MUL_MAT_ID(ffn_moe_down-77): 382 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 524 us MUL_MAT_ID(ffn_moe_down-78): 343 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 367 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 378 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 405 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 381 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 430 us MUL_MAT_ID(ffn_moe_down-83): 552 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 341 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 416 us MUL_MAT_ID(ffn_moe_down-85): 561 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 404 us MUL_MAT_ID(ffn_moe_down-86): 215 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 414 us MUL_MAT_ID(ffn_moe_down-87): 225 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 389 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 391 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 392 us MUL_MAT_ID(ffn_moe_down-90): 238 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 462 us MUL_MAT_ID(ffn_moe_down-91): 215 us ggml_barrier(...): 219 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 471 us MUL_MAT_ID(ffn_moe_down-25): 331 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 391 us MUL_MAT_ID(ffn_moe_down-26): 198 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 501 us MUL_MAT_ID(ffn_moe_down-27): 230 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 407 us MUL_MAT_ID(ffn_moe_down-28): 193 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 516 us MUL_MAT_ID(ffn_moe_down-29): 323 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 745 us MUL_MAT_ID(ffn_moe_down-30): 250 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 395 us MUL_MAT_ID(ffn_moe_down-31): 260 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 390 us MUL_MAT_ID(ffn_moe_down-32): 207 us ggml_barrier(...): 356 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 429 us MUL_MAT_ID(ffn_moe_down-34): 207 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 465 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 419 us MUL_MAT_ID(ffn_moe_down-36): 291 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 509 us MUL_MAT_ID(ffn_moe_down-37): 313 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 466 us MUL_MAT_ID(ffn_moe_down-38): 349 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 513 us MUL_MAT_ID(ffn_moe_down-39): 265 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 471 us MUL_MAT_ID(ffn_moe_down-40): 251 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 436 us MUL_MAT_ID(ffn_moe_down-41): 185 us ggml_barrier(...): 447 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 465 us MUL_MAT_ID(ffn_moe_down-42): 331 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 447 us MUL_MAT_ID(ffn_moe_down-43): 226 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 386 us MUL_MAT_ID(ffn_moe_down-44): 225 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 471 us MUL_MAT_ID(ffn_moe_down-45): 217 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 386 us MUL_MAT_ID(ffn_moe_down-46): 487 us ggml_barrier(...): 373 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 430 us MUL_MAT_ID(ffn_moe_down-47): 246 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 380 us MUL_MAT_ID(ffn_moe_down-48): 303 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 342 us MUL_MAT_ID(ffn_moe_down-49): 170 us ggml_barrier(...): 400 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 419 us MUL_MAT_ID(ffn_moe_down-50): 688 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 415 us MUL_MAT_ID(ffn_moe_down-51): 234 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 445 us MUL_MAT_ID(ffn_moe_down-52): 386 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 418 us MUL_MAT_ID(ffn_moe_down-53): 371 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 423 us MUL_MAT_ID(ffn_moe_down-54): 616 us ggml_barrier(...): 395 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 359 us MUL_MAT_ID(ffn_moe_down-55): 212 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 408 us MUL_MAT_ID(ffn_moe_down-56): 260 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 378 us MUL_MAT_ID(ffn_moe_down-57): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 385 us MUL_MAT_ID(ffn_moe_down-58): 266 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 489 us MUL_MAT_ID(ffn_moe_down-59): 256 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 776 us MUL_MAT_ID(ffn_moe_down-60): 267 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 396 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 400 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 356 us MUL_MAT_ID(ffn_moe_down-63): 214 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 413 us MUL_MAT_ID(ffn_moe_down-64): 209 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 423 us MUL_MAT_ID(ffn_moe_down-65): 200 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 402 us MUL_MAT_ID(ffn_moe_down-66): 222 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 388 us MUL_MAT_ID(ffn_moe_down-67): 244 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 374 us MUL_MAT_ID(ffn_moe_down-68): 182 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 538 us MUL_MAT_ID(ffn_moe_down-69): 266 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 402 us MUL_MAT_ID(ffn_moe_down-70): 186 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 452 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 466 us MUL_MAT_ID(ffn_moe_down-72): 281 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 428 us MUL_MAT_ID(ffn_moe_down-73): 315 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 532 us MUL_MAT_ID(ffn_moe_down-74): 273 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 410 us MUL_MAT_ID(ffn_moe_down-75): 293 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 382 us MUL_MAT_ID(ffn_moe_down-76): 206 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 402 us MUL_MAT_ID(ffn_moe_down-77): 199 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 404 us MUL_MAT_ID(ffn_moe_down-78): 264 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 467 us MUL_MAT_ID(ffn_moe_down-79): 287 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 376 us MUL_MAT_ID(ffn_moe_down-80): 243 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 410 us MUL_MAT_ID(ffn_moe_down-81): 253 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 391 us MUL_MAT_ID(ffn_moe_down-83): 256 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 450 us MUL_MAT_ID(ffn_moe_down-84): 301 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 379 us MUL_MAT_ID(ffn_moe_down-85): 556 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 363 us MUL_MAT_ID(ffn_moe_down-86): 237 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 438 us MUL_MAT_ID(ffn_moe_down-87): 279 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 430 us MUL_MAT_ID(ffn_moe_down-88): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 516 us MUL_MAT_ID(ffn_moe_down-89): 232 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 402 us MUL_MAT_ID(ffn_moe_down-90): 251 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 403 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 42 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 427 us MUL_MAT_ID(ffn_moe_down-25): 281 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 408 us MUL_MAT_ID(ffn_moe_down-26): 254 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 409 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 423 us MUL_MAT_ID(ffn_moe_down-28): 227 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 498 us MUL_MAT_ID(ffn_moe_down-29): 585 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 405 us MUL_MAT_ID(ffn_moe_down-30): 277 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 580 us MUL_MAT_ID(ffn_moe_down-31): 238 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 370 us MUL_MAT_ID(ffn_moe_down-32): 199 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 534 us MUL_MAT_ID(ffn_moe_down-33): 294 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 435 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 409 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 498 us MUL_MAT_ID(ffn_moe_down-36): 202 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 388 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 375 us MUL_MAT_ID(ffn_moe_down-38): 183 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 396 us MUL_MAT_ID(ffn_moe_down-39): 251 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 443 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 465 us MUL_MAT_ID(ffn_moe_down-41): 189 us ggml_barrier(...): 364 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 738 us MUL_MAT_ID(ffn_moe_down-42): 318 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 358 us MUL_MAT_ID(ffn_moe_down-43): 285 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 420 us MUL_MAT_ID(ffn_moe_down-44): 179 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 425 us MUL_MAT_ID(ffn_moe_down-45): 226 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 400 us MUL_MAT_ID(ffn_moe_down-46): 198 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 420 us MUL_MAT_ID(ffn_moe_down-47): 215 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 425 us MUL_MAT_ID(ffn_moe_down-48): 269 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 461 us MUL_MAT_ID(ffn_moe_down-49): 263 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 399 us MUL_MAT_ID(ffn_moe_down-50): 571 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 463 us MUL_MAT_ID(ffn_moe_down-51): 177 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 448 us MUL_MAT_ID(ffn_moe_down-52): 205 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 438 us MUL_MAT_ID(ffn_moe_down-53): 206 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 521 us MUL_MAT_ID(ffn_moe_down-54): 274 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 439 us MUL_MAT_ID(ffn_moe_down-55): 223 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 424 us MUL_MAT_ID(ffn_moe_down-56): 337 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 420 us MUL_MAT_ID(ffn_moe_down-57): 300 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 447 us MUL_MAT_ID(ffn_moe_down-58): 525 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 421 us MUL_MAT_ID(ffn_moe_down-59): 221 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 469 us MUL_MAT_ID(ffn_moe_down-60): 214 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 426 us MUL_MAT_ID(ffn_moe_down-61): 308 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 392 us MUL_MAT_ID(ffn_moe_down-62): 284 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 430 us MUL_MAT_ID(ffn_moe_down-63): 252 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 409 us MUL_MAT_ID(ffn_moe_down-64): 223 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 505 us MUL_MAT_ID(ffn_moe_down-65): 206 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 451 us MUL_MAT_ID(ffn_moe_down-66): 441 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 401 us MUL_MAT_ID(ffn_moe_down-67): 244 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 783 us MUL_MAT_ID(ffn_moe_down-68): 319 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 422 us MUL_MAT_ID(ffn_moe_down-69): 284 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 432 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 412 us MUL_MAT_ID(ffn_moe_down-71): 255 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 500 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 446 us MUL_MAT_ID(ffn_moe_down-73): 284 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 475 us MUL_MAT_ID(ffn_moe_down-74): 277 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 512 us MUL_MAT_ID(ffn_moe_down-75): 266 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 405 us MUL_MAT_ID(ffn_moe_down-76): 219 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 456 us MUL_MAT_ID(ffn_moe_down-77): 546 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 546 us MUL_MAT_ID(ffn_moe_down-78): 229 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 419 us MUL_MAT_ID(ffn_moe_down-79): 183 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 381 us MUL_MAT_ID(ffn_moe_down-80): 234 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 464 us MUL_MAT_ID(ffn_moe_down-81): 200 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 484 us MUL_MAT_ID(ffn_moe_down-82): 259 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 378 us MUL_MAT_ID(ffn_moe_down-83): 214 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 441 us MUL_MAT_ID(ffn_moe_down-84): 290 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 422 us MUL_MAT_ID(ffn_moe_down-85): 225 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 382 us MUL_MAT_ID(ffn_moe_down-86): 221 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 467 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 452 us MUL_MAT_ID(ffn_moe_down-88): 263 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 394 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 395 us MUL_MAT_ID(ffn_moe_down-90): 225 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 424 us MUL_MAT_ID(ffn_moe_down-91): 228 us ggml_barrier(...): 49 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 550 us MUL_MAT_ID(ffn_moe_down-25): 208 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 416 us MUL_MAT_ID(ffn_moe_down-26): 199 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 389 us MUL_MAT_ID(ffn_moe_down-27): 188 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 702 us MUL_MAT_ID(ffn_moe_down-28): 292 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 468 us MUL_MAT_ID(ffn_moe_down-29): 197 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 469 us MUL_MAT_ID(ffn_moe_down-30): 209 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 386 us MUL_MAT_ID(ffn_moe_down-31): 339 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 389 us MUL_MAT_ID(ffn_moe_down-32): 188 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 350 us MUL_MAT_ID(ffn_moe_down-33): 566 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 361 us MUL_MAT_ID(ffn_moe_down-34): 218 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 406 us MUL_MAT_ID(ffn_moe_down-35): 201 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 349 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 378 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 431 us MUL_MAT_ID(ffn_moe_down-38): 320 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 405 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 395 us MUL_MAT_ID(ffn_moe_down-40): 491 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 376 us MUL_MAT_ID(ffn_moe_down-41): 209 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 558 us MUL_MAT_ID(ffn_moe_down-42): 334 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 395 us MUL_MAT_ID(ffn_moe_down-43): 198 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 435 us MUL_MAT_ID(ffn_moe_down-44): 184 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 405 us MUL_MAT_ID(ffn_moe_down-45): 235 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 480 us MUL_MAT_ID(ffn_moe_down-46): 181 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 429 us MUL_MAT_ID(ffn_moe_down-47): 389 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 377 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 345 us MUL_MAT_ID(ffn_moe_down-49): 269 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 506 us MUL_MAT_ID(ffn_moe_down-50): 334 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 503 us MUL_MAT_ID(ffn_moe_down-51): 243 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 510 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 434 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 353 us MUL_MAT_ID(ffn_moe_down-54): 305 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 374 us MUL_MAT_ID(ffn_moe_down-55): 368 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 471 us MUL_MAT_ID(ffn_moe_down-56): 346 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 362 us MUL_MAT_ID(ffn_moe_down-57): 218 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 389 us MUL_MAT_ID(ffn_moe_down-58): 206 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 418 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 442 us MUL_MAT_ID(ffn_moe_down-60): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 410 us MUL_MAT_ID(ffn_moe_down-61): 221 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 364 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 201 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 417 us MUL_MAT_ID(ffn_moe_down-64): 202 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 437 us MUL_MAT_ID(ffn_moe_down-65): 406 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 415 us MUL_MAT_ID(ffn_moe_down-66): 185 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 426 us MUL_MAT_ID(ffn_moe_down-67): 209 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 494 us MUL_MAT_ID(ffn_moe_down-68): 213 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 764 us MUL_MAT_ID(ffn_moe_down-69): 228 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 406 us MUL_MAT_ID(ffn_moe_down-70): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 435 us MUL_MAT_ID(ffn_moe_down-71): 174 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 415 us MUL_MAT_ID(ffn_moe_down-72): 254 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 412 us MUL_MAT_ID(ffn_moe_down-73): 296 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 472 us MUL_MAT_ID(ffn_moe_down-74): 216 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 410 us MUL_MAT_ID(ffn_moe_down-75): 202 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 460 us MUL_MAT_ID(ffn_moe_down-76): 205 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 400 us MUL_MAT_ID(ffn_moe_down-77): 213 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 443 us MUL_MAT_ID(ffn_moe_down-78): 230 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 491 us MUL_MAT_ID(ffn_moe_down-79): 323 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 444 us MUL_MAT_ID(ffn_moe_down-80): 290 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 373 us MUL_MAT_ID(ffn_moe_down-81): 236 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 439 us MUL_MAT_ID(ffn_moe_down-82): 225 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 375 us MUL_MAT_ID(ffn_moe_down-83): 174 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 422 us MUL_MAT_ID(ffn_moe_down-84): 224 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 434 us MUL_MAT_ID(ffn_moe_down-85): 417 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 425 us MUL_MAT_ID(ffn_moe_down-86): 196 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 489 us MUL_MAT_ID(ffn_moe_down-87): 206 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 557 us MUL_MAT_ID(ffn_moe_down-88): 231 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 388 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 471 us MUL_MAT_ID(ffn_moe_down-90): 229 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 369 us MUL_MAT_ID(ffn_moe_down-91): 674 us ggml_barrier(...): 47 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 411 us MUL_MAT_ID(ffn_moe_down-25): 220 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 407 us MUL_MAT_ID(ffn_moe_down-26): 217 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 457 us MUL_MAT_ID(ffn_moe_down-27): 208 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 452 us MUL_MAT_ID(ffn_moe_down-28): 207 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 416 us MUL_MAT_ID(ffn_moe_down-29): 190 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 530 us MUL_MAT_ID(ffn_moe_down-30): 256 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 521 us MUL_MAT_ID(ffn_moe_down-31): 234 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 399 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 506 us MUL_MAT_ID(ffn_moe_down-33): 413 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 405 us MUL_MAT_ID(ffn_moe_down-34): 185 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 521 us MUL_MAT_ID(ffn_moe_down-35): 197 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 465 us MUL_MAT_ID(ffn_moe_down-36): 233 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 382 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 422 us MUL_MAT_ID(ffn_moe_down-38): 285 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 460 us MUL_MAT_ID(ffn_moe_down-39): 288 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 411 us MUL_MAT_ID(ffn_moe_down-40): 222 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 394 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 431 us MUL_MAT_ID(ffn_moe_down-42): 212 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 478 us MUL_MAT_ID(ffn_moe_down-43): 523 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 377 us MUL_MAT_ID(ffn_moe_down-44): 590 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 397 us MUL_MAT_ID(ffn_moe_down-45): 274 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 516 us MUL_MAT_ID(ffn_moe_down-46): 195 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 406 us MUL_MAT_ID(ffn_moe_down-47): 206 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 361 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 408 us MUL_MAT_ID(ffn_moe_down-49): 273 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 568 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 403 us MUL_MAT_ID(ffn_moe_down-51): 262 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 399 us MUL_MAT_ID(ffn_moe_down-52): 353 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 453 us MUL_MAT_ID(ffn_moe_down-53): 216 us ggml_barrier(...): 411 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 391 us MUL_MAT_ID(ffn_moe_down-54): 243 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 411 us MUL_MAT_ID(ffn_moe_down-55): 218 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 404 us MUL_MAT_ID(ffn_moe_down-56): 248 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 462 us MUL_MAT_ID(ffn_moe_down-57): 260 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 401 us MUL_MAT_ID(ffn_moe_down-58): 250 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 423 us MUL_MAT_ID(ffn_moe_down-59): 232 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 429 us MUL_MAT_ID(ffn_moe_down-60): 202 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 422 us MUL_MAT_ID(ffn_moe_down-61): 262 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 372 us MUL_MAT_ID(ffn_moe_down-62): 275 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 506 us MUL_MAT_ID(ffn_moe_down-63): 316 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 400 us MUL_MAT_ID(ffn_moe_down-64): 387 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 458 us MUL_MAT_ID(ffn_moe_down-65): 256 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 425 us MUL_MAT_ID(ffn_moe_down-66): 419 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 482 us MUL_MAT_ID(ffn_moe_down-67): 205 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 371 us MUL_MAT_ID(ffn_moe_down-68): 218 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 452 us MUL_MAT_ID(ffn_moe_down-69): 224 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 368 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 430 us MUL_MAT_ID(ffn_moe_down-71): 301 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 415 us MUL_MAT_ID(ffn_moe_down-72): 303 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 425 us MUL_MAT_ID(ffn_moe_down-73): 270 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 462 us MUL_MAT_ID(ffn_moe_down-74): 214 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 390 us MUL_MAT_ID(ffn_moe_down-75): 189 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 433 us MUL_MAT_ID(ffn_moe_down-76): 570 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 402 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 429 us MUL_MAT_ID(ffn_moe_down-78): 325 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 431 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 414 us MUL_MAT_ID(ffn_moe_down-80): 276 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 403 us MUL_MAT_ID(ffn_moe_down-81): 291 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 423 us MUL_MAT_ID(ffn_moe_down-82): 187 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 426 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 390 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 416 us MUL_MAT_ID(ffn_moe_down-85): 240 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 443 us MUL_MAT_ID(ffn_moe_down-86): 416 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 478 us MUL_MAT_ID(ffn_moe_down-87): 206 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 433 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 403 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 396 us MUL_MAT_ID(ffn_moe_down-90): 487 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 419 us MUL_MAT_ID(ffn_moe_down-91): 252 us ggml_barrier(...): 16 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 409 us MUL_MAT_ID(ffn_moe_down-25): 336 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 390 us MUL_MAT_ID(ffn_moe_down-26): 246 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 427 us MUL_MAT_ID(ffn_moe_down-27): 267 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 416 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 378 us MUL_MAT_ID(ffn_moe_down-29): 228 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 444 us MUL_MAT_ID(ffn_moe_down-30): 261 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 390 us MUL_MAT_ID(ffn_moe_down-31): 180 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 409 us MUL_MAT_ID(ffn_moe_down-32): 275 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 437 us MUL_MAT_ID(ffn_moe_down-33): 260 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 485 us MUL_MAT_ID(ffn_moe_down-34): 229 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 416 us MUL_MAT_ID(ffn_moe_down-35): 237 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 369 us MUL_MAT_ID(ffn_moe_down-36): 311 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 432 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 370 us MUL_MAT_ID(ffn_moe_down-38): 360 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 386 us MUL_MAT_ID(ffn_moe_down-39): 586 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 429 us MUL_MAT_ID(ffn_moe_down-40): 354 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 440 us MUL_MAT_ID(ffn_moe_down-41): 195 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 440 us MUL_MAT_ID(ffn_moe_down-42): 180 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 393 us MUL_MAT_ID(ffn_moe_down-43): 199 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 337 us MUL_MAT_ID(ffn_moe_down-44): 258 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 415 us MUL_MAT_ID(ffn_moe_down-45): 209 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 384 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 380 us MUL_MAT_ID(ffn_moe_down-47): 253 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 452 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 373 us MUL_MAT_ID(ffn_moe_down-49): 222 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 383 us MUL_MAT_ID(ffn_moe_down-50): 210 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 381 us MUL_MAT_ID(ffn_moe_down-51): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 539 us MUL_MAT_ID(ffn_moe_down-52): 447 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 437 us MUL_MAT_ID(ffn_moe_down-53): 246 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 353 us MUL_MAT_ID(ffn_moe_down-54): 187 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 387 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 432 us MUL_MAT_ID(ffn_moe_down-56): 365 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 428 us MUL_MAT_ID(ffn_moe_down-57): 186 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 437 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 396 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 436 us MUL_MAT_ID(ffn_moe_down-60): 210 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 508 us MUL_MAT_ID(ffn_moe_down-61): 278 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 441 us MUL_MAT_ID(ffn_moe_down-62): 203 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 362 us MUL_MAT_ID(ffn_moe_down-63): 473 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 414 us MUL_MAT_ID(ffn_moe_down-64): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 424 us MUL_MAT_ID(ffn_moe_down-65): 258 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 406 us MUL_MAT_ID(ffn_moe_down-66): 225 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 412 us MUL_MAT_ID(ffn_moe_down-67): 184 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 412 us MUL_MAT_ID(ffn_moe_down-68): 190 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 328 us MUL_MAT_ID(ffn_moe_down-69): 204 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 389 us MUL_MAT_ID(ffn_moe_down-70): 266 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 419 us MUL_MAT_ID(ffn_moe_down-71): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 473 us MUL_MAT_ID(ffn_moe_down-72): 290 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 583 us MUL_MAT_ID(ffn_moe_down-73): 327 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 381 us MUL_MAT_ID(ffn_moe_down-74): 215 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 414 us MUL_MAT_ID(ffn_moe_down-75): 543 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 413 us MUL_MAT_ID(ffn_moe_down-76): 310 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 510 us MUL_MAT_ID(ffn_moe_down-77): 179 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 435 us MUL_MAT_ID(ffn_moe_down-78): 680 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 376 us MUL_MAT_ID(ffn_moe_down-79): 295 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 417 us MUL_MAT_ID(ffn_moe_down-80): 376 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 377 us MUL_MAT_ID(ffn_moe_down-81): 208 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 391 us MUL_MAT_ID(ffn_moe_down-82): 367 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 448 us MUL_MAT_ID(ffn_moe_down-83): 416 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 472 us MUL_MAT_ID(ffn_moe_down-84): 270 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 423 us MUL_MAT_ID(ffn_moe_down-85): 243 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 480 us MUL_MAT_ID(ffn_moe_down-86): 221 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 429 us MUL_MAT_ID(ffn_moe_down-87): 271 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 193 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 440 us MUL_MAT_ID(ffn_moe_down-89): 203 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 339 us MUL_MAT_ID(ffn_moe_down-90): 656 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 429 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 18 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 432 us MUL_MAT_ID(ffn_moe_down-25): 310 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 406 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 507 us MUL_MAT_ID(ffn_moe_down-27): 303 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 409 us MUL_MAT_ID(ffn_moe_down-28): 314 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 679 us MUL_MAT_ID(ffn_moe_down-29): 183 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 415 us MUL_MAT_ID(ffn_moe_down-30): 181 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 441 us MUL_MAT_ID(ffn_moe_down-31): 184 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 439 us MUL_MAT_ID(ffn_moe_down-32): 311 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 484 us MUL_MAT_ID(ffn_moe_down-33): 187 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 451 us MUL_MAT_ID(ffn_moe_down-34): 180 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 411 us MUL_MAT_ID(ffn_moe_down-35): 205 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 431 us MUL_MAT_ID(ffn_moe_down-36): 323 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 447 us MUL_MAT_ID(ffn_moe_down-37): 238 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 411 us MUL_MAT_ID(ffn_moe_down-38): 322 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 403 us MUL_MAT_ID(ffn_moe_down-39): 277 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 396 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 543 us MUL_MAT_ID(ffn_moe_down-41): 221 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 398 us MUL_MAT_ID(ffn_moe_down-42): 220 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 399 us MUL_MAT_ID(ffn_moe_down-43): 174 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 389 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 517 us MUL_MAT_ID(ffn_moe_down-45): 278 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 517 us MUL_MAT_ID(ffn_moe_down-46): 271 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 416 us MUL_MAT_ID(ffn_moe_down-47): 230 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 392 us MUL_MAT_ID(ffn_moe_down-48): 198 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 410 us MUL_MAT_ID(ffn_moe_down-49): 204 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 378 us MUL_MAT_ID(ffn_moe_down-50): 388 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 408 us MUL_MAT_ID(ffn_moe_down-51): 553 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 434 us MUL_MAT_ID(ffn_moe_down-52): 214 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 467 us MUL_MAT_ID(ffn_moe_down-53): 297 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 455 us MUL_MAT_ID(ffn_moe_down-54): 315 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 404 us MUL_MAT_ID(ffn_moe_down-55): 270 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 509 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 522 us MUL_MAT_ID(ffn_moe_down-57): 178 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 414 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 435 us MUL_MAT_ID(ffn_moe_down-59): 259 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 356 us MUL_MAT_ID(ffn_moe_down-60): 201 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 415 us MUL_MAT_ID(ffn_moe_down-61): 282 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 447 us MUL_MAT_ID(ffn_moe_down-62): 266 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 390 us MUL_MAT_ID(ffn_moe_down-63): 225 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 376 us MUL_MAT_ID(ffn_moe_down-64): 183 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 451 us MUL_MAT_ID(ffn_moe_down-65): 176 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 440 us MUL_MAT_ID(ffn_moe_down-66): 310 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 478 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 417 us MUL_MAT_ID(ffn_moe_down-68): 339 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 376 us MUL_MAT_ID(ffn_moe_down-69): 203 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 439 us MUL_MAT_ID(ffn_moe_down-70): 230 us ggml_barrier(...): 335 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 422 us MUL_MAT_ID(ffn_moe_down-71): 263 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 414 us MUL_MAT_ID(ffn_moe_down-72): 189 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 403 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 459 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 442 us MUL_MAT_ID(ffn_moe_down-75): 266 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 432 us MUL_MAT_ID(ffn_moe_down-76): 278 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 418 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 351 us MUL_MAT_ID(ffn_moe_down-78): 240 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 440 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 515 us MUL_MAT_ID(ffn_moe_down-80): 251 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 533 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 702 us MUL_MAT_ID(ffn_moe_down-82): 258 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 416 us MUL_MAT_ID(ffn_moe_down-83): 191 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 440 us MUL_MAT_ID(ffn_moe_down-84): 247 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 479 us MUL_MAT_ID(ffn_moe_down-85): 319 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 413 us MUL_MAT_ID(ffn_moe_down-86): 300 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 530 us MUL_MAT_ID(ffn_moe_down-87): 211 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 372 us MUL_MAT_ID(ffn_moe_down-88): 270 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 400 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 388 us MUL_MAT_ID(ffn_moe_down-90): 431 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 419 us MUL_MAT_ID(ffn_moe_down-91): 270 us ggml_barrier(...): 156 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 482 us MUL_MAT_ID(ffn_moe_down-25): 197 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 463 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 384 us MUL_MAT_ID(ffn_moe_down-27): 306 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 376 us MUL_MAT_ID(ffn_moe_down-28): 243 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 583 us MUL_MAT_ID(ffn_moe_down-29): 347 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 522 us MUL_MAT_ID(ffn_moe_down-30): 347 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 455 us MUL_MAT_ID(ffn_moe_down-31): 461 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 651 us MUL_MAT_ID(ffn_moe_down-32): 273 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 395 us MUL_MAT_ID(ffn_moe_down-33): 245 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 385 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 445 us MUL_MAT_ID(ffn_moe_down-35): 218 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 460 us MUL_MAT_ID(ffn_moe_down-36): 210 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 441 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 387 us MUL_MAT_ID(ffn_moe_down-38): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 403 us MUL_MAT_ID(ffn_moe_down-39): 249 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 369 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 400 us MUL_MAT_ID(ffn_moe_down-41): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 416 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 425 us MUL_MAT_ID(ffn_moe_down-43): 207 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 367 us MUL_MAT_ID(ffn_moe_down-44): 325 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 443 us MUL_MAT_ID(ffn_moe_down-45): 289 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 389 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 441 us MUL_MAT_ID(ffn_moe_down-47): 259 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 457 us MUL_MAT_ID(ffn_moe_down-48): 253 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 485 us MUL_MAT_ID(ffn_moe_down-49): 187 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 410 us MUL_MAT_ID(ffn_moe_down-50): 217 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 429 us MUL_MAT_ID(ffn_moe_down-51): 154 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 399 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 400 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 368 us MUL_MAT_ID(ffn_moe_down-54): 245 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 349 us MUL_MAT_ID(ffn_moe_down-55): 267 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 528 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 466 us MUL_MAT_ID(ffn_moe_down-57): 224 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 396 us MUL_MAT_ID(ffn_moe_down-58): 217 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 501 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 534 us MUL_MAT_ID(ffn_moe_down-60): 191 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 405 us MUL_MAT_ID(ffn_moe_down-61): 194 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 517 us MUL_MAT_ID(ffn_moe_down-62): 301 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 464 us MUL_MAT_ID(ffn_moe_down-63): 281 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 454 us MUL_MAT_ID(ffn_moe_down-64): 309 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 470 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 433 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 462 us MUL_MAT_ID(ffn_moe_down-67): 443 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 444 us MUL_MAT_ID(ffn_moe_down-68): 294 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 425 us MUL_MAT_ID(ffn_moe_down-69): 299 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 431 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 416 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 445 us MUL_MAT_ID(ffn_moe_down-72): 216 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 405 us MUL_MAT_ID(ffn_moe_down-73): 220 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 400 us MUL_MAT_ID(ffn_moe_down-74): 254 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 406 us MUL_MAT_ID(ffn_moe_down-75): 176 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 393 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 403 us MUL_MAT_ID(ffn_moe_down-77): 235 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 475 us MUL_MAT_ID(ffn_moe_down-78): 239 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 607 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 402 us MUL_MAT_ID(ffn_moe_down-80): 219 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 472 us MUL_MAT_ID(ffn_moe_down-81): 199 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 500 us MUL_MAT_ID(ffn_moe_down-82): 202 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 669 us MUL_MAT_ID(ffn_moe_down-83): 262 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 417 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 513 us MUL_MAT_ID(ffn_moe_down-85): 285 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 407 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 424 us MUL_MAT_ID(ffn_moe_down-88): 267 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 398 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 392 us MUL_MAT_ID(ffn_moe_down-90): 524 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 513 us MUL_MAT_ID(ffn_moe_down-91): 350 us ggml_barrier(...): 126 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 516 us MUL_MAT_ID(ffn_moe_down-25): 240 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 450 us MUL_MAT_ID(ffn_moe_down-26): 302 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 428 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 358 us MUL_MAT_ID(ffn_moe_down-28): 187 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 369 us MUL_MAT_ID(ffn_moe_down-29): 190 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 408 us MUL_MAT_ID(ffn_moe_down-30): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 457 us MUL_MAT_ID(ffn_moe_down-31): 214 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 435 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 448 us MUL_MAT_ID(ffn_moe_down-33): 202 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 485 us MUL_MAT_ID(ffn_moe_down-34): 247 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 384 us MUL_MAT_ID(ffn_moe_down-35): 183 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 365 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 407 us MUL_MAT_ID(ffn_moe_down-37): 294 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 433 us MUL_MAT_ID(ffn_moe_down-38): 235 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 409 us MUL_MAT_ID(ffn_moe_down-39): 356 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 660 us MUL_MAT_ID(ffn_moe_down-40): 350 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 353 us MUL_MAT_ID(ffn_moe_down-41): 573 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 372 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 403 us MUL_MAT_ID(ffn_moe_down-43): 209 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 430 us MUL_MAT_ID(ffn_moe_down-44): 454 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 382 us MUL_MAT_ID(ffn_moe_down-45): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 381 us MUL_MAT_ID(ffn_moe_down-46): 215 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 423 us MUL_MAT_ID(ffn_moe_down-47): 286 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 432 us MUL_MAT_ID(ffn_moe_down-48): 247 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 271 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 456 us MUL_MAT_ID(ffn_moe_down-50): 193 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 369 us MUL_MAT_ID(ffn_moe_down-51): 189 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 406 us MUL_MAT_ID(ffn_moe_down-52): 420 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 410 us MUL_MAT_ID(ffn_moe_down-53): 551 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 422 us MUL_MAT_ID(ffn_moe_down-54): 215 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 421 us MUL_MAT_ID(ffn_moe_down-55): 193 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 393 us MUL_MAT_ID(ffn_moe_down-56): 219 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 479 us MUL_MAT_ID(ffn_moe_down-57): 179 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 385 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 425 us MUL_MAT_ID(ffn_moe_down-59): 208 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 402 us MUL_MAT_ID(ffn_moe_down-60): 207 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 517 us MUL_MAT_ID(ffn_moe_down-61): 180 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 459 us MUL_MAT_ID(ffn_moe_down-62): 346 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 425 us MUL_MAT_ID(ffn_moe_down-63): 243 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 406 us MUL_MAT_ID(ffn_moe_down-64): 177 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 438 us MUL_MAT_ID(ffn_moe_down-65): 294 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 524 us MUL_MAT_ID(ffn_moe_down-66): 274 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 388 us MUL_MAT_ID(ffn_moe_down-67): 312 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 424 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 447 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 379 us MUL_MAT_ID(ffn_moe_down-70): 238 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 418 us MUL_MAT_ID(ffn_moe_down-71): 239 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 405 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 477 us MUL_MAT_ID(ffn_moe_down-73): 599 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 522 us MUL_MAT_ID(ffn_moe_down-74): 206 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 405 us MUL_MAT_ID(ffn_moe_down-75): 409 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 435 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 478 us MUL_MAT_ID(ffn_moe_down-77): 246 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 483 us MUL_MAT_ID(ffn_moe_down-78): 268 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 499 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 410 us MUL_MAT_ID(ffn_moe_down-80): 381 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 455 us MUL_MAT_ID(ffn_moe_down-81): 485 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 404 us MUL_MAT_ID(ffn_moe_down-82): 224 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 403 us MUL_MAT_ID(ffn_moe_down-83): 218 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 375 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 399 us MUL_MAT_ID(ffn_moe_down-85): 254 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 425 us MUL_MAT_ID(ffn_moe_down-86): 414 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 417 us MUL_MAT_ID(ffn_moe_down-87): 212 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 502 us MUL_MAT_ID(ffn_moe_down-88): 227 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 376 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 421 us MUL_MAT_ID(ffn_moe_down-90): 281 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 386 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 41 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 367 us MUL_MAT_ID(ffn_moe_down-25): 228 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 617 us MUL_MAT_ID(ffn_moe_down-26): 328 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 433 us MUL_MAT_ID(ffn_moe_down-27): 288 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 425 us MUL_MAT_ID(ffn_moe_down-28): 167 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 394 us MUL_MAT_ID(ffn_moe_down-29): 204 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 384 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 555 us MUL_MAT_ID(ffn_moe_down-31): 408 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 383 us MUL_MAT_ID(ffn_moe_down-32): 242 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 434 us MUL_MAT_ID(ffn_moe_down-33): 454 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 397 us MUL_MAT_ID(ffn_moe_down-34): 190 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 435 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 447 us MUL_MAT_ID(ffn_moe_down-36): 263 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 455 us MUL_MAT_ID(ffn_moe_down-37): 212 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 424 us MUL_MAT_ID(ffn_moe_down-38): 199 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 428 us MUL_MAT_ID(ffn_moe_down-39): 393 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 385 us MUL_MAT_ID(ffn_moe_down-40): 225 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 585 us MUL_MAT_ID(ffn_moe_down-41): 258 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 412 us MUL_MAT_ID(ffn_moe_down-42): 393 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 555 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 440 us MUL_MAT_ID(ffn_moe_down-44): 220 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 392 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 394 us MUL_MAT_ID(ffn_moe_down-46): 257 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 416 us MUL_MAT_ID(ffn_moe_down-47): 390 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 340 us MUL_MAT_ID(ffn_moe_down-48): 172 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 410 us MUL_MAT_ID(ffn_moe_down-49): 185 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 377 us MUL_MAT_ID(ffn_moe_down-50): 205 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 409 us MUL_MAT_ID(ffn_moe_down-51): 213 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 389 us MUL_MAT_ID(ffn_moe_down-52): 218 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 395 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 379 us MUL_MAT_ID(ffn_moe_down-54): 219 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 395 us MUL_MAT_ID(ffn_moe_down-55): 194 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 358 us MUL_MAT_ID(ffn_moe_down-56): 244 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 389 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 467 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 540 us MUL_MAT_ID(ffn_moe_down-59): 271 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 669 us MUL_MAT_ID(ffn_moe_down-60): 313 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 370 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 397 us MUL_MAT_ID(ffn_moe_down-62): 192 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 425 us MUL_MAT_ID(ffn_moe_down-63): 256 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 404 us MUL_MAT_ID(ffn_moe_down-64): 208 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 404 us MUL_MAT_ID(ffn_moe_down-65): 226 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 393 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 418 us MUL_MAT_ID(ffn_moe_down-67): 522 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 396 us MUL_MAT_ID(ffn_moe_down-68): 300 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 445 us MUL_MAT_ID(ffn_moe_down-69): 291 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 422 us MUL_MAT_ID(ffn_moe_down-70): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 408 us MUL_MAT_ID(ffn_moe_down-71): 270 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 409 us MUL_MAT_ID(ffn_moe_down-72): 172 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 433 us MUL_MAT_ID(ffn_moe_down-73): 298 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 380 us MUL_MAT_ID(ffn_moe_down-74): 204 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 369 us MUL_MAT_ID(ffn_moe_down-75): 247 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 391 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 395 us MUL_MAT_ID(ffn_moe_down-77): 582 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 414 us MUL_MAT_ID(ffn_moe_down-78): 311 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 434 us MUL_MAT_ID(ffn_moe_down-79): 227 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 430 us MUL_MAT_ID(ffn_moe_down-80): 518 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 408 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 470 us MUL_MAT_ID(ffn_moe_down-82): 297 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 457 us MUL_MAT_ID(ffn_moe_down-83): 211 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 455 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 399 us MUL_MAT_ID(ffn_moe_down-85): 228 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 418 us MUL_MAT_ID(ffn_moe_down-86): 231 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 368 us MUL_MAT_ID(ffn_moe_down-87): 198 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 420 us MUL_MAT_ID(ffn_moe_down-88): 206 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 396 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 430 us MUL_MAT_ID(ffn_moe_down-90): 653 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 402 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 21 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 493 us MUL_MAT_ID(ffn_moe_down-25): 299 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 493 us MUL_MAT_ID(ffn_moe_down-26): 191 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 366 us MUL_MAT_ID(ffn_moe_down-27): 405 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 482 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 447 us MUL_MAT_ID(ffn_moe_down-29): 234 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 391 us MUL_MAT_ID(ffn_moe_down-30): 329 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 433 us MUL_MAT_ID(ffn_moe_down-31): 248 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 430 us MUL_MAT_ID(ffn_moe_down-32): 305 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 425 us MUL_MAT_ID(ffn_moe_down-33): 634 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 395 us MUL_MAT_ID(ffn_moe_down-34): 269 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 483 us MUL_MAT_ID(ffn_moe_down-35): 200 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 370 us MUL_MAT_ID(ffn_moe_down-36): 229 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 500 us MUL_MAT_ID(ffn_moe_down-37): 257 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 529 us MUL_MAT_ID(ffn_moe_down-38): 229 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 438 us MUL_MAT_ID(ffn_moe_down-39): 299 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 489 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 408 us MUL_MAT_ID(ffn_moe_down-41): 303 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 492 us MUL_MAT_ID(ffn_moe_down-42): 213 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 427 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 393 us MUL_MAT_ID(ffn_moe_down-44): 185 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 382 us MUL_MAT_ID(ffn_moe_down-45): 227 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 541 us MUL_MAT_ID(ffn_moe_down-46): 380 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 585 us MUL_MAT_ID(ffn_moe_down-47): 227 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 542 us MUL_MAT_ID(ffn_moe_down-48): 290 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 478 us MUL_MAT_ID(ffn_moe_down-49): 249 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 441 us MUL_MAT_ID(ffn_moe_down-50): 498 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 378 us MUL_MAT_ID(ffn_moe_down-51): 215 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 470 us MUL_MAT_ID(ffn_moe_down-52): 214 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 402 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 432 us MUL_MAT_ID(ffn_moe_down-54): 306 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 411 us MUL_MAT_ID(ffn_moe_down-55): 229 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 493 us MUL_MAT_ID(ffn_moe_down-56): 221 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 401 us MUL_MAT_ID(ffn_moe_down-57): 262 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 416 us MUL_MAT_ID(ffn_moe_down-58): 206 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 446 us MUL_MAT_ID(ffn_moe_down-59): 252 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 515 us MUL_MAT_ID(ffn_moe_down-60): 180 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 397 us MUL_MAT_ID(ffn_moe_down-61): 199 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 499 us MUL_MAT_ID(ffn_moe_down-62): 206 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 453 us MUL_MAT_ID(ffn_moe_down-63): 388 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 402 us MUL_MAT_ID(ffn_moe_down-64): 262 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 541 us MUL_MAT_ID(ffn_moe_down-65): 211 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 433 us MUL_MAT_ID(ffn_moe_down-66): 198 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 437 us MUL_MAT_ID(ffn_moe_down-67): 186 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 381 us MUL_MAT_ID(ffn_moe_down-68): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 387 us MUL_MAT_ID(ffn_moe_down-69): 228 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 463 us MUL_MAT_ID(ffn_moe_down-70): 277 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 402 us MUL_MAT_ID(ffn_moe_down-71): 242 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 326 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 476 us MUL_MAT_ID(ffn_moe_down-73): 241 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 399 us MUL_MAT_ID(ffn_moe_down-74): 259 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 425 us MUL_MAT_ID(ffn_moe_down-75): 262 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 469 us MUL_MAT_ID(ffn_moe_down-76): 272 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 560 us MUL_MAT_ID(ffn_moe_down-77): 214 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 394 us MUL_MAT_ID(ffn_moe_down-78): 205 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 463 us MUL_MAT_ID(ffn_moe_down-79): 217 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 399 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 466 us MUL_MAT_ID(ffn_moe_down-81): 280 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 408 us MUL_MAT_ID(ffn_moe_down-82): 247 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 395 us MUL_MAT_ID(ffn_moe_down-83): 268 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 451 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 406 us MUL_MAT_ID(ffn_moe_down-85): 748 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 392 us MUL_MAT_ID(ffn_moe_down-86): 217 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 412 us MUL_MAT_ID(ffn_moe_down-87): 494 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 383 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 377 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 470 us MUL_MAT_ID(ffn_moe_down-90): 240 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 399 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 39 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 384 us MUL_MAT_ID(ffn_moe_down-25): 427 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 399 us MUL_MAT_ID(ffn_moe_down-26): 202 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 376 us MUL_MAT_ID(ffn_moe_down-27): 228 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 456 us MUL_MAT_ID(ffn_moe_down-28): 228 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 408 us MUL_MAT_ID(ffn_moe_down-29): 194 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 344 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 386 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 390 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 396 us MUL_MAT_ID(ffn_moe_down-33): 188 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 452 us MUL_MAT_ID(ffn_moe_down-34): 249 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 397 us MUL_MAT_ID(ffn_moe_down-35): 300 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 412 us MUL_MAT_ID(ffn_moe_down-36): 489 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 378 us MUL_MAT_ID(ffn_moe_down-37): 242 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 384 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 407 us MUL_MAT_ID(ffn_moe_down-39): 253 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 494 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 364 us MUL_MAT_ID(ffn_moe_down-41): 435 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 388 us MUL_MAT_ID(ffn_moe_down-42): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 374 us MUL_MAT_ID(ffn_moe_down-43): 540 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 427 us MUL_MAT_ID(ffn_moe_down-44): 207 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 447 us MUL_MAT_ID(ffn_moe_down-45): 203 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 414 us MUL_MAT_ID(ffn_moe_down-46): 235 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 454 us MUL_MAT_ID(ffn_moe_down-47): 277 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 387 us MUL_MAT_ID(ffn_moe_down-48): 265 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 420 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 189 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 460 us MUL_MAT_ID(ffn_moe_down-51): 217 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 676 us MUL_MAT_ID(ffn_moe_down-52): 328 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 646 us MUL_MAT_ID(ffn_moe_down-53): 270 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 367 us MUL_MAT_ID(ffn_moe_down-54): 311 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 395 us MUL_MAT_ID(ffn_moe_down-55): 215 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 502 us MUL_MAT_ID(ffn_moe_down-56): 179 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 393 us MUL_MAT_ID(ffn_moe_down-57): 226 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 476 us MUL_MAT_ID(ffn_moe_down-58): 210 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 434 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 443 us MUL_MAT_ID(ffn_moe_down-60): 250 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 387 us MUL_MAT_ID(ffn_moe_down-61): 201 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 656 us MUL_MAT_ID(ffn_moe_down-62): 388 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 360 us MUL_MAT_ID(ffn_moe_down-63): 198 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 383 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 397 us MUL_MAT_ID(ffn_moe_down-65): 198 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 446 us MUL_MAT_ID(ffn_moe_down-66): 209 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 331 us MUL_MAT_ID(ffn_moe_down-67): 269 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 471 us MUL_MAT_ID(ffn_moe_down-68): 205 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 382 us MUL_MAT_ID(ffn_moe_down-69): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 366 us MUL_MAT_ID(ffn_moe_down-70): 186 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 399 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 535 us MUL_MAT_ID(ffn_moe_down-72): 204 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 430 us MUL_MAT_ID(ffn_moe_down-73): 197 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 438 us MUL_MAT_ID(ffn_moe_down-74): 283 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 373 us MUL_MAT_ID(ffn_moe_down-75): 349 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 421 us MUL_MAT_ID(ffn_moe_down-76): 213 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 395 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 458 us MUL_MAT_ID(ffn_moe_down-78): 215 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 394 us MUL_MAT_ID(ffn_moe_down-79): 250 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 394 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 370 us MUL_MAT_ID(ffn_moe_down-81): 246 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 420 us MUL_MAT_ID(ffn_moe_down-82): 209 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 401 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 419 us MUL_MAT_ID(ffn_moe_down-84): 245 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 400 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 413 us MUL_MAT_ID(ffn_moe_down-86): 244 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 389 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 362 us MUL_MAT_ID(ffn_moe_down-88): 197 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 390 us MUL_MAT_ID(ffn_moe_down-89): 363 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 387 us MUL_MAT_ID(ffn_moe_down-90): 252 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 418 us MUL_MAT_ID(ffn_moe_down-91): 424 us ggml_barrier(...): 254 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 370 us MUL_MAT_ID(ffn_moe_down-25): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 393 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 509 us MUL_MAT_ID(ffn_moe_down-27): 172 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 478 us MUL_MAT_ID(ffn_moe_down-28): 296 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 411 us MUL_MAT_ID(ffn_moe_down-29): 201 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 524 us MUL_MAT_ID(ffn_moe_down-30): 201 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 401 us MUL_MAT_ID(ffn_moe_down-31): 267 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 434 us MUL_MAT_ID(ffn_moe_down-32): 249 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 411 us MUL_MAT_ID(ffn_moe_down-33): 187 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 389 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 381 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 348 us MUL_MAT_ID(ffn_moe_down-36): 214 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 398 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 375 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 384 us MUL_MAT_ID(ffn_moe_down-39): 244 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 378 us MUL_MAT_ID(ffn_moe_down-40): 237 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 424 us MUL_MAT_ID(ffn_moe_down-41): 189 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 422 us MUL_MAT_ID(ffn_moe_down-42): 272 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 404 us MUL_MAT_ID(ffn_moe_down-43): 202 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 378 us MUL_MAT_ID(ffn_moe_down-44): 200 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 409 us MUL_MAT_ID(ffn_moe_down-45): 209 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 421 us MUL_MAT_ID(ffn_moe_down-46): 206 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 377 us MUL_MAT_ID(ffn_moe_down-47): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 508 us MUL_MAT_ID(ffn_moe_down-48): 240 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 387 us MUL_MAT_ID(ffn_moe_down-49): 204 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 370 us MUL_MAT_ID(ffn_moe_down-50): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 376 us MUL_MAT_ID(ffn_moe_down-51): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 405 us MUL_MAT_ID(ffn_moe_down-52): 222 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 397 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 362 us MUL_MAT_ID(ffn_moe_down-54): 209 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 400 us MUL_MAT_ID(ffn_moe_down-55): 211 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 419 us MUL_MAT_ID(ffn_moe_down-56): 216 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 490 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 427 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 367 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 364 us MUL_MAT_ID(ffn_moe_down-60): 258 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 398 us MUL_MAT_ID(ffn_moe_down-61): 286 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 646 us MUL_MAT_ID(ffn_moe_down-62): 254 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 394 us MUL_MAT_ID(ffn_moe_down-63): 222 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 451 us MUL_MAT_ID(ffn_moe_down-64): 186 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 354 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 386 us MUL_MAT_ID(ffn_moe_down-66): 211 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 404 us MUL_MAT_ID(ffn_moe_down-67): 266 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 342 us MUL_MAT_ID(ffn_moe_down-68): 324 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 369 us MUL_MAT_ID(ffn_moe_down-69): 306 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 423 us MUL_MAT_ID(ffn_moe_down-70): 189 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 390 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 415 us MUL_MAT_ID(ffn_moe_down-72): 204 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 398 us MUL_MAT_ID(ffn_moe_down-73): 201 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 389 us MUL_MAT_ID(ffn_moe_down-74): 189 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 451 us MUL_MAT_ID(ffn_moe_down-75): 233 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 381 us MUL_MAT_ID(ffn_moe_down-76): 271 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 407 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 451 us MUL_MAT_ID(ffn_moe_down-78): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 454 us MUL_MAT_ID(ffn_moe_down-79): 206 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 365 us MUL_MAT_ID(ffn_moe_down-80): 219 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 421 us MUL_MAT_ID(ffn_moe_down-81): 182 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 376 us MUL_MAT_ID(ffn_moe_down-82): 259 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 422 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 349 us MUL_MAT_ID(ffn_moe_down-84): 249 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 414 us MUL_MAT_ID(ffn_moe_down-85): 286 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 464 us MUL_MAT_ID(ffn_moe_down-86): 218 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 459 us MUL_MAT_ID(ffn_moe_down-87): 178 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 451 us MUL_MAT_ID(ffn_moe_down-88): 422 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 363 us MUL_MAT_ID(ffn_moe_down-89): 403 us ggml_barrier(...): 342 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 645 us MUL_MAT_ID(ffn_moe_down-90): 263 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 476 us MUL_MAT_ID(ffn_moe_down-91): 263 us ggml_barrier(...): 158 us GET_ROWS(inp_embd): 11 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 423 us MUL_MAT_ID(ffn_moe_down-25): 327 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 478 us MUL_MAT_ID(ffn_moe_down-26): 310 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 394 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 393 us MUL_MAT_ID(ffn_moe_down-28): 247 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 437 us MUL_MAT_ID(ffn_moe_down-29): 426 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 357 us MUL_MAT_ID(ffn_moe_down-30): 202 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 435 us MUL_MAT_ID(ffn_moe_down-31): 188 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 412 us MUL_MAT_ID(ffn_moe_down-32): 210 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 423 us MUL_MAT_ID(ffn_moe_down-33): 218 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 457 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 452 us MUL_MAT_ID(ffn_moe_down-35): 306 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 413 us MUL_MAT_ID(ffn_moe_down-36): 245 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 458 us MUL_MAT_ID(ffn_moe_down-37): 347 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 486 us MUL_MAT_ID(ffn_moe_down-38): 569 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 510 us MUL_MAT_ID(ffn_moe_down-39): 274 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 404 us MUL_MAT_ID(ffn_moe_down-40): 252 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 497 us MUL_MAT_ID(ffn_moe_down-41): 181 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 426 us MUL_MAT_ID(ffn_moe_down-42): 218 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 388 us MUL_MAT_ID(ffn_moe_down-43): 211 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 461 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 479 us MUL_MAT_ID(ffn_moe_down-45): 192 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 505 us MUL_MAT_ID(ffn_moe_down-46): 257 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 440 us MUL_MAT_ID(ffn_moe_down-47): 212 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 453 us MUL_MAT_ID(ffn_moe_down-48): 243 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 411 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 378 us MUL_MAT_ID(ffn_moe_down-50): 291 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 417 us MUL_MAT_ID(ffn_moe_down-51): 179 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 440 us MUL_MAT_ID(ffn_moe_down-52): 280 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 406 us MUL_MAT_ID(ffn_moe_down-53): 206 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 407 us MUL_MAT_ID(ffn_moe_down-54): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 406 us MUL_MAT_ID(ffn_moe_down-56): 186 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 374 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 422 us MUL_MAT_ID(ffn_moe_down-58): 214 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 395 us MUL_MAT_ID(ffn_moe_down-59): 271 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 409 us MUL_MAT_ID(ffn_moe_down-60): 469 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 391 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 343 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 420 us MUL_MAT_ID(ffn_moe_down-62): 332 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 463 us MUL_MAT_ID(ffn_moe_down-63): 342 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 421 us MUL_MAT_ID(ffn_moe_down-64): 242 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 371 us MUL_MAT_ID(ffn_moe_down-65): 186 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 523 us MUL_MAT_ID(ffn_moe_down-66): 398 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 636 us MUL_MAT_ID(ffn_moe_down-67): 418 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 547 us MUL_MAT_ID(ffn_moe_down-68): 387 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 368 us MUL_MAT_ID(ffn_moe_down-69): 213 us ggml_barrier(...): 341 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 376 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 370 us MUL_MAT_ID(ffn_moe_down-71): 232 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 266 us ggml_barrier(...): 353 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 380 us MUL_MAT_ID(ffn_moe_down-73): 223 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 500 us MUL_MAT_ID(ffn_moe_down-74): 227 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 472 us MUL_MAT_ID(ffn_moe_down-75): 210 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 400 us MUL_MAT_ID(ffn_moe_down-76): 298 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 398 us MUL_MAT_ID(ffn_moe_down-77): 295 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 441 us MUL_MAT_ID(ffn_moe_down-78): 283 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 393 us MUL_MAT_ID(ffn_moe_down-79): 205 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 440 us MUL_MAT_ID(ffn_moe_down-80): 204 us ggml_barrier(...): 338 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 557 us MUL_MAT_ID(ffn_moe_down-81): 223 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 493 us MUL_MAT_ID(ffn_moe_down-82): 327 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 460 us MUL_MAT_ID(ffn_moe_down-83): 216 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 426 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 429 us MUL_MAT_ID(ffn_moe_down-85): 719 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 568 us MUL_MAT_ID(ffn_moe_down-86): 206 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 402 us MUL_MAT_ID(ffn_moe_down-87): 316 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 497 us MUL_MAT_ID(ffn_moe_down-88): 214 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 473 us MUL_MAT_ID(ffn_moe_down-89): 204 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 482 us MUL_MAT_ID(ffn_moe_down-90): 443 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 473 us MUL_MAT_ID(ffn_moe_down-91): 325 us ggml_barrier(...): 113 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 453 us MUL_MAT_ID(ffn_moe_down-25): 211 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 420 us MUL_MAT_ID(ffn_moe_down-26): 183 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 462 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 405 us MUL_MAT_ID(ffn_moe_down-28): 605 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 564 us MUL_MAT_ID(ffn_moe_down-29): 444 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 338 us MUL_MAT_ID(ffn_moe_down-30): 202 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 452 us MUL_MAT_ID(ffn_moe_down-31): 330 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 399 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 363 us MUL_MAT_ID(ffn_moe_down-33): 181 us ggml_barrier(...): 550 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 407 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 397 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 551 us MUL_MAT_ID(ffn_moe_down-36): 334 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 375 us MUL_MAT_ID(ffn_moe_down-37): 241 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 405 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 401 us MUL_MAT_ID(ffn_moe_down-39): 252 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 581 us MUL_MAT_ID(ffn_moe_down-40): 285 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 423 us MUL_MAT_ID(ffn_moe_down-41): 191 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 487 us MUL_MAT_ID(ffn_moe_down-42): 257 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 568 us MUL_MAT_ID(ffn_moe_down-43): 192 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 412 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 439 us MUL_MAT_ID(ffn_moe_down-45): 307 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 423 us MUL_MAT_ID(ffn_moe_down-46): 217 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 420 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 413 us MUL_MAT_ID(ffn_moe_down-48): 213 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 396 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 386 us MUL_MAT_ID(ffn_moe_down-50): 193 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 408 us MUL_MAT_ID(ffn_moe_down-51): 225 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 413 us MUL_MAT_ID(ffn_moe_down-52): 298 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 402 us MUL_MAT_ID(ffn_moe_down-53): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 394 us MUL_MAT_ID(ffn_moe_down-54): 250 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 379 us MUL_MAT_ID(ffn_moe_down-55): 284 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 438 us MUL_MAT_ID(ffn_moe_down-56): 300 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 472 us MUL_MAT_ID(ffn_moe_down-57): 179 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 359 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 408 us MUL_MAT_ID(ffn_moe_down-59): 189 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 459 us MUL_MAT_ID(ffn_moe_down-60): 162 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 646 us MUL_MAT_ID(ffn_moe_down-61): 266 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 357 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 378 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 409 us MUL_MAT_ID(ffn_moe_down-64): 264 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 543 us MUL_MAT_ID(ffn_moe_down-65): 259 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 429 us MUL_MAT_ID(ffn_moe_down-66): 390 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 410 us MUL_MAT_ID(ffn_moe_down-67): 544 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 378 us MUL_MAT_ID(ffn_moe_down-68): 257 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 531 us MUL_MAT_ID(ffn_moe_down-69): 244 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 579 us MUL_MAT_ID(ffn_moe_down-70): 335 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 489 us MUL_MAT_ID(ffn_moe_down-71): 187 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 364 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 421 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 446 us MUL_MAT_ID(ffn_moe_down-74): 181 us ggml_barrier(...): 352 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 456 us MUL_MAT_ID(ffn_moe_down-75): 179 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 419 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 388 us MUL_MAT_ID(ffn_moe_down-77): 224 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 473 us MUL_MAT_ID(ffn_moe_down-78): 198 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 403 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 444 us MUL_MAT_ID(ffn_moe_down-80): 304 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 368 us MUL_MAT_ID(ffn_moe_down-81): 242 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 456 us MUL_MAT_ID(ffn_moe_down-82): 274 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 426 us MUL_MAT_ID(ffn_moe_down-83): 292 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 477 us MUL_MAT_ID(ffn_moe_down-84): 262 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 390 us MUL_MAT_ID(ffn_moe_down-85): 423 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 414 us MUL_MAT_ID(ffn_moe_down-86): 342 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 431 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 412 us MUL_MAT_ID(ffn_moe_down-88): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 418 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 441 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 433 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 13 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 537 us MUL_MAT_ID(ffn_moe_down-25): 230 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 493 us MUL_MAT_ID(ffn_moe_down-26): 281 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 423 us MUL_MAT_ID(ffn_moe_down-27): 223 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 498 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 446 us MUL_MAT_ID(ffn_moe_down-29): 230 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 376 us MUL_MAT_ID(ffn_moe_down-30): 266 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 372 us MUL_MAT_ID(ffn_moe_down-31): 186 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 409 us MUL_MAT_ID(ffn_moe_down-32): 213 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 445 us MUL_MAT_ID(ffn_moe_down-33): 269 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 452 us MUL_MAT_ID(ffn_moe_down-34): 186 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 419 us MUL_MAT_ID(ffn_moe_down-35): 248 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 411 us MUL_MAT_ID(ffn_moe_down-36): 222 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 410 us MUL_MAT_ID(ffn_moe_down-37): 275 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 510 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 391 us MUL_MAT_ID(ffn_moe_down-39): 246 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 406 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 555 us MUL_MAT_ID(ffn_moe_down-41): 362 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 461 us MUL_MAT_ID(ffn_moe_down-42): 167 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 422 us MUL_MAT_ID(ffn_moe_down-43): 262 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 563 us MUL_MAT_ID(ffn_moe_down-44): 210 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 389 us MUL_MAT_ID(ffn_moe_down-45): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 440 us MUL_MAT_ID(ffn_moe_down-46): 205 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 382 us MUL_MAT_ID(ffn_moe_down-47): 304 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 430 us MUL_MAT_ID(ffn_moe_down-48): 219 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 365 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 421 us MUL_MAT_ID(ffn_moe_down-50): 744 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 442 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 408 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 349 us MUL_MAT_ID(ffn_moe_down-53): 285 us ggml_barrier(...): 402 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 425 us MUL_MAT_ID(ffn_moe_down-54): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 419 us MUL_MAT_ID(ffn_moe_down-55): 255 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 458 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 394 us MUL_MAT_ID(ffn_moe_down-57): 193 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 416 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 480 us MUL_MAT_ID(ffn_moe_down-59): 213 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 511 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 490 us MUL_MAT_ID(ffn_moe_down-61): 310 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 367 us MUL_MAT_ID(ffn_moe_down-62): 218 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 373 us MUL_MAT_ID(ffn_moe_down-63): 525 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 396 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 344 us MUL_MAT_ID(ffn_moe_down-65): 196 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 430 us MUL_MAT_ID(ffn_moe_down-66): 239 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 376 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 404 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 357 us MUL_MAT_ID(ffn_moe_down-69): 198 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 412 us MUL_MAT_ID(ffn_moe_down-70): 205 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 379 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 402 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 210 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 418 us MUL_MAT_ID(ffn_moe_down-74): 587 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 479 us MUL_MAT_ID(ffn_moe_down-75): 244 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 502 us MUL_MAT_ID(ffn_moe_down-76): 462 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 541 us MUL_MAT_ID(ffn_moe_down-77): 211 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 363 us MUL_MAT_ID(ffn_moe_down-78): 583 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 380 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 433 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 358 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 379 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 403 us MUL_MAT_ID(ffn_moe_down-83): 189 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 410 us MUL_MAT_ID(ffn_moe_down-84): 226 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 620 us MUL_MAT_ID(ffn_moe_down-85): 251 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 429 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 565 us MUL_MAT_ID(ffn_moe_down-87): 383 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 340 us MUL_MAT_ID(ffn_moe_down-88): 328 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 442 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 398 us MUL_MAT_ID(ffn_moe_down-90): 225 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 418 us MUL_MAT_ID(ffn_moe_down-91): 385 us ggml_barrier(...): 119 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 384 us MUL_MAT_ID(ffn_moe_down-25): 214 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 442 us MUL_MAT_ID(ffn_moe_down-26): 259 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 395 us MUL_MAT_ID(ffn_moe_down-27): 232 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 451 us MUL_MAT_ID(ffn_moe_down-28): 162 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 456 us MUL_MAT_ID(ffn_moe_down-29): 218 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 475 us MUL_MAT_ID(ffn_moe_down-30): 167 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 578 us MUL_MAT_ID(ffn_moe_down-31): 229 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 378 us MUL_MAT_ID(ffn_moe_down-32): 666 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 434 us MUL_MAT_ID(ffn_moe_down-33): 261 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 382 us MUL_MAT_ID(ffn_moe_down-34): 273 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 369 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 461 us MUL_MAT_ID(ffn_moe_down-36): 265 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 650 us MUL_MAT_ID(ffn_moe_down-37): 275 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 402 us MUL_MAT_ID(ffn_moe_down-38): 173 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 418 us MUL_MAT_ID(ffn_moe_down-39): 677 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 348 us MUL_MAT_ID(ffn_moe_down-40): 579 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 360 us MUL_MAT_ID(ffn_moe_down-41): 199 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 427 us MUL_MAT_ID(ffn_moe_down-42): 264 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 398 us MUL_MAT_ID(ffn_moe_down-43): 584 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 382 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 376 us MUL_MAT_ID(ffn_moe_down-45): 236 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 376 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 381 us MUL_MAT_ID(ffn_moe_down-47): 184 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 329 us MUL_MAT_ID(ffn_moe_down-48): 209 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 492 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 410 us MUL_MAT_ID(ffn_moe_down-50): 248 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 363 us MUL_MAT_ID(ffn_moe_down-51): 193 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 432 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 455 us MUL_MAT_ID(ffn_moe_down-53): 232 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 407 us MUL_MAT_ID(ffn_moe_down-54): 684 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 417 us MUL_MAT_ID(ffn_moe_down-55): 414 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 373 us MUL_MAT_ID(ffn_moe_down-56): 174 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 391 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 424 us MUL_MAT_ID(ffn_moe_down-58): 434 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 379 us MUL_MAT_ID(ffn_moe_down-59): 238 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 389 us MUL_MAT_ID(ffn_moe_down-60): 214 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 407 us MUL_MAT_ID(ffn_moe_down-61): 421 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 432 us MUL_MAT_ID(ffn_moe_down-62): 256 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 441 us MUL_MAT_ID(ffn_moe_down-63): 214 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 393 us MUL_MAT_ID(ffn_moe_down-64): 230 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 545 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 496 us MUL_MAT_ID(ffn_moe_down-66): 205 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 446 us MUL_MAT_ID(ffn_moe_down-67): 298 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 426 us MUL_MAT_ID(ffn_moe_down-68): 322 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 448 us MUL_MAT_ID(ffn_moe_down-69): 272 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 549 us MUL_MAT_ID(ffn_moe_down-70): 419 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 420 us MUL_MAT_ID(ffn_moe_down-71): 276 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 535 us MUL_MAT_ID(ffn_moe_down-72): 171 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 421 us MUL_MAT_ID(ffn_moe_down-73): 202 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 389 us MUL_MAT_ID(ffn_moe_down-74): 172 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 383 us MUL_MAT_ID(ffn_moe_down-75): 300 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 441 us MUL_MAT_ID(ffn_moe_down-76): 204 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 440 us MUL_MAT_ID(ffn_moe_down-77): 191 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 456 us MUL_MAT_ID(ffn_moe_down-78): 229 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 431 us MUL_MAT_ID(ffn_moe_down-79): 267 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 635 us MUL_MAT_ID(ffn_moe_down-80): 245 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 414 us MUL_MAT_ID(ffn_moe_down-81): 216 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 471 us MUL_MAT_ID(ffn_moe_down-82): 228 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 373 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 416 us MUL_MAT_ID(ffn_moe_down-84): 226 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 442 us MUL_MAT_ID(ffn_moe_down-85): 226 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 367 us MUL_MAT_ID(ffn_moe_down-86): 180 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 402 us MUL_MAT_ID(ffn_moe_down-87): 243 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 364 us MUL_MAT_ID(ffn_moe_down-88): 383 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 395 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 476 us MUL_MAT_ID(ffn_moe_down-90): 249 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 451 us MUL_MAT_ID(ffn_moe_down-91): 211 us ggml_barrier(...): 253 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 441 us MUL_MAT_ID(ffn_moe_down-25): 191 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 717 us MUL_MAT_ID(ffn_moe_down-26): 326 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 364 us MUL_MAT_ID(ffn_moe_down-27): 164 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 414 us MUL_MAT_ID(ffn_moe_down-28): 265 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 448 us MUL_MAT_ID(ffn_moe_down-29): 202 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 382 us MUL_MAT_ID(ffn_moe_down-30): 219 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 394 us MUL_MAT_ID(ffn_moe_down-31): 417 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 496 us MUL_MAT_ID(ffn_moe_down-32): 221 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 405 us MUL_MAT_ID(ffn_moe_down-33): 205 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 466 us MUL_MAT_ID(ffn_moe_down-34): 285 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 373 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 489 us MUL_MAT_ID(ffn_moe_down-36): 197 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 398 us MUL_MAT_ID(ffn_moe_down-37): 254 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 438 us MUL_MAT_ID(ffn_moe_down-38): 519 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 370 us MUL_MAT_ID(ffn_moe_down-39): 275 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 400 us MUL_MAT_ID(ffn_moe_down-40): 364 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 421 us MUL_MAT_ID(ffn_moe_down-41): 273 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 391 us MUL_MAT_ID(ffn_moe_down-42): 212 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 401 us MUL_MAT_ID(ffn_moe_down-43): 572 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 389 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 404 us MUL_MAT_ID(ffn_moe_down-45): 217 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 434 us MUL_MAT_ID(ffn_moe_down-46): 247 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 437 us MUL_MAT_ID(ffn_moe_down-47): 183 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 363 us MUL_MAT_ID(ffn_moe_down-48): 209 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 384 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 420 us MUL_MAT_ID(ffn_moe_down-50): 276 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 408 us MUL_MAT_ID(ffn_moe_down-51): 240 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 407 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 339 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 716 us MUL_MAT_ID(ffn_moe_down-54): 337 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 361 us MUL_MAT_ID(ffn_moe_down-55): 172 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 520 us MUL_MAT_ID(ffn_moe_down-56): 210 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 381 us MUL_MAT_ID(ffn_moe_down-57): 170 us ggml_barrier(...): 371 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 368 us MUL_MAT_ID(ffn_moe_down-58): 285 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 469 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 425 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 451 us MUL_MAT_ID(ffn_moe_down-61): 477 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 446 us MUL_MAT_ID(ffn_moe_down-62): 275 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 438 us MUL_MAT_ID(ffn_moe_down-63): 304 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 407 us MUL_MAT_ID(ffn_moe_down-64): 407 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 468 us MUL_MAT_ID(ffn_moe_down-65): 318 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 397 us MUL_MAT_ID(ffn_moe_down-66): 208 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 539 us MUL_MAT_ID(ffn_moe_down-67): 217 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 404 us MUL_MAT_ID(ffn_moe_down-68): 182 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 402 us MUL_MAT_ID(ffn_moe_down-69): 207 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 469 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 524 us MUL_MAT_ID(ffn_moe_down-71): 209 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 425 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 473 us MUL_MAT_ID(ffn_moe_down-73): 229 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 422 us MUL_MAT_ID(ffn_moe_down-74): 159 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 404 us MUL_MAT_ID(ffn_moe_down-75): 180 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 521 us MUL_MAT_ID(ffn_moe_down-76): 214 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 375 us MUL_MAT_ID(ffn_moe_down-77): 295 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 396 us MUL_MAT_ID(ffn_moe_down-78): 409 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 495 us MUL_MAT_ID(ffn_moe_down-79): 225 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 396 us MUL_MAT_ID(ffn_moe_down-80): 208 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 461 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 454 us MUL_MAT_ID(ffn_moe_down-82): 249 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 441 us MUL_MAT_ID(ffn_moe_down-83): 286 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 675 us MUL_MAT_ID(ffn_moe_down-84): 379 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 695 us MUL_MAT_ID(ffn_moe_down-85): 282 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 420 us MUL_MAT_ID(ffn_moe_down-86): 195 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 429 us MUL_MAT_ID(ffn_moe_down-87): 199 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 500 us MUL_MAT_ID(ffn_moe_down-88): 200 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 399 us MUL_MAT_ID(ffn_moe_down-89): 187 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 470 us MUL_MAT_ID(ffn_moe_down-90): 243 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 463 us MUL_MAT_ID(ffn_moe_down-91): 314 us ggml_barrier(...): 156 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 407 us MUL_MAT_ID(ffn_moe_down-25): 204 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 695 us MUL_MAT_ID(ffn_moe_down-26): 261 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 370 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 349 us MUL_MAT_ID(ffn_moe_down-28): 212 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 407 us MUL_MAT_ID(ffn_moe_down-29): 281 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 361 us MUL_MAT_ID(ffn_moe_down-30): 248 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 537 us MUL_MAT_ID(ffn_moe_down-31): 223 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 397 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 409 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 363 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 377 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 469 us MUL_MAT_ID(ffn_moe_down-36): 249 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 481 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 427 us MUL_MAT_ID(ffn_moe_down-38): 223 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 364 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 392 us MUL_MAT_ID(ffn_moe_down-40): 223 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 392 us MUL_MAT_ID(ffn_moe_down-41): 350 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 517 us MUL_MAT_ID(ffn_moe_down-42): 163 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 396 us MUL_MAT_ID(ffn_moe_down-43): 198 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 377 us MUL_MAT_ID(ffn_moe_down-44): 316 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 401 us MUL_MAT_ID(ffn_moe_down-45): 232 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 374 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 408 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 340 us MUL_MAT_ID(ffn_moe_down-48): 185 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 373 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 363 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 350 us MUL_MAT_ID(ffn_moe_down-51): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 336 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 379 us MUL_MAT_ID(ffn_moe_down-53): 199 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 381 us MUL_MAT_ID(ffn_moe_down-54): 324 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 401 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 408 us MUL_MAT_ID(ffn_moe_down-56): 245 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 361 us MUL_MAT_ID(ffn_moe_down-57): 211 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 350 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 368 us MUL_MAT_ID(ffn_moe_down-59): 206 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 482 us MUL_MAT_ID(ffn_moe_down-60): 167 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 403 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 417 us MUL_MAT_ID(ffn_moe_down-62): 227 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 407 us MUL_MAT_ID(ffn_moe_down-63): 628 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 394 us MUL_MAT_ID(ffn_moe_down-64): 222 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 384 us MUL_MAT_ID(ffn_moe_down-65): 206 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 383 us MUL_MAT_ID(ffn_moe_down-66): 186 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 340 us MUL_MAT_ID(ffn_moe_down-67): 200 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 419 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 544 us MUL_MAT_ID(ffn_moe_down-69): 209 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 358 us MUL_MAT_ID(ffn_moe_down-70): 213 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 396 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 364 us MUL_MAT_ID(ffn_moe_down-72): 209 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 406 us MUL_MAT_ID(ffn_moe_down-73): 199 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 385 us MUL_MAT_ID(ffn_moe_down-74): 216 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 442 us MUL_MAT_ID(ffn_moe_down-75): 250 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 379 us MUL_MAT_ID(ffn_moe_down-76): 372 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 383 us MUL_MAT_ID(ffn_moe_down-77): 408 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 401 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 439 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 433 us MUL_MAT_ID(ffn_moe_down-80): 257 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 497 us MUL_MAT_ID(ffn_moe_down-81): 278 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 462 us MUL_MAT_ID(ffn_moe_down-82): 222 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 409 us MUL_MAT_ID(ffn_moe_down-83): 208 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 413 us MUL_MAT_ID(ffn_moe_down-84): 247 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 529 us MUL_MAT_ID(ffn_moe_down-85): 373 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 387 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 399 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 443 us MUL_MAT_ID(ffn_moe_down-88): 221 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 470 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 425 us MUL_MAT_ID(ffn_moe_down-90): 236 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 460 us MUL_MAT_ID(ffn_moe_down-91): 248 us ggml_barrier(...): 89 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 423 us MUL_MAT_ID(ffn_moe_down-25): 191 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 433 us MUL_MAT_ID(ffn_moe_down-26): 411 us ggml_barrier(...): 346 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 396 us MUL_MAT_ID(ffn_moe_down-27): 292 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 367 us MUL_MAT_ID(ffn_moe_down-28): 241 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 414 us MUL_MAT_ID(ffn_moe_down-29): 187 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 392 us MUL_MAT_ID(ffn_moe_down-30): 206 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 527 us MUL_MAT_ID(ffn_moe_down-31): 268 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 396 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 404 us MUL_MAT_ID(ffn_moe_down-33): 375 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 396 us MUL_MAT_ID(ffn_moe_down-34): 189 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 379 us MUL_MAT_ID(ffn_moe_down-35): 200 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 486 us MUL_MAT_ID(ffn_moe_down-36): 328 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 395 us MUL_MAT_ID(ffn_moe_down-37): 323 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 370 us MUL_MAT_ID(ffn_moe_down-38): 474 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 369 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 487 us MUL_MAT_ID(ffn_moe_down-40): 207 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 374 us MUL_MAT_ID(ffn_moe_down-41): 187 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 367 us MUL_MAT_ID(ffn_moe_down-42): 464 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 408 us MUL_MAT_ID(ffn_moe_down-43): 206 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 393 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 377 us MUL_MAT_ID(ffn_moe_down-45): 221 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 466 us MUL_MAT_ID(ffn_moe_down-46): 330 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 377 us MUL_MAT_ID(ffn_moe_down-47): 373 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 398 us MUL_MAT_ID(ffn_moe_down-48): 241 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 413 us MUL_MAT_ID(ffn_moe_down-49): 172 us ggml_barrier(...): 399 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 523 us MUL_MAT_ID(ffn_moe_down-50): 207 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 424 us MUL_MAT_ID(ffn_moe_down-51): 283 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 460 us MUL_MAT_ID(ffn_moe_down-52): 232 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 454 us MUL_MAT_ID(ffn_moe_down-53): 305 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 364 us MUL_MAT_ID(ffn_moe_down-54): 250 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 462 us MUL_MAT_ID(ffn_moe_down-55): 194 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 436 us MUL_MAT_ID(ffn_moe_down-56): 207 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 423 us MUL_MAT_ID(ffn_moe_down-57): 206 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 387 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 391 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 464 us MUL_MAT_ID(ffn_moe_down-60): 297 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 418 us MUL_MAT_ID(ffn_moe_down-61): 208 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 349 us MUL_MAT_ID(ffn_moe_down-62): 192 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 396 us MUL_MAT_ID(ffn_moe_down-63): 204 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 410 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 418 us MUL_MAT_ID(ffn_moe_down-65): 202 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 389 us MUL_MAT_ID(ffn_moe_down-66): 257 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 523 us MUL_MAT_ID(ffn_moe_down-67): 224 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 467 us MUL_MAT_ID(ffn_moe_down-68): 295 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 397 us MUL_MAT_ID(ffn_moe_down-69): 240 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 411 us MUL_MAT_ID(ffn_moe_down-70): 235 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 471 us MUL_MAT_ID(ffn_moe_down-71): 280 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 375 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 434 us MUL_MAT_ID(ffn_moe_down-73): 232 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 410 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 422 us MUL_MAT_ID(ffn_moe_down-75): 212 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 434 us MUL_MAT_ID(ffn_moe_down-76): 280 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 508 us MUL_MAT_ID(ffn_moe_down-78): 230 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 436 us MUL_MAT_ID(ffn_moe_down-79): 249 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 647 us MUL_MAT_ID(ffn_moe_down-80): 328 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 358 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 412 us MUL_MAT_ID(ffn_moe_down-82): 206 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 475 us MUL_MAT_ID(ffn_moe_down-83): 248 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 447 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 348 us MUL_MAT_ID(ffn_moe_down-85): 213 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 374 us MUL_MAT_ID(ffn_moe_down-86): 208 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 414 us MUL_MAT_ID(ffn_moe_down-87): 182 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 404 us MUL_MAT_ID(ffn_moe_down-88): 231 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 419 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 403 us MUL_MAT_ID(ffn_moe_down-90): 248 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 517 us MUL_MAT_ID(ffn_moe_down-91): 479 us ggml_barrier(...): 62 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 397 us MUL_MAT_ID(ffn_moe_down-25): 306 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 432 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 464 us MUL_MAT_ID(ffn_moe_down-27): 185 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 376 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 385 us MUL_MAT_ID(ffn_moe_down-29): 587 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 375 us MUL_MAT_ID(ffn_moe_down-30): 205 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 416 us MUL_MAT_ID(ffn_moe_down-31): 184 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 424 us MUL_MAT_ID(ffn_moe_down-32): 199 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 522 us MUL_MAT_ID(ffn_moe_down-33): 210 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 534 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 398 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 386 us MUL_MAT_ID(ffn_moe_down-36): 190 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 441 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 459 us MUL_MAT_ID(ffn_moe_down-38): 227 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 393 us MUL_MAT_ID(ffn_moe_down-39): 714 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 446 us MUL_MAT_ID(ffn_moe_down-40): 257 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 403 us MUL_MAT_ID(ffn_moe_down-41): 182 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 445 us MUL_MAT_ID(ffn_moe_down-42): 219 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 440 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 423 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 414 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 373 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 395 us MUL_MAT_ID(ffn_moe_down-47): 210 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 499 us MUL_MAT_ID(ffn_moe_down-48): 334 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 418 us MUL_MAT_ID(ffn_moe_down-49): 209 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 553 us MUL_MAT_ID(ffn_moe_down-50): 364 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 415 us MUL_MAT_ID(ffn_moe_down-51): 202 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 399 us MUL_MAT_ID(ffn_moe_down-52): 186 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 457 us MUL_MAT_ID(ffn_moe_down-53): 473 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 378 us MUL_MAT_ID(ffn_moe_down-54): 258 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 410 us MUL_MAT_ID(ffn_moe_down-55): 187 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 375 us MUL_MAT_ID(ffn_moe_down-56): 189 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 512 us MUL_MAT_ID(ffn_moe_down-57): 187 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 460 us MUL_MAT_ID(ffn_moe_down-58): 263 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 456 us MUL_MAT_ID(ffn_moe_down-59): 312 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 481 us MUL_MAT_ID(ffn_moe_down-60): 289 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 394 us MUL_MAT_ID(ffn_moe_down-61): 542 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 364 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 377 us MUL_MAT_ID(ffn_moe_down-63): 200 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 413 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 431 us MUL_MAT_ID(ffn_moe_down-65): 212 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 402 us MUL_MAT_ID(ffn_moe_down-66): 222 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 468 us MUL_MAT_ID(ffn_moe_down-67): 674 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 656 us MUL_MAT_ID(ffn_moe_down-68): 211 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 442 us MUL_MAT_ID(ffn_moe_down-69): 175 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 385 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 397 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 375 us MUL_MAT_ID(ffn_moe_down-72): 203 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 516 us MUL_MAT_ID(ffn_moe_down-73): 202 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 389 us MUL_MAT_ID(ffn_moe_down-74): 175 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 432 us MUL_MAT_ID(ffn_moe_down-75): 184 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 417 us MUL_MAT_ID(ffn_moe_down-76): 175 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 353 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 475 us MUL_MAT_ID(ffn_moe_down-78): 209 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 369 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 360 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 419 us MUL_MAT_ID(ffn_moe_down-80): 204 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 496 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 380 us MUL_MAT_ID(ffn_moe_down-82): 242 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 430 us MUL_MAT_ID(ffn_moe_down-83): 212 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 400 us MUL_MAT_ID(ffn_moe_down-84): 255 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 397 us MUL_MAT_ID(ffn_moe_down-85): 230 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 416 us MUL_MAT_ID(ffn_moe_down-86): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 351 us MUL_MAT_ID(ffn_moe_down-87): 288 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 468 us MUL_MAT_ID(ffn_moe_down-88): 173 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 387 us MUL_MAT_ID(ffn_moe_down-89): 410 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 373 us MUL_MAT_ID(ffn_moe_down-90): 243 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 397 us MUL_MAT_ID(ffn_moe_down-91): 315 us ggml_barrier(...): 176 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 443 us MUL_MAT_ID(ffn_moe_down-25): 436 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 399 us MUL_MAT_ID(ffn_moe_down-26): 285 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 411 us MUL_MAT_ID(ffn_moe_down-27): 303 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 416 us MUL_MAT_ID(ffn_moe_down-28): 185 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 460 us MUL_MAT_ID(ffn_moe_down-29): 187 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 454 us MUL_MAT_ID(ffn_moe_down-30): 197 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 404 us MUL_MAT_ID(ffn_moe_down-31): 393 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 407 us MUL_MAT_ID(ffn_moe_down-32): 256 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 522 us MUL_MAT_ID(ffn_moe_down-33): 287 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 414 us MUL_MAT_ID(ffn_moe_down-34): 209 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 350 us MUL_MAT_ID(ffn_moe_down-35): 213 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 489 us MUL_MAT_ID(ffn_moe_down-36): 186 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 422 us MUL_MAT_ID(ffn_moe_down-37): 252 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 192 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 383 us MUL_MAT_ID(ffn_moe_down-39): 736 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 379 us MUL_MAT_ID(ffn_moe_down-40): 480 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 469 us MUL_MAT_ID(ffn_moe_down-41): 217 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 472 us MUL_MAT_ID(ffn_moe_down-42): 212 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 383 us MUL_MAT_ID(ffn_moe_down-43): 212 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 432 us MUL_MAT_ID(ffn_moe_down-44): 280 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 415 us MUL_MAT_ID(ffn_moe_down-45): 412 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 368 us MUL_MAT_ID(ffn_moe_down-46): 214 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 392 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 391 us MUL_MAT_ID(ffn_moe_down-48): 374 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 392 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 364 us MUL_MAT_ID(ffn_moe_down-50): 345 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 441 us MUL_MAT_ID(ffn_moe_down-51): 213 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 387 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 394 us MUL_MAT_ID(ffn_moe_down-53): 182 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 551 us MUL_MAT_ID(ffn_moe_down-54): 243 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 387 us MUL_MAT_ID(ffn_moe_down-55): 374 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 417 us MUL_MAT_ID(ffn_moe_down-56): 201 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 342 us MUL_MAT_ID(ffn_moe_down-57): 193 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 380 us MUL_MAT_ID(ffn_moe_down-58): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 462 us MUL_MAT_ID(ffn_moe_down-59): 190 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 348 us MUL_MAT_ID(ffn_moe_down-60): 202 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 419 us MUL_MAT_ID(ffn_moe_down-61): 268 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 368 us MUL_MAT_ID(ffn_moe_down-62): 209 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 384 us MUL_MAT_ID(ffn_moe_down-63): 201 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 391 us MUL_MAT_ID(ffn_moe_down-64): 212 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 364 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 401 us MUL_MAT_ID(ffn_moe_down-66): 413 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 449 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 450 us MUL_MAT_ID(ffn_moe_down-68): 200 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 395 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 350 us MUL_MAT_ID(ffn_moe_down-70): 579 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 369 us MUL_MAT_ID(ffn_moe_down-71): 203 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 352 us MUL_MAT_ID(ffn_moe_down-72): 240 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 264 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 505 us MUL_MAT_ID(ffn_moe_down-74): 185 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 408 us MUL_MAT_ID(ffn_moe_down-75): 567 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 388 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 377 us MUL_MAT_ID(ffn_moe_down-77): 183 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 383 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 667 us MUL_MAT_ID(ffn_moe_down-79): 265 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 379 us MUL_MAT_ID(ffn_moe_down-80): 211 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 385 us MUL_MAT_ID(ffn_moe_down-81): 209 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 489 us MUL_MAT_ID(ffn_moe_down-82): 283 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 431 us MUL_MAT_ID(ffn_moe_down-83): 584 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 374 us MUL_MAT_ID(ffn_moe_down-84): 243 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 429 us MUL_MAT_ID(ffn_moe_down-85): 708 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 380 us MUL_MAT_ID(ffn_moe_down-86): 200 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 404 us MUL_MAT_ID(ffn_moe_down-87): 228 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 446 us MUL_MAT_ID(ffn_moe_down-88): 287 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 402 us MUL_MAT_ID(ffn_moe_down-89): 209 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 390 us MUL_MAT_ID(ffn_moe_down-90): 241 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 430 us MUL_MAT_ID(ffn_moe_down-91): 355 us ggml_barrier(...): 25 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 405 us MUL_MAT_ID(ffn_moe_down-25): 196 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 448 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 401 us MUL_MAT_ID(ffn_moe_down-27): 208 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 341 us MUL_MAT_ID(ffn_moe_down-28): 242 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 431 us MUL_MAT_ID(ffn_moe_down-29): 261 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 441 us MUL_MAT_ID(ffn_moe_down-30): 212 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 792 us MUL_MAT_ID(ffn_moe_down-31): 424 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 694 us MUL_MAT_ID(ffn_moe_down-32): 322 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 377 us MUL_MAT_ID(ffn_moe_down-33): 216 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 384 us MUL_MAT_ID(ffn_moe_down-34): 186 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 422 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 457 us MUL_MAT_ID(ffn_moe_down-36): 573 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 383 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 378 us MUL_MAT_ID(ffn_moe_down-38): 507 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 397 us MUL_MAT_ID(ffn_moe_down-39): 424 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 500 us MUL_MAT_ID(ffn_moe_down-40): 448 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 656 us MUL_MAT_ID(ffn_moe_down-41): 283 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 365 us MUL_MAT_ID(ffn_moe_down-42): 189 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 407 us MUL_MAT_ID(ffn_moe_down-43): 486 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 398 us MUL_MAT_ID(ffn_moe_down-44): 200 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 408 us MUL_MAT_ID(ffn_moe_down-45): 192 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 350 us MUL_MAT_ID(ffn_moe_down-46): 205 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 434 us MUL_MAT_ID(ffn_moe_down-47): 209 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 398 us MUL_MAT_ID(ffn_moe_down-48): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 384 us MUL_MAT_ID(ffn_moe_down-49): 347 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 404 us MUL_MAT_ID(ffn_moe_down-50): 217 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 430 us MUL_MAT_ID(ffn_moe_down-51): 212 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 390 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 382 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 393 us MUL_MAT_ID(ffn_moe_down-54): 224 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 679 us MUL_MAT_ID(ffn_moe_down-55): 213 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 400 us MUL_MAT_ID(ffn_moe_down-56): 228 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 429 us MUL_MAT_ID(ffn_moe_down-57): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 394 us MUL_MAT_ID(ffn_moe_down-58): 279 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 508 us MUL_MAT_ID(ffn_moe_down-59): 249 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 447 us MUL_MAT_ID(ffn_moe_down-60): 179 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 398 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 418 us MUL_MAT_ID(ffn_moe_down-62): 241 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 356 us MUL_MAT_ID(ffn_moe_down-63): 311 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 461 us MUL_MAT_ID(ffn_moe_down-64): 303 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 432 us MUL_MAT_ID(ffn_moe_down-65): 184 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 487 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 429 us MUL_MAT_ID(ffn_moe_down-67): 192 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 430 us MUL_MAT_ID(ffn_moe_down-68): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 436 us MUL_MAT_ID(ffn_moe_down-69): 265 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 443 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 384 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 269 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 400 us MUL_MAT_ID(ffn_moe_down-73): 654 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 394 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 429 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 421 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 498 us MUL_MAT_ID(ffn_moe_down-77): 500 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 457 us MUL_MAT_ID(ffn_moe_down-78): 189 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 348 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 421 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 478 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 380 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 403 us MUL_MAT_ID(ffn_moe_down-82): 212 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 431 us MUL_MAT_ID(ffn_moe_down-83): 227 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 394 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 453 us MUL_MAT_ID(ffn_moe_down-85): 611 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 412 us MUL_MAT_ID(ffn_moe_down-86): 189 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 407 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 426 us MUL_MAT_ID(ffn_moe_down-88): 204 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 442 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 379 us MUL_MAT_ID(ffn_moe_down-90): 547 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 386 us MUL_MAT_ID(ffn_moe_down-91): 264 us ggml_barrier(...): 274 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 429 us MUL_MAT_ID(ffn_moe_down-25): 227 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 384 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 422 us MUL_MAT_ID(ffn_moe_down-27): 277 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 378 us MUL_MAT_ID(ffn_moe_down-28): 214 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 408 us MUL_MAT_ID(ffn_moe_down-29): 207 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 402 us MUL_MAT_ID(ffn_moe_down-30): 176 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 386 us MUL_MAT_ID(ffn_moe_down-31): 204 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 457 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 427 us MUL_MAT_ID(ffn_moe_down-33): 191 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 490 us MUL_MAT_ID(ffn_moe_down-34): 171 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 461 us MUL_MAT_ID(ffn_moe_down-35): 268 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 388 us MUL_MAT_ID(ffn_moe_down-36): 309 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 465 us MUL_MAT_ID(ffn_moe_down-37): 303 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 776 us MUL_MAT_ID(ffn_moe_down-38): 345 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 389 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 402 us MUL_MAT_ID(ffn_moe_down-40): 245 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 357 us MUL_MAT_ID(ffn_moe_down-41): 177 us ggml_barrier(...): 383 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 491 us MUL_MAT_ID(ffn_moe_down-42): 193 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 403 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 406 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 400 us MUL_MAT_ID(ffn_moe_down-45): 316 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 383 us MUL_MAT_ID(ffn_moe_down-46): 179 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 470 us MUL_MAT_ID(ffn_moe_down-47): 284 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 416 us MUL_MAT_ID(ffn_moe_down-48): 440 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 425 us MUL_MAT_ID(ffn_moe_down-49): 238 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 397 us MUL_MAT_ID(ffn_moe_down-50): 204 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 430 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 377 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 431 us MUL_MAT_ID(ffn_moe_down-53): 591 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 387 us MUL_MAT_ID(ffn_moe_down-54): 597 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 480 us MUL_MAT_ID(ffn_moe_down-55): 434 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 391 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 379 us MUL_MAT_ID(ffn_moe_down-57): 187 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 392 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 433 us MUL_MAT_ID(ffn_moe_down-59): 303 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 452 us MUL_MAT_ID(ffn_moe_down-60): 233 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 425 us MUL_MAT_ID(ffn_moe_down-61): 228 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 441 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 360 us MUL_MAT_ID(ffn_moe_down-63): 552 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 492 us MUL_MAT_ID(ffn_moe_down-64): 228 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 418 us MUL_MAT_ID(ffn_moe_down-65): 211 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 392 us MUL_MAT_ID(ffn_moe_down-66): 432 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 392 us MUL_MAT_ID(ffn_moe_down-67): 184 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 473 us MUL_MAT_ID(ffn_moe_down-68): 188 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 375 us MUL_MAT_ID(ffn_moe_down-69): 207 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 356 us MUL_MAT_ID(ffn_moe_down-70): 270 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 667 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 417 us MUL_MAT_ID(ffn_moe_down-72): 182 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 402 us MUL_MAT_ID(ffn_moe_down-73): 196 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 419 us MUL_MAT_ID(ffn_moe_down-74): 186 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 396 us MUL_MAT_ID(ffn_moe_down-75): 215 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 579 us MUL_MAT_ID(ffn_moe_down-76): 331 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 463 us MUL_MAT_ID(ffn_moe_down-77): 207 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 501 us MUL_MAT_ID(ffn_moe_down-78): 214 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 437 us MUL_MAT_ID(ffn_moe_down-79): 234 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 417 us MUL_MAT_ID(ffn_moe_down-80): 303 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 442 us MUL_MAT_ID(ffn_moe_down-81): 279 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 392 us MUL_MAT_ID(ffn_moe_down-82): 597 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 413 us MUL_MAT_ID(ffn_moe_down-83): 210 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 397 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 558 us MUL_MAT_ID(ffn_moe_down-85): 244 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 392 us MUL_MAT_ID(ffn_moe_down-86): 230 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 415 us MUL_MAT_ID(ffn_moe_down-87): 204 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 449 us MUL_MAT_ID(ffn_moe_down-88): 226 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 414 us MUL_MAT_ID(ffn_moe_down-89): 186 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 397 us MUL_MAT_ID(ffn_moe_down-90): 639 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 375 us MUL_MAT_ID(ffn_moe_down-91): 278 us ggml_barrier(...): 50 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 387 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 385 us MUL_MAT_ID(ffn_moe_down-26): 184 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 522 us MUL_MAT_ID(ffn_moe_down-27): 280 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 185 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 423 us MUL_MAT_ID(ffn_moe_down-29): 173 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 386 us MUL_MAT_ID(ffn_moe_down-30): 284 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 388 us MUL_MAT_ID(ffn_moe_down-31): 226 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 394 us MUL_MAT_ID(ffn_moe_down-32): 213 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 421 us MUL_MAT_ID(ffn_moe_down-33): 184 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 437 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 380 us MUL_MAT_ID(ffn_moe_down-35): 221 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 345 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 477 us MUL_MAT_ID(ffn_moe_down-37): 264 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 368 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 438 us MUL_MAT_ID(ffn_moe_down-39): 253 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 364 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 497 us MUL_MAT_ID(ffn_moe_down-41): 235 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 473 us MUL_MAT_ID(ffn_moe_down-42): 447 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 535 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 433 us MUL_MAT_ID(ffn_moe_down-44): 328 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 402 us MUL_MAT_ID(ffn_moe_down-45): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 348 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 452 us MUL_MAT_ID(ffn_moe_down-47): 276 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 395 us MUL_MAT_ID(ffn_moe_down-48): 228 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 433 us MUL_MAT_ID(ffn_moe_down-49): 210 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 405 us MUL_MAT_ID(ffn_moe_down-50): 205 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 347 us MUL_MAT_ID(ffn_moe_down-51): 478 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 408 us MUL_MAT_ID(ffn_moe_down-52): 295 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 456 us MUL_MAT_ID(ffn_moe_down-53): 284 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 410 us MUL_MAT_ID(ffn_moe_down-54): 289 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 370 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 394 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 595 us MUL_MAT_ID(ffn_moe_down-57): 394 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 396 us MUL_MAT_ID(ffn_moe_down-58): 223 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 401 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 408 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 406 us MUL_MAT_ID(ffn_moe_down-61): 202 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 398 us MUL_MAT_ID(ffn_moe_down-62): 208 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 470 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 508 us MUL_MAT_ID(ffn_moe_down-64): 177 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 440 us MUL_MAT_ID(ffn_moe_down-65): 163 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 539 us MUL_MAT_ID(ffn_moe_down-66): 336 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 424 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 445 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 377 us MUL_MAT_ID(ffn_moe_down-69): 231 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 392 us MUL_MAT_ID(ffn_moe_down-70): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 462 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 528 us MUL_MAT_ID(ffn_moe_down-72): 164 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 406 us MUL_MAT_ID(ffn_moe_down-73): 179 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 439 us MUL_MAT_ID(ffn_moe_down-74): 207 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 429 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 417 us MUL_MAT_ID(ffn_moe_down-76): 230 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 422 us MUL_MAT_ID(ffn_moe_down-77): 290 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 435 us MUL_MAT_ID(ffn_moe_down-78): 159 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 383 us MUL_MAT_ID(ffn_moe_down-79): 173 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 432 us MUL_MAT_ID(ffn_moe_down-80): 213 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 516 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 375 us MUL_MAT_ID(ffn_moe_down-82): 272 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 461 us MUL_MAT_ID(ffn_moe_down-83): 585 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 371 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 424 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 393 us MUL_MAT_ID(ffn_moe_down-86): 218 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 403 us MUL_MAT_ID(ffn_moe_down-87): 206 us ggml_barrier(...): 335 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 440 us MUL_MAT_ID(ffn_moe_down-88): 231 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 423 us MUL_MAT_ID(ffn_moe_down-89): 463 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 653 us MUL_MAT_ID(ffn_moe_down-90): 258 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 440 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 35 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 381 us MUL_MAT_ID(ffn_moe_down-25): 206 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 402 us MUL_MAT_ID(ffn_moe_down-26): 262 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 370 us MUL_MAT_ID(ffn_moe_down-27): 240 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 382 us MUL_MAT_ID(ffn_moe_down-28): 379 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 411 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 357 us MUL_MAT_ID(ffn_moe_down-30): 655 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 430 us MUL_MAT_ID(ffn_moe_down-31): 192 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 486 us MUL_MAT_ID(ffn_moe_down-32): 171 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 423 us MUL_MAT_ID(ffn_moe_down-33): 222 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 389 us MUL_MAT_ID(ffn_moe_down-34): 215 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 391 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 380 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 445 us MUL_MAT_ID(ffn_moe_down-37): 208 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 432 us MUL_MAT_ID(ffn_moe_down-38): 183 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 431 us MUL_MAT_ID(ffn_moe_down-39): 260 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 424 us MUL_MAT_ID(ffn_moe_down-40): 243 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 384 us MUL_MAT_ID(ffn_moe_down-41): 239 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 505 us MUL_MAT_ID(ffn_moe_down-42): 272 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 374 us MUL_MAT_ID(ffn_moe_down-43): 242 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 374 us MUL_MAT_ID(ffn_moe_down-44): 193 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 476 us MUL_MAT_ID(ffn_moe_down-45): 175 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 405 us MUL_MAT_ID(ffn_moe_down-46): 237 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 505 us MUL_MAT_ID(ffn_moe_down-47): 445 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 566 us MUL_MAT_ID(ffn_moe_down-48): 217 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 404 us MUL_MAT_ID(ffn_moe_down-49): 201 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 400 us MUL_MAT_ID(ffn_moe_down-50): 240 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 460 us MUL_MAT_ID(ffn_moe_down-51): 194 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 471 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 428 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 425 us MUL_MAT_ID(ffn_moe_down-54): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 379 us MUL_MAT_ID(ffn_moe_down-55): 409 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 410 us MUL_MAT_ID(ffn_moe_down-56): 418 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 452 us MUL_MAT_ID(ffn_moe_down-57): 242 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 372 us MUL_MAT_ID(ffn_moe_down-58): 204 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 421 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 383 us MUL_MAT_ID(ffn_moe_down-60): 195 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 389 us MUL_MAT_ID(ffn_moe_down-61): 203 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 476 us MUL_MAT_ID(ffn_moe_down-62): 179 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 389 us MUL_MAT_ID(ffn_moe_down-63): 218 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 405 us MUL_MAT_ID(ffn_moe_down-64): 245 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 380 us MUL_MAT_ID(ffn_moe_down-65): 250 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 409 us MUL_MAT_ID(ffn_moe_down-66): 249 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 373 us MUL_MAT_ID(ffn_moe_down-67): 424 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 378 us MUL_MAT_ID(ffn_moe_down-68): 240 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 411 us MUL_MAT_ID(ffn_moe_down-69): 183 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 379 us MUL_MAT_ID(ffn_moe_down-70): 188 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 368 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 368 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 468 us MUL_MAT_ID(ffn_moe_down-74): 426 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 372 us MUL_MAT_ID(ffn_moe_down-75): 218 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 380 us MUL_MAT_ID(ffn_moe_down-76): 173 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 655 us MUL_MAT_ID(ffn_moe_down-77): 344 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 372 us MUL_MAT_ID(ffn_moe_down-78): 172 us ggml_barrier(...): 363 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 444 us MUL_MAT_ID(ffn_moe_down-79): 354 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 392 us MUL_MAT_ID(ffn_moe_down-80): 248 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 429 us MUL_MAT_ID(ffn_moe_down-81): 212 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 413 us MUL_MAT_ID(ffn_moe_down-82): 192 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 476 us MUL_MAT_ID(ffn_moe_down-83): 204 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 406 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 533 us MUL_MAT_ID(ffn_moe_down-85): 260 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 496 us MUL_MAT_ID(ffn_moe_down-86): 641 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 521 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 455 us MUL_MAT_ID(ffn_moe_down-88): 199 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 498 us MUL_MAT_ID(ffn_moe_down-89): 179 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 464 us MUL_MAT_ID(ffn_moe_down-90): 237 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 414 us MUL_MAT_ID(ffn_moe_down-91): 472 us ggml_barrier(...): 35 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 406 us MUL_MAT_ID(ffn_moe_down-25): 213 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 410 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 520 us MUL_MAT_ID(ffn_moe_down-27): 227 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 409 us MUL_MAT_ID(ffn_moe_down-28): 435 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 515 us MUL_MAT_ID(ffn_moe_down-29): 316 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 402 us MUL_MAT_ID(ffn_moe_down-30): 260 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 440 us MUL_MAT_ID(ffn_moe_down-31): 202 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 359 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 359 us MUL_MAT_ID(ffn_moe_down-33): 191 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 393 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 389 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 332 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 416 us MUL_MAT_ID(ffn_moe_down-37): 308 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 396 us MUL_MAT_ID(ffn_moe_down-38): 216 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 429 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 384 us MUL_MAT_ID(ffn_moe_down-40): 217 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 394 us MUL_MAT_ID(ffn_moe_down-41): 198 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 427 us MUL_MAT_ID(ffn_moe_down-42): 215 us ggml_barrier(...): 335 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 403 us MUL_MAT_ID(ffn_moe_down-43): 187 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 401 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 444 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 389 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 428 us MUL_MAT_ID(ffn_moe_down-47): 367 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 354 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 465 us MUL_MAT_ID(ffn_moe_down-49): 207 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 413 us MUL_MAT_ID(ffn_moe_down-50): 182 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 402 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 462 us MUL_MAT_ID(ffn_moe_down-52): 351 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 419 us MUL_MAT_ID(ffn_moe_down-53): 427 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 366 us MUL_MAT_ID(ffn_moe_down-54): 235 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 449 us MUL_MAT_ID(ffn_moe_down-55): 244 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 685 us MUL_MAT_ID(ffn_moe_down-56): 564 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 556 us MUL_MAT_ID(ffn_moe_down-57): 218 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 515 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 402 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 429 us MUL_MAT_ID(ffn_moe_down-60): 544 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 390 us MUL_MAT_ID(ffn_moe_down-61): 378 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 372 us MUL_MAT_ID(ffn_moe_down-62): 288 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 418 us MUL_MAT_ID(ffn_moe_down-63): 185 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 408 us MUL_MAT_ID(ffn_moe_down-64): 179 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 479 us MUL_MAT_ID(ffn_moe_down-65): 307 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 650 us MUL_MAT_ID(ffn_moe_down-66): 255 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 354 us MUL_MAT_ID(ffn_moe_down-67): 257 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 375 us MUL_MAT_ID(ffn_moe_down-68): 211 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 440 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 411 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 369 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 378 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 408 us MUL_MAT_ID(ffn_moe_down-73): 237 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 381 us MUL_MAT_ID(ffn_moe_down-74): 244 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 426 us MUL_MAT_ID(ffn_moe_down-75): 219 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 425 us MUL_MAT_ID(ffn_moe_down-76): 185 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 394 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 425 us MUL_MAT_ID(ffn_moe_down-78): 272 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 410 us MUL_MAT_ID(ffn_moe_down-79): 381 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 545 us MUL_MAT_ID(ffn_moe_down-80): 212 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 397 us MUL_MAT_ID(ffn_moe_down-81): 316 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 406 us MUL_MAT_ID(ffn_moe_down-82): 199 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 425 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 438 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 434 us MUL_MAT_ID(ffn_moe_down-85): 293 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 419 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 359 us MUL_MAT_ID(ffn_moe_down-87): 238 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 426 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 443 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 465 us MUL_MAT_ID(ffn_moe_down-90): 514 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 409 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 142 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 401 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 353 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 629 us MUL_MAT_ID(ffn_moe_down-26): 481 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 355 us MUL_MAT_ID(ffn_moe_down-27): 414 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 364 us MUL_MAT_ID(ffn_moe_down-28): 331 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 430 us MUL_MAT_ID(ffn_moe_down-29): 188 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 383 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 400 us MUL_MAT_ID(ffn_moe_down-31): 201 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 401 us MUL_MAT_ID(ffn_moe_down-32): 206 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 404 us MUL_MAT_ID(ffn_moe_down-33): 169 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 425 us MUL_MAT_ID(ffn_moe_down-34): 202 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 401 us MUL_MAT_ID(ffn_moe_down-35): 220 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 370 us MUL_MAT_ID(ffn_moe_down-36): 205 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 411 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 376 us MUL_MAT_ID(ffn_moe_down-38): 181 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 400 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 548 us MUL_MAT_ID(ffn_moe_down-40): 314 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 404 us MUL_MAT_ID(ffn_moe_down-41): 189 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 441 us MUL_MAT_ID(ffn_moe_down-42): 211 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 407 us MUL_MAT_ID(ffn_moe_down-43): 202 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 322 us MUL_MAT_ID(ffn_moe_down-44): 209 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 407 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 395 us MUL_MAT_ID(ffn_moe_down-46): 226 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 407 us MUL_MAT_ID(ffn_moe_down-47): 401 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 400 us MUL_MAT_ID(ffn_moe_down-48): 196 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 407 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 410 us MUL_MAT_ID(ffn_moe_down-50): 199 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 384 us MUL_MAT_ID(ffn_moe_down-51): 345 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 395 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 513 us MUL_MAT_ID(ffn_moe_down-53): 355 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 426 us MUL_MAT_ID(ffn_moe_down-54): 188 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 374 us MUL_MAT_ID(ffn_moe_down-55): 581 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 413 us MUL_MAT_ID(ffn_moe_down-56): 208 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 379 us MUL_MAT_ID(ffn_moe_down-57): 268 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 395 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 376 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 361 us MUL_MAT_ID(ffn_moe_down-60): 442 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 536 us MUL_MAT_ID(ffn_moe_down-61): 309 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 434 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 414 us MUL_MAT_ID(ffn_moe_down-63): 183 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 429 us MUL_MAT_ID(ffn_moe_down-64): 308 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 489 us MUL_MAT_ID(ffn_moe_down-65): 195 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 400 us MUL_MAT_ID(ffn_moe_down-66): 244 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 394 us MUL_MAT_ID(ffn_moe_down-67): 210 us ggml_barrier(...): 353 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 388 us MUL_MAT_ID(ffn_moe_down-68): 206 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 456 us MUL_MAT_ID(ffn_moe_down-69): 236 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 392 us MUL_MAT_ID(ffn_moe_down-70): 365 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 376 us MUL_MAT_ID(ffn_moe_down-71): 209 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 477 us MUL_MAT_ID(ffn_moe_down-72): 219 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 487 us MUL_MAT_ID(ffn_moe_down-73): 184 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 441 us MUL_MAT_ID(ffn_moe_down-74): 408 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 537 us MUL_MAT_ID(ffn_moe_down-75): 206 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 418 us MUL_MAT_ID(ffn_moe_down-76): 247 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 422 us MUL_MAT_ID(ffn_moe_down-77): 524 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 400 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 421 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 394 us MUL_MAT_ID(ffn_moe_down-80): 291 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 403 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 533 us MUL_MAT_ID(ffn_moe_down-82): 590 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 372 us MUL_MAT_ID(ffn_moe_down-83): 185 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 440 us MUL_MAT_ID(ffn_moe_down-84): 255 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 370 us MUL_MAT_ID(ffn_moe_down-85): 377 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 362 us MUL_MAT_ID(ffn_moe_down-86): 197 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 421 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 423 us MUL_MAT_ID(ffn_moe_down-88): 200 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 372 us MUL_MAT_ID(ffn_moe_down-89): 347 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 388 us MUL_MAT_ID(ffn_moe_down-90): 255 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 398 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 38 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 378 us MUL_MAT_ID(ffn_moe_down-25): 206 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 417 us MUL_MAT_ID(ffn_moe_down-26): 218 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 507 us MUL_MAT_ID(ffn_moe_down-27): 292 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 382 us MUL_MAT_ID(ffn_moe_down-28): 210 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 458 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 456 us MUL_MAT_ID(ffn_moe_down-30): 273 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 441 us MUL_MAT_ID(ffn_moe_down-31): 233 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 393 us MUL_MAT_ID(ffn_moe_down-32): 185 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 433 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 392 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 448 us MUL_MAT_ID(ffn_moe_down-35): 400 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 400 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 392 us MUL_MAT_ID(ffn_moe_down-37): 321 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 396 us MUL_MAT_ID(ffn_moe_down-38): 241 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 455 us MUL_MAT_ID(ffn_moe_down-39): 270 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 466 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 422 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 423 us MUL_MAT_ID(ffn_moe_down-42): 201 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 399 us MUL_MAT_ID(ffn_moe_down-43): 206 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 466 us MUL_MAT_ID(ffn_moe_down-44): 200 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 421 us MUL_MAT_ID(ffn_moe_down-45): 479 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 464 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 471 us MUL_MAT_ID(ffn_moe_down-47): 294 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 407 us MUL_MAT_ID(ffn_moe_down-48): 176 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 503 us MUL_MAT_ID(ffn_moe_down-49): 229 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 406 us MUL_MAT_ID(ffn_moe_down-50): 779 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 547 us MUL_MAT_ID(ffn_moe_down-51): 199 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 419 us MUL_MAT_ID(ffn_moe_down-52): 203 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 396 us MUL_MAT_ID(ffn_moe_down-53): 425 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 418 us MUL_MAT_ID(ffn_moe_down-54): 393 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 378 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 396 us MUL_MAT_ID(ffn_moe_down-56): 191 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 373 us MUL_MAT_ID(ffn_moe_down-57): 252 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 394 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 511 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 388 us MUL_MAT_ID(ffn_moe_down-61): 199 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 352 us MUL_MAT_ID(ffn_moe_down-62): 236 us ggml_barrier(...): 367 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 365 us MUL_MAT_ID(ffn_moe_down-63): 223 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 360 us MUL_MAT_ID(ffn_moe_down-64): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 392 us MUL_MAT_ID(ffn_moe_down-65): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 387 us MUL_MAT_ID(ffn_moe_down-66): 182 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 444 us MUL_MAT_ID(ffn_moe_down-67): 261 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 414 us MUL_MAT_ID(ffn_moe_down-68): 220 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 717 us MUL_MAT_ID(ffn_moe_down-69): 197 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 430 us MUL_MAT_ID(ffn_moe_down-70): 214 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 417 us MUL_MAT_ID(ffn_moe_down-71): 282 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 179 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 459 us MUL_MAT_ID(ffn_moe_down-73): 242 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 382 us MUL_MAT_ID(ffn_moe_down-74): 214 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 398 us MUL_MAT_ID(ffn_moe_down-75): 220 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 448 us MUL_MAT_ID(ffn_moe_down-76): 171 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 396 us MUL_MAT_ID(ffn_moe_down-77): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 342 us MUL_MAT_ID(ffn_moe_down-78): 202 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 448 us MUL_MAT_ID(ffn_moe_down-79): 215 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 350 us MUL_MAT_ID(ffn_moe_down-80): 235 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 429 us MUL_MAT_ID(ffn_moe_down-81): 209 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 400 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 381 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 402 us MUL_MAT_ID(ffn_moe_down-84): 289 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 385 us MUL_MAT_ID(ffn_moe_down-85): 230 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 659 us MUL_MAT_ID(ffn_moe_down-86): 305 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 399 us MUL_MAT_ID(ffn_moe_down-87): 189 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 385 us MUL_MAT_ID(ffn_moe_down-88): 290 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 387 us MUL_MAT_ID(ffn_moe_down-89): 195 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 399 us MUL_MAT_ID(ffn_moe_down-90): 448 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 387 us MUL_MAT_ID(ffn_moe_down-91): 258 us ggml_barrier(...): 33 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 392 us MUL_MAT_ID(ffn_moe_down-25): 203 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 395 us MUL_MAT_ID(ffn_moe_down-26): 426 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 358 us MUL_MAT_ID(ffn_moe_down-27): 214 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 467 us MUL_MAT_ID(ffn_moe_down-28): 283 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 391 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 451 us MUL_MAT_ID(ffn_moe_down-30): 183 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 386 us MUL_MAT_ID(ffn_moe_down-31): 210 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 400 us MUL_MAT_ID(ffn_moe_down-32): 218 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 420 us MUL_MAT_ID(ffn_moe_down-33): 183 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 391 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 418 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 386 us MUL_MAT_ID(ffn_moe_down-36): 188 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 344 us MUL_MAT_ID(ffn_moe_down-37): 670 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 367 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 488 us MUL_MAT_ID(ffn_moe_down-39): 365 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 447 us MUL_MAT_ID(ffn_moe_down-40): 255 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 557 us MUL_MAT_ID(ffn_moe_down-41): 205 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 846 us MUL_MAT_ID(ffn_moe_down-42): 338 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 498 us MUL_MAT_ID(ffn_moe_down-43): 256 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 450 us MUL_MAT_ID(ffn_moe_down-44): 208 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 399 us MUL_MAT_ID(ffn_moe_down-45): 387 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 354 us MUL_MAT_ID(ffn_moe_down-46): 186 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 490 us MUL_MAT_ID(ffn_moe_down-47): 404 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 406 us MUL_MAT_ID(ffn_moe_down-48): 199 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 448 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 465 us MUL_MAT_ID(ffn_moe_down-50): 357 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 509 us MUL_MAT_ID(ffn_moe_down-51): 633 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 652 us MUL_MAT_ID(ffn_moe_down-52): 1334 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 946 us MUL_MAT_ID(ffn_moe_down-53): 520 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 867 us MUL_MAT_ID(ffn_moe_down-54): 587 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 350 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 547 us MUL_MAT_ID(ffn_moe_down-56): 251 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 359 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 411 us MUL_MAT_ID(ffn_moe_down-58): 252 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 387 us MUL_MAT_ID(ffn_moe_down-59): 211 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 374 us MUL_MAT_ID(ffn_moe_down-60): 206 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 495 us MUL_MAT_ID(ffn_moe_down-61): 171 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 371 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 364 us MUL_MAT_ID(ffn_moe_down-63): 285 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 377 us MUL_MAT_ID(ffn_moe_down-64): 267 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 413 us MUL_MAT_ID(ffn_moe_down-65): 229 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 394 us MUL_MAT_ID(ffn_moe_down-66): 245 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 378 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 537 us MUL_MAT_ID(ffn_moe_down-68): 459 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 391 us MUL_MAT_ID(ffn_moe_down-69): 614 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 420 us MUL_MAT_ID(ffn_moe_down-70): 209 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 443 us MUL_MAT_ID(ffn_moe_down-71): 267 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 456 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 402 us MUL_MAT_ID(ffn_moe_down-73): 206 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 395 us MUL_MAT_ID(ffn_moe_down-74): 225 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 409 us MUL_MAT_ID(ffn_moe_down-75): 288 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 478 us MUL_MAT_ID(ffn_moe_down-76): 263 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 381 us MUL_MAT_ID(ffn_moe_down-77): 185 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 440 us MUL_MAT_ID(ffn_moe_down-78): 279 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 451 us MUL_MAT_ID(ffn_moe_down-79): 187 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 527 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 389 us MUL_MAT_ID(ffn_moe_down-81): 184 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 386 us MUL_MAT_ID(ffn_moe_down-82): 226 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 501 us MUL_MAT_ID(ffn_moe_down-83): 221 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 409 us MUL_MAT_ID(ffn_moe_down-84): 254 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 436 us MUL_MAT_ID(ffn_moe_down-85): 471 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 452 us MUL_MAT_ID(ffn_moe_down-86): 220 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 518 us MUL_MAT_ID(ffn_moe_down-87): 384 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 470 us MUL_MAT_ID(ffn_moe_down-88): 207 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 473 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 385 us MUL_MAT_ID(ffn_moe_down-90): 240 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 519 us MUL_MAT_ID(ffn_moe_down-91): 266 us ggml_barrier(...): 166 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 394 us MUL_MAT_ID(ffn_moe_down-25): 180 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 410 us MUL_MAT_ID(ffn_moe_down-26): 305 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 425 us MUL_MAT_ID(ffn_moe_down-27): 206 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 384 us MUL_MAT_ID(ffn_moe_down-28): 187 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 474 us MUL_MAT_ID(ffn_moe_down-30): 236 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 409 us MUL_MAT_ID(ffn_moe_down-31): 188 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 390 us MUL_MAT_ID(ffn_moe_down-32): 190 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 355 us MUL_MAT_ID(ffn_moe_down-33): 235 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 397 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 379 us MUL_MAT_ID(ffn_moe_down-35): 228 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 437 us MUL_MAT_ID(ffn_moe_down-36): 386 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 395 us MUL_MAT_ID(ffn_moe_down-37): 666 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 348 us MUL_MAT_ID(ffn_moe_down-38): 243 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 439 us MUL_MAT_ID(ffn_moe_down-39): 241 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 405 us MUL_MAT_ID(ffn_moe_down-40): 286 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 348 us MUL_MAT_ID(ffn_moe_down-41): 452 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 447 us MUL_MAT_ID(ffn_moe_down-42): 211 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 413 us MUL_MAT_ID(ffn_moe_down-43): 417 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 377 us MUL_MAT_ID(ffn_moe_down-44): 409 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 369 us MUL_MAT_ID(ffn_moe_down-45): 218 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 354 us MUL_MAT_ID(ffn_moe_down-46): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 477 us MUL_MAT_ID(ffn_moe_down-47): 228 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 411 us MUL_MAT_ID(ffn_moe_down-48): 217 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 410 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 395 us MUL_MAT_ID(ffn_moe_down-50): 232 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 410 us MUL_MAT_ID(ffn_moe_down-51): 218 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 504 us MUL_MAT_ID(ffn_moe_down-52): 540 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 400 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 385 us MUL_MAT_ID(ffn_moe_down-54): 205 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 652 us MUL_MAT_ID(ffn_moe_down-55): 286 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 465 us MUL_MAT_ID(ffn_moe_down-56): 306 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 385 us MUL_MAT_ID(ffn_moe_down-57): 198 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 421 us MUL_MAT_ID(ffn_moe_down-58): 215 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 389 us MUL_MAT_ID(ffn_moe_down-59): 202 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 361 us MUL_MAT_ID(ffn_moe_down-60): 202 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 517 us MUL_MAT_ID(ffn_moe_down-61): 214 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 393 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 416 us MUL_MAT_ID(ffn_moe_down-63): 204 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 438 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 412 us MUL_MAT_ID(ffn_moe_down-65): 399 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 427 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 375 us MUL_MAT_ID(ffn_moe_down-67): 216 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 421 us MUL_MAT_ID(ffn_moe_down-68): 583 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 370 us MUL_MAT_ID(ffn_moe_down-69): 300 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 398 us MUL_MAT_ID(ffn_moe_down-70): 220 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 491 us MUL_MAT_ID(ffn_moe_down-71): 432 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 379 us MUL_MAT_ID(ffn_moe_down-72): 231 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 431 us MUL_MAT_ID(ffn_moe_down-73): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 396 us MUL_MAT_ID(ffn_moe_down-74): 592 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 436 us MUL_MAT_ID(ffn_moe_down-75): 222 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 431 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 426 us MUL_MAT_ID(ffn_moe_down-77): 209 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 729 us MUL_MAT_ID(ffn_moe_down-78): 344 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 396 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 380 us MUL_MAT_ID(ffn_moe_down-80): 328 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 358 us MUL_MAT_ID(ffn_moe_down-81): 209 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 410 us MUL_MAT_ID(ffn_moe_down-82): 569 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 408 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 372 us MUL_MAT_ID(ffn_moe_down-84): 250 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 405 us MUL_MAT_ID(ffn_moe_down-85): 250 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 390 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 356 us MUL_MAT_ID(ffn_moe_down-87): 212 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 395 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 396 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 687 us MUL_MAT_ID(ffn_moe_down-90): 289 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 367 us MUL_MAT_ID(ffn_moe_down-91): 245 us ggml_barrier(...): 144 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 374 us MUL_MAT_ID(ffn_moe_down-25): 309 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 438 us MUL_MAT_ID(ffn_moe_down-26): 221 us ggml_barrier(...): 356 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 477 us MUL_MAT_ID(ffn_moe_down-27): 203 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 403 us MUL_MAT_ID(ffn_moe_down-28): 264 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 412 us MUL_MAT_ID(ffn_moe_down-29): 201 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 438 us MUL_MAT_ID(ffn_moe_down-30): 176 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 407 us MUL_MAT_ID(ffn_moe_down-31): 274 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 402 us MUL_MAT_ID(ffn_moe_down-32): 236 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 247 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 401 us MUL_MAT_ID(ffn_moe_down-34): 273 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 435 us MUL_MAT_ID(ffn_moe_down-35): 287 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 410 us MUL_MAT_ID(ffn_moe_down-36): 540 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 415 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 411 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 488 us MUL_MAT_ID(ffn_moe_down-39): 273 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 401 us MUL_MAT_ID(ffn_moe_down-40): 219 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 416 us MUL_MAT_ID(ffn_moe_down-41): 202 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 578 us MUL_MAT_ID(ffn_moe_down-42): 258 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 412 us MUL_MAT_ID(ffn_moe_down-43): 250 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 395 us MUL_MAT_ID(ffn_moe_down-44): 206 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 498 us MUL_MAT_ID(ffn_moe_down-45): 418 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 359 us MUL_MAT_ID(ffn_moe_down-46): 213 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 380 us MUL_MAT_ID(ffn_moe_down-47): 202 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 515 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 374 us MUL_MAT_ID(ffn_moe_down-49): 197 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 400 us MUL_MAT_ID(ffn_moe_down-50): 219 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 442 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 440 us MUL_MAT_ID(ffn_moe_down-52): 182 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 398 us MUL_MAT_ID(ffn_moe_down-53): 204 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 449 us MUL_MAT_ID(ffn_moe_down-54): 224 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 397 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 421 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 477 us MUL_MAT_ID(ffn_moe_down-57): 210 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 373 us MUL_MAT_ID(ffn_moe_down-58): 182 us ggml_barrier(...): 343 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 363 us MUL_MAT_ID(ffn_moe_down-59): 188 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 372 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 374 us MUL_MAT_ID(ffn_moe_down-61): 188 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 388 us MUL_MAT_ID(ffn_moe_down-62): 203 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 433 us MUL_MAT_ID(ffn_moe_down-63): 649 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 349 us MUL_MAT_ID(ffn_moe_down-64): 260 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 410 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 384 us MUL_MAT_ID(ffn_moe_down-66): 184 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 375 us MUL_MAT_ID(ffn_moe_down-67): 218 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 434 us MUL_MAT_ID(ffn_moe_down-68): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 381 us MUL_MAT_ID(ffn_moe_down-69): 206 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 415 us MUL_MAT_ID(ffn_moe_down-70): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 385 us MUL_MAT_ID(ffn_moe_down-71): 187 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 402 us MUL_MAT_ID(ffn_moe_down-72): 188 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 367 us MUL_MAT_ID(ffn_moe_down-73): 248 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 431 us MUL_MAT_ID(ffn_moe_down-74): 253 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 404 us MUL_MAT_ID(ffn_moe_down-75): 205 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 476 us MUL_MAT_ID(ffn_moe_down-76): 184 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 392 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 457 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 399 us MUL_MAT_ID(ffn_moe_down-80): 189 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 427 us MUL_MAT_ID(ffn_moe_down-81): 245 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 397 us MUL_MAT_ID(ffn_moe_down-82): 189 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 379 us MUL_MAT_ID(ffn_moe_down-83): 501 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 429 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 409 us MUL_MAT_ID(ffn_moe_down-85): 227 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 430 us MUL_MAT_ID(ffn_moe_down-86): 208 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 744 us MUL_MAT_ID(ffn_moe_down-87): 329 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 357 us MUL_MAT_ID(ffn_moe_down-88): 217 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 429 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 375 us MUL_MAT_ID(ffn_moe_down-90): 222 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 393 us MUL_MAT_ID(ffn_moe_down-91): 265 us ggml_barrier(...): 42 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 575 us MUL_MAT_ID(ffn_moe_down-25): 283 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 468 us MUL_MAT_ID(ffn_moe_down-26): 288 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 448 us MUL_MAT_ID(ffn_moe_down-27): 176 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 411 us MUL_MAT_ID(ffn_moe_down-28): 423 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 410 us MUL_MAT_ID(ffn_moe_down-29): 266 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 422 us MUL_MAT_ID(ffn_moe_down-30): 372 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 405 us MUL_MAT_ID(ffn_moe_down-31): 237 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 505 us MUL_MAT_ID(ffn_moe_down-32): 223 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 361 us MUL_MAT_ID(ffn_moe_down-33): 204 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 664 us MUL_MAT_ID(ffn_moe_down-34): 221 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 448 us MUL_MAT_ID(ffn_moe_down-35): 195 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 418 us MUL_MAT_ID(ffn_moe_down-36): 229 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 432 us MUL_MAT_ID(ffn_moe_down-37): 300 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 392 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 427 us MUL_MAT_ID(ffn_moe_down-39): 260 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 448 us MUL_MAT_ID(ffn_moe_down-40): 267 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 433 us MUL_MAT_ID(ffn_moe_down-41): 199 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 475 us MUL_MAT_ID(ffn_moe_down-42): 368 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 503 us MUL_MAT_ID(ffn_moe_down-43): 232 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 392 us MUL_MAT_ID(ffn_moe_down-44): 201 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 396 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 432 us MUL_MAT_ID(ffn_moe_down-46): 203 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 392 us MUL_MAT_ID(ffn_moe_down-47): 212 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 362 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 520 us MUL_MAT_ID(ffn_moe_down-49): 233 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 476 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 412 us MUL_MAT_ID(ffn_moe_down-51): 392 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 507 us MUL_MAT_ID(ffn_moe_down-52): 231 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 364 us MUL_MAT_ID(ffn_moe_down-53): 191 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 390 us MUL_MAT_ID(ffn_moe_down-54): 209 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 409 us MUL_MAT_ID(ffn_moe_down-55): 184 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 462 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 366 us MUL_MAT_ID(ffn_moe_down-57): 179 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 391 us MUL_MAT_ID(ffn_moe_down-58): 210 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 430 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 501 us MUL_MAT_ID(ffn_moe_down-60): 445 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 486 us MUL_MAT_ID(ffn_moe_down-61): 210 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 461 us MUL_MAT_ID(ffn_moe_down-62): 206 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 444 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 426 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 367 us MUL_MAT_ID(ffn_moe_down-65): 469 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 423 us MUL_MAT_ID(ffn_moe_down-66): 214 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 474 us MUL_MAT_ID(ffn_moe_down-67): 355 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 449 us MUL_MAT_ID(ffn_moe_down-68): 234 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 397 us MUL_MAT_ID(ffn_moe_down-69): 186 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 399 us MUL_MAT_ID(ffn_moe_down-70): 197 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 358 us MUL_MAT_ID(ffn_moe_down-71): 232 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 377 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 411 us MUL_MAT_ID(ffn_moe_down-73): 338 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 372 us MUL_MAT_ID(ffn_moe_down-74): 177 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 447 us MUL_MAT_ID(ffn_moe_down-75): 247 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 422 us MUL_MAT_ID(ffn_moe_down-76): 183 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 491 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 346 us MUL_MAT_ID(ffn_moe_down-78): 206 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 358 us MUL_MAT_ID(ffn_moe_down-79): 178 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 391 us MUL_MAT_ID(ffn_moe_down-80): 562 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 391 us MUL_MAT_ID(ffn_moe_down-81): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 391 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 360 us MUL_MAT_ID(ffn_moe_down-83): 190 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 414 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 437 us MUL_MAT_ID(ffn_moe_down-85): 225 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 373 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 431 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 516 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 414 us MUL_MAT_ID(ffn_moe_down-89): 187 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 392 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 357 us MUL_MAT_ID(ffn_moe_down-91): 784 us ggml_barrier(...): 96 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 343 us ggml_barrier(...): 562 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 439 us MUL_MAT_ID(ffn_moe_down-26): 328 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 410 us MUL_MAT_ID(ffn_moe_down-27): 337 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 500 us MUL_MAT_ID(ffn_moe_down-28): 212 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 453 us MUL_MAT_ID(ffn_moe_down-29): 201 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 504 us MUL_MAT_ID(ffn_moe_down-30): 166 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 442 us MUL_MAT_ID(ffn_moe_down-31): 262 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 410 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 398 us MUL_MAT_ID(ffn_moe_down-33): 184 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 443 us MUL_MAT_ID(ffn_moe_down-34): 162 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 650 us MUL_MAT_ID(ffn_moe_down-35): 264 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 513 us MUL_MAT_ID(ffn_moe_down-36): 175 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 369 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 401 us MUL_MAT_ID(ffn_moe_down-38): 191 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 368 us MUL_MAT_ID(ffn_moe_down-39): 584 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 388 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 513 us MUL_MAT_ID(ffn_moe_down-41): 559 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 403 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 407 us MUL_MAT_ID(ffn_moe_down-43): 210 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 367 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 336 us MUL_MAT_ID(ffn_moe_down-45): 215 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 385 us MUL_MAT_ID(ffn_moe_down-46): 186 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 384 us MUL_MAT_ID(ffn_moe_down-47): 210 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 441 us MUL_MAT_ID(ffn_moe_down-48): 181 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 433 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 444 us MUL_MAT_ID(ffn_moe_down-50): 208 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 406 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 446 us MUL_MAT_ID(ffn_moe_down-52): 215 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 387 us MUL_MAT_ID(ffn_moe_down-53): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 448 us MUL_MAT_ID(ffn_moe_down-54): 190 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 469 us MUL_MAT_ID(ffn_moe_down-55): 380 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 414 us MUL_MAT_ID(ffn_moe_down-56): 276 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 398 us MUL_MAT_ID(ffn_moe_down-57): 204 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 390 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 424 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 400 us MUL_MAT_ID(ffn_moe_down-60): 205 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 392 us MUL_MAT_ID(ffn_moe_down-61): 256 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 399 us MUL_MAT_ID(ffn_moe_down-62): 300 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 383 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 380 us MUL_MAT_ID(ffn_moe_down-64): 248 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 401 us MUL_MAT_ID(ffn_moe_down-65): 233 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 423 us MUL_MAT_ID(ffn_moe_down-66): 187 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 444 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 404 us MUL_MAT_ID(ffn_moe_down-68): 298 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 393 us MUL_MAT_ID(ffn_moe_down-69): 540 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 386 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 401 us MUL_MAT_ID(ffn_moe_down-71): 213 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 395 us MUL_MAT_ID(ffn_moe_down-72): 212 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 396 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 413 us MUL_MAT_ID(ffn_moe_down-74): 209 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 443 us MUL_MAT_ID(ffn_moe_down-75): 211 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 454 us MUL_MAT_ID(ffn_moe_down-76): 216 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 186 us ggml_barrier(...): 441 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 392 us MUL_MAT_ID(ffn_moe_down-78): 190 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 402 us MUL_MAT_ID(ffn_moe_down-79): 178 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 467 us MUL_MAT_ID(ffn_moe_down-80): 200 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 427 us MUL_MAT_ID(ffn_moe_down-81): 209 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 491 us MUL_MAT_ID(ffn_moe_down-82): 197 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 457 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 401 us MUL_MAT_ID(ffn_moe_down-84): 244 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 450 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 407 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 410 us MUL_MAT_ID(ffn_moe_down-87): 361 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 402 us MUL_MAT_ID(ffn_moe_down-88): 199 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 421 us MUL_MAT_ID(ffn_moe_down-89): 177 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 404 us MUL_MAT_ID(ffn_moe_down-90): 609 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 370 us MUL_MAT_ID(ffn_moe_down-91): 742 us ggml_barrier(...): 61 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 472 us MUL_MAT_ID(ffn_moe_down-25): 524 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 449 us MUL_MAT_ID(ffn_moe_down-26): 345 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 433 us MUL_MAT_ID(ffn_moe_down-27): 267 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 518 us MUL_MAT_ID(ffn_moe_down-28): 307 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 458 us MUL_MAT_ID(ffn_moe_down-29): 258 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 433 us MUL_MAT_ID(ffn_moe_down-30): 339 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 413 us MUL_MAT_ID(ffn_moe_down-31): 214 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 411 us MUL_MAT_ID(ffn_moe_down-32): 254 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 413 us MUL_MAT_ID(ffn_moe_down-33): 222 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 406 us MUL_MAT_ID(ffn_moe_down-34): 216 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 536 us MUL_MAT_ID(ffn_moe_down-35): 212 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 400 us MUL_MAT_ID(ffn_moe_down-36): 508 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 397 us MUL_MAT_ID(ffn_moe_down-37): 241 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 416 us MUL_MAT_ID(ffn_moe_down-38): 374 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 343 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 385 us MUL_MAT_ID(ffn_moe_down-40): 192 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 436 us MUL_MAT_ID(ffn_moe_down-41): 394 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 497 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 423 us MUL_MAT_ID(ffn_moe_down-43): 204 us ggml_barrier(...): 343 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 401 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 402 us MUL_MAT_ID(ffn_moe_down-45): 220 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 417 us MUL_MAT_ID(ffn_moe_down-46): 211 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 427 us MUL_MAT_ID(ffn_moe_down-47): 198 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 395 us MUL_MAT_ID(ffn_moe_down-48): 203 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 400 us MUL_MAT_ID(ffn_moe_down-49): 382 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 381 us MUL_MAT_ID(ffn_moe_down-50): 210 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 424 us MUL_MAT_ID(ffn_moe_down-51): 202 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 396 us MUL_MAT_ID(ffn_moe_down-52): 181 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 407 us MUL_MAT_ID(ffn_moe_down-53): 247 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 425 us MUL_MAT_ID(ffn_moe_down-54): 210 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 406 us MUL_MAT_ID(ffn_moe_down-55): 217 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 393 us MUL_MAT_ID(ffn_moe_down-57): 177 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 413 us MUL_MAT_ID(ffn_moe_down-58): 202 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 439 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 399 us MUL_MAT_ID(ffn_moe_down-60): 202 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 474 us MUL_MAT_ID(ffn_moe_down-61): 178 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 412 us MUL_MAT_ID(ffn_moe_down-62): 279 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 536 us MUL_MAT_ID(ffn_moe_down-63): 214 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 383 us MUL_MAT_ID(ffn_moe_down-64): 228 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 385 us MUL_MAT_ID(ffn_moe_down-65): 223 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 404 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 337 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 186 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 396 us MUL_MAT_ID(ffn_moe_down-69): 429 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 492 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 644 us MUL_MAT_ID(ffn_moe_down-71): 227 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 373 us MUL_MAT_ID(ffn_moe_down-72): 255 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 372 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 362 us MUL_MAT_ID(ffn_moe_down-74): 200 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 422 us MUL_MAT_ID(ffn_moe_down-75): 217 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 457 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 363 us MUL_MAT_ID(ffn_moe_down-77): 210 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 493 us MUL_MAT_ID(ffn_moe_down-78): 319 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 420 us MUL_MAT_ID(ffn_moe_down-79): 337 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 374 us MUL_MAT_ID(ffn_moe_down-80): 289 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 470 us MUL_MAT_ID(ffn_moe_down-81): 280 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 416 us MUL_MAT_ID(ffn_moe_down-82): 184 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 386 us MUL_MAT_ID(ffn_moe_down-83): 186 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 438 us MUL_MAT_ID(ffn_moe_down-84): 310 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 384 us MUL_MAT_ID(ffn_moe_down-85): 252 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 404 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 400 us MUL_MAT_ID(ffn_moe_down-87): 223 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 390 us MUL_MAT_ID(ffn_moe_down-88): 188 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 421 us MUL_MAT_ID(ffn_moe_down-89): 414 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 376 us MUL_MAT_ID(ffn_moe_down-90): 692 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 349 us MUL_MAT_ID(ffn_moe_down-91): 257 us ggml_barrier(...): 50 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 381 us MUL_MAT_ID(ffn_moe_down-25): 185 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 410 us MUL_MAT_ID(ffn_moe_down-26): 203 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 385 us MUL_MAT_ID(ffn_moe_down-27): 214 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 387 us MUL_MAT_ID(ffn_moe_down-28): 195 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 387 us MUL_MAT_ID(ffn_moe_down-29): 186 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 360 us MUL_MAT_ID(ffn_moe_down-30): 184 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 406 us MUL_MAT_ID(ffn_moe_down-31): 259 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 432 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 412 us MUL_MAT_ID(ffn_moe_down-33): 189 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 417 us MUL_MAT_ID(ffn_moe_down-34): 224 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 372 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 387 us MUL_MAT_ID(ffn_moe_down-36): 207 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 367 us MUL_MAT_ID(ffn_moe_down-37): 241 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 404 us MUL_MAT_ID(ffn_moe_down-38): 557 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 424 us MUL_MAT_ID(ffn_moe_down-39): 748 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 343 us MUL_MAT_ID(ffn_moe_down-40): 669 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 385 us MUL_MAT_ID(ffn_moe_down-41): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 345 us MUL_MAT_ID(ffn_moe_down-42): 278 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 466 us MUL_MAT_ID(ffn_moe_down-43): 460 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 351 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 417 us MUL_MAT_ID(ffn_moe_down-45): 202 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 419 us MUL_MAT_ID(ffn_moe_down-46): 411 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 665 us MUL_MAT_ID(ffn_moe_down-47): 271 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 419 us MUL_MAT_ID(ffn_moe_down-48): 233 us ggml_barrier(...): 346 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 409 us MUL_MAT_ID(ffn_moe_down-49): 180 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 421 us MUL_MAT_ID(ffn_moe_down-50): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 418 us MUL_MAT_ID(ffn_moe_down-51): 256 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 446 us MUL_MAT_ID(ffn_moe_down-52): 255 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 417 us MUL_MAT_ID(ffn_moe_down-53): 178 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 364 us MUL_MAT_ID(ffn_moe_down-54): 205 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 282 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 416 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 396 us MUL_MAT_ID(ffn_moe_down-57): 178 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 411 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 414 us MUL_MAT_ID(ffn_moe_down-59): 206 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 377 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 372 us MUL_MAT_ID(ffn_moe_down-61): 201 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 402 us MUL_MAT_ID(ffn_moe_down-62): 207 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 398 us MUL_MAT_ID(ffn_moe_down-63): 205 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 367 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 393 us MUL_MAT_ID(ffn_moe_down-65): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 393 us MUL_MAT_ID(ffn_moe_down-66): 175 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 425 us MUL_MAT_ID(ffn_moe_down-67): 225 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 404 us MUL_MAT_ID(ffn_moe_down-68): 211 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 394 us MUL_MAT_ID(ffn_moe_down-69): 188 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 467 us MUL_MAT_ID(ffn_moe_down-70): 295 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 394 us MUL_MAT_ID(ffn_moe_down-71): 199 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 383 us MUL_MAT_ID(ffn_moe_down-72): 278 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 448 us MUL_MAT_ID(ffn_moe_down-73): 263 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 439 us MUL_MAT_ID(ffn_moe_down-74): 305 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 417 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 411 us MUL_MAT_ID(ffn_moe_down-76): 388 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 608 us MUL_MAT_ID(ffn_moe_down-77): 304 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 439 us MUL_MAT_ID(ffn_moe_down-78): 270 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 393 us MUL_MAT_ID(ffn_moe_down-79): 209 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 372 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 440 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 402 us MUL_MAT_ID(ffn_moe_down-82): 207 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 343 us MUL_MAT_ID(ffn_moe_down-83): 218 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 514 us MUL_MAT_ID(ffn_moe_down-84): 223 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 417 us MUL_MAT_ID(ffn_moe_down-85): 418 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 381 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 418 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 425 us MUL_MAT_ID(ffn_moe_down-88): 371 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 418 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 398 us MUL_MAT_ID(ffn_moe_down-90): 534 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 385 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 5 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 202 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 468 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 419 us MUL_MAT_ID(ffn_moe_down-27): 331 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 513 us MUL_MAT_ID(ffn_moe_down-28): 387 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 771 us MUL_MAT_ID(ffn_moe_down-29): 330 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 362 us MUL_MAT_ID(ffn_moe_down-30): 168 us ggml_barrier(...): 552 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 520 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 382 us MUL_MAT_ID(ffn_moe_down-32): 178 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 417 us MUL_MAT_ID(ffn_moe_down-33): 223 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 382 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 375 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 392 us MUL_MAT_ID(ffn_moe_down-36): 190 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 345 us MUL_MAT_ID(ffn_moe_down-37): 219 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 421 us MUL_MAT_ID(ffn_moe_down-38): 219 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 441 us MUL_MAT_ID(ffn_moe_down-39): 315 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 364 us MUL_MAT_ID(ffn_moe_down-40): 217 us ggml_barrier(...): 353 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 392 us MUL_MAT_ID(ffn_moe_down-41): 188 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 336 us MUL_MAT_ID(ffn_moe_down-42): 244 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 504 us MUL_MAT_ID(ffn_moe_down-43): 164 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 393 us MUL_MAT_ID(ffn_moe_down-44): 184 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 436 us MUL_MAT_ID(ffn_moe_down-45): 248 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 372 us MUL_MAT_ID(ffn_moe_down-46): 195 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 381 us MUL_MAT_ID(ffn_moe_down-47): 212 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 475 us MUL_MAT_ID(ffn_moe_down-48): 256 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 353 us MUL_MAT_ID(ffn_moe_down-49): 382 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 387 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 347 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 388 us MUL_MAT_ID(ffn_moe_down-52): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 359 us MUL_MAT_ID(ffn_moe_down-53): 213 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 390 us MUL_MAT_ID(ffn_moe_down-54): 179 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 424 us MUL_MAT_ID(ffn_moe_down-55): 443 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 382 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 357 us MUL_MAT_ID(ffn_moe_down-57): 206 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 414 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 351 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 381 us MUL_MAT_ID(ffn_moe_down-60): 385 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 392 us MUL_MAT_ID(ffn_moe_down-61): 203 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 395 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 342 us MUL_MAT_ID(ffn_moe_down-63): 195 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 347 us MUL_MAT_ID(ffn_moe_down-64): 210 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 404 us MUL_MAT_ID(ffn_moe_down-65): 344 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 427 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 422 us MUL_MAT_ID(ffn_moe_down-67): 212 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 410 us MUL_MAT_ID(ffn_moe_down-68): 188 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 398 us MUL_MAT_ID(ffn_moe_down-69): 205 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 447 us MUL_MAT_ID(ffn_moe_down-70): 212 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 423 us MUL_MAT_ID(ffn_moe_down-71): 177 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 400 us MUL_MAT_ID(ffn_moe_down-72): 207 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 432 us MUL_MAT_ID(ffn_moe_down-73): 521 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 405 us MUL_MAT_ID(ffn_moe_down-74): 208 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 428 us MUL_MAT_ID(ffn_moe_down-75): 188 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 390 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 425 us MUL_MAT_ID(ffn_moe_down-77): 600 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 377 us MUL_MAT_ID(ffn_moe_down-78): 210 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 391 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 336 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 375 us MUL_MAT_ID(ffn_moe_down-81): 186 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 335 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 444 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 413 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 415 us MUL_MAT_ID(ffn_moe_down-85): 663 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 439 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 423 us MUL_MAT_ID(ffn_moe_down-87): 556 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 426 us MUL_MAT_ID(ffn_moe_down-88): 184 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 445 us MUL_MAT_ID(ffn_moe_down-89): 295 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 407 us MUL_MAT_ID(ffn_moe_down-90): 221 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 425 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 28 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 421 us MUL_MAT_ID(ffn_moe_down-25): 410 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 467 us MUL_MAT_ID(ffn_moe_down-26): 306 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 416 us MUL_MAT_ID(ffn_moe_down-27): 209 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 451 us MUL_MAT_ID(ffn_moe_down-28): 235 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 421 us MUL_MAT_ID(ffn_moe_down-29): 216 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 427 us MUL_MAT_ID(ffn_moe_down-30): 177 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 420 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 471 us MUL_MAT_ID(ffn_moe_down-32): 230 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 411 us MUL_MAT_ID(ffn_moe_down-33): 188 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 417 us MUL_MAT_ID(ffn_moe_down-34): 203 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 414 us MUL_MAT_ID(ffn_moe_down-35): 178 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 440 us MUL_MAT_ID(ffn_moe_down-36): 156 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 409 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 461 us MUL_MAT_ID(ffn_moe_down-38): 231 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 392 us MUL_MAT_ID(ffn_moe_down-39): 277 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 443 us MUL_MAT_ID(ffn_moe_down-40): 322 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 686 us MUL_MAT_ID(ffn_moe_down-41): 276 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 389 us MUL_MAT_ID(ffn_moe_down-42): 212 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 358 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 400 us MUL_MAT_ID(ffn_moe_down-44): 185 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 439 us MUL_MAT_ID(ffn_moe_down-45): 217 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 551 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 467 us MUL_MAT_ID(ffn_moe_down-47): 312 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 422 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 479 us MUL_MAT_ID(ffn_moe_down-49): 256 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 374 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 479 us MUL_MAT_ID(ffn_moe_down-51): 223 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 379 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 438 us MUL_MAT_ID(ffn_moe_down-53): 357 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 391 us MUL_MAT_ID(ffn_moe_down-54): 201 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 386 us MUL_MAT_ID(ffn_moe_down-55): 203 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 429 us MUL_MAT_ID(ffn_moe_down-56): 223 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 417 us MUL_MAT_ID(ffn_moe_down-57): 433 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 377 us MUL_MAT_ID(ffn_moe_down-58): 206 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 505 us MUL_MAT_ID(ffn_moe_down-59): 249 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 496 us MUL_MAT_ID(ffn_moe_down-60): 225 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 495 us MUL_MAT_ID(ffn_moe_down-61): 199 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 404 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 357 us MUL_MAT_ID(ffn_moe_down-63): 214 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 356 us MUL_MAT_ID(ffn_moe_down-64): 201 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 372 us MUL_MAT_ID(ffn_moe_down-65): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 412 us MUL_MAT_ID(ffn_moe_down-66): 185 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 470 us MUL_MAT_ID(ffn_moe_down-67): 245 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 409 us MUL_MAT_ID(ffn_moe_down-68): 206 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 350 us MUL_MAT_ID(ffn_moe_down-69): 194 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 472 us MUL_MAT_ID(ffn_moe_down-70): 209 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 364 us MUL_MAT_ID(ffn_moe_down-71): 201 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 428 us MUL_MAT_ID(ffn_moe_down-72): 291 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 392 us MUL_MAT_ID(ffn_moe_down-73): 196 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 353 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 406 us MUL_MAT_ID(ffn_moe_down-75): 203 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 352 us MUL_MAT_ID(ffn_moe_down-76): 386 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 469 us MUL_MAT_ID(ffn_moe_down-77): 189 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 405 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 380 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 384 us MUL_MAT_ID(ffn_moe_down-80): 184 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 438 us MUL_MAT_ID(ffn_moe_down-81): 288 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 404 us MUL_MAT_ID(ffn_moe_down-82): 316 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 371 us MUL_MAT_ID(ffn_moe_down-83): 189 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 419 us MUL_MAT_ID(ffn_moe_down-84): 384 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 418 us MUL_MAT_ID(ffn_moe_down-85): 242 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 404 us MUL_MAT_ID(ffn_moe_down-86): 204 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 384 us MUL_MAT_ID(ffn_moe_down-87): 210 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 458 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 387 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 603 us MUL_MAT_ID(ffn_moe_down-90): 253 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 407 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 47 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 442 us MUL_MAT_ID(ffn_moe_down-25): 266 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 377 us MUL_MAT_ID(ffn_moe_down-26): 254 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 426 us MUL_MAT_ID(ffn_moe_down-27): 220 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 450 us MUL_MAT_ID(ffn_moe_down-28): 209 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 459 us MUL_MAT_ID(ffn_moe_down-29): 317 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 384 us MUL_MAT_ID(ffn_moe_down-30): 198 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 411 us MUL_MAT_ID(ffn_moe_down-31): 215 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 442 us MUL_MAT_ID(ffn_moe_down-32): 190 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 443 us MUL_MAT_ID(ffn_moe_down-33): 229 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 428 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 513 us MUL_MAT_ID(ffn_moe_down-35): 232 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 384 us MUL_MAT_ID(ffn_moe_down-36): 220 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 364 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 473 us MUL_MAT_ID(ffn_moe_down-38): 211 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 433 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 429 us MUL_MAT_ID(ffn_moe_down-40): 234 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 461 us MUL_MAT_ID(ffn_moe_down-41): 325 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 388 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 381 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 418 us MUL_MAT_ID(ffn_moe_down-43): 236 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 447 us MUL_MAT_ID(ffn_moe_down-44): 251 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 428 us MUL_MAT_ID(ffn_moe_down-45): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 517 us MUL_MAT_ID(ffn_moe_down-46): 184 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 373 us MUL_MAT_ID(ffn_moe_down-47): 263 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 486 us MUL_MAT_ID(ffn_moe_down-48): 241 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 457 us MUL_MAT_ID(ffn_moe_down-49): 423 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 446 us MUL_MAT_ID(ffn_moe_down-50): 277 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 399 us MUL_MAT_ID(ffn_moe_down-51): 313 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 397 us MUL_MAT_ID(ffn_moe_down-52): 408 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 423 us MUL_MAT_ID(ffn_moe_down-53): 216 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 382 us MUL_MAT_ID(ffn_moe_down-54): 225 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 411 us MUL_MAT_ID(ffn_moe_down-55): 241 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 369 us MUL_MAT_ID(ffn_moe_down-56): 204 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 417 us MUL_MAT_ID(ffn_moe_down-57): 277 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 386 us MUL_MAT_ID(ffn_moe_down-58): 185 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 573 us MUL_MAT_ID(ffn_moe_down-59): 252 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 416 us MUL_MAT_ID(ffn_moe_down-60): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 397 us MUL_MAT_ID(ffn_moe_down-61): 318 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 399 us MUL_MAT_ID(ffn_moe_down-62): 303 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 372 us MUL_MAT_ID(ffn_moe_down-63): 184 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 336 us MUL_MAT_ID(ffn_moe_down-64): 184 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 492 us MUL_MAT_ID(ffn_moe_down-65): 176 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 421 us MUL_MAT_ID(ffn_moe_down-66): 212 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 398 us MUL_MAT_ID(ffn_moe_down-67): 196 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 388 us MUL_MAT_ID(ffn_moe_down-68): 618 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 406 us MUL_MAT_ID(ffn_moe_down-69): 260 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 403 us MUL_MAT_ID(ffn_moe_down-70): 171 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 392 us MUL_MAT_ID(ffn_moe_down-71): 207 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 423 us MUL_MAT_ID(ffn_moe_down-72): 204 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 354 us MUL_MAT_ID(ffn_moe_down-73): 228 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 382 us MUL_MAT_ID(ffn_moe_down-74): 235 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 403 us MUL_MAT_ID(ffn_moe_down-75): 208 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 416 us MUL_MAT_ID(ffn_moe_down-76): 532 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 536 us MUL_MAT_ID(ffn_moe_down-77): 204 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 403 us MUL_MAT_ID(ffn_moe_down-78): 209 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 471 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 418 us MUL_MAT_ID(ffn_moe_down-80): 210 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 483 us MUL_MAT_ID(ffn_moe_down-81): 171 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 399 us MUL_MAT_ID(ffn_moe_down-82): 187 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 436 us MUL_MAT_ID(ffn_moe_down-83): 236 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 412 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 407 us MUL_MAT_ID(ffn_moe_down-85): 429 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 366 us MUL_MAT_ID(ffn_moe_down-86): 184 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 424 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 405 us MUL_MAT_ID(ffn_moe_down-88): 220 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 341 us MUL_MAT_ID(ffn_moe_down-89): 191 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 470 us MUL_MAT_ID(ffn_moe_down-90): 259 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 379 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 47 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 421 us MUL_MAT_ID(ffn_moe_down-25): 177 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 372 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 345 us MUL_MAT_ID(ffn_moe_down-27): 422 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 427 us MUL_MAT_ID(ffn_moe_down-28): 187 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 492 us MUL_MAT_ID(ffn_moe_down-29): 214 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 400 us MUL_MAT_ID(ffn_moe_down-30): 265 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 411 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 338 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 437 us MUL_MAT_ID(ffn_moe_down-33): 215 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 382 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 397 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 459 us MUL_MAT_ID(ffn_moe_down-36): 203 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 469 us MUL_MAT_ID(ffn_moe_down-37): 212 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 456 us MUL_MAT_ID(ffn_moe_down-38): 174 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 403 us MUL_MAT_ID(ffn_moe_down-39): 242 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 438 us MUL_MAT_ID(ffn_moe_down-40): 660 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 417 us MUL_MAT_ID(ffn_moe_down-41): 310 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 465 us MUL_MAT_ID(ffn_moe_down-42): 252 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 394 us MUL_MAT_ID(ffn_moe_down-43): 413 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 366 us MUL_MAT_ID(ffn_moe_down-44): 198 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 402 us MUL_MAT_ID(ffn_moe_down-45): 215 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 402 us MUL_MAT_ID(ffn_moe_down-46): 207 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 436 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 396 us MUL_MAT_ID(ffn_moe_down-48): 214 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 391 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 417 us MUL_MAT_ID(ffn_moe_down-50): 227 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 392 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 401 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 373 us MUL_MAT_ID(ffn_moe_down-53): 190 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 413 us MUL_MAT_ID(ffn_moe_down-54): 435 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 366 us MUL_MAT_ID(ffn_moe_down-55): 212 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 407 us MUL_MAT_ID(ffn_moe_down-56): 223 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 414 us MUL_MAT_ID(ffn_moe_down-57): 210 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 638 us MUL_MAT_ID(ffn_moe_down-58): 343 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 561 us MUL_MAT_ID(ffn_moe_down-59): 331 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 469 us MUL_MAT_ID(ffn_moe_down-60): 230 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 435 us MUL_MAT_ID(ffn_moe_down-61): 234 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 400 us MUL_MAT_ID(ffn_moe_down-62): 183 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 387 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 442 us MUL_MAT_ID(ffn_moe_down-64): 218 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 389 us MUL_MAT_ID(ffn_moe_down-65): 213 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 440 us MUL_MAT_ID(ffn_moe_down-66): 229 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 439 us MUL_MAT_ID(ffn_moe_down-67): 196 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 491 us MUL_MAT_ID(ffn_moe_down-68): 755 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 413 us MUL_MAT_ID(ffn_moe_down-69): 200 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 396 us MUL_MAT_ID(ffn_moe_down-70): 397 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 481 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 411 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 340 us MUL_MAT_ID(ffn_moe_down-73): 203 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 422 us MUL_MAT_ID(ffn_moe_down-74): 225 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 434 us MUL_MAT_ID(ffn_moe_down-75): 237 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 422 us MUL_MAT_ID(ffn_moe_down-76): 401 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 385 us MUL_MAT_ID(ffn_moe_down-77): 210 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 366 us MUL_MAT_ID(ffn_moe_down-78): 202 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 439 us MUL_MAT_ID(ffn_moe_down-79): 197 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 387 us MUL_MAT_ID(ffn_moe_down-80): 225 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 417 us MUL_MAT_ID(ffn_moe_down-81): 208 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 579 us MUL_MAT_ID(ffn_moe_down-82): 337 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 372 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 396 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 513 us MUL_MAT_ID(ffn_moe_down-85): 326 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 346 us MUL_MAT_ID(ffn_moe_down-86): 183 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 425 us MUL_MAT_ID(ffn_moe_down-87): 204 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 391 us MUL_MAT_ID(ffn_moe_down-88): 213 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 394 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 456 us MUL_MAT_ID(ffn_moe_down-90): 253 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 395 us MUL_MAT_ID(ffn_moe_down-91): 233 us ggml_barrier(...): 32 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 469 us MUL_MAT_ID(ffn_moe_down-25): 280 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 437 us MUL_MAT_ID(ffn_moe_down-26): 264 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 478 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 395 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 549 us MUL_MAT_ID(ffn_moe_down-29): 344 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 347 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 425 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 362 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 493 us MUL_MAT_ID(ffn_moe_down-33): 264 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 423 us MUL_MAT_ID(ffn_moe_down-34): 203 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 388 us MUL_MAT_ID(ffn_moe_down-35): 253 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 426 us MUL_MAT_ID(ffn_moe_down-36): 596 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 462 us MUL_MAT_ID(ffn_moe_down-37): 296 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 480 us MUL_MAT_ID(ffn_moe_down-38): 177 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 457 us MUL_MAT_ID(ffn_moe_down-39): 241 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 365 us MUL_MAT_ID(ffn_moe_down-40): 234 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 449 us MUL_MAT_ID(ffn_moe_down-41): 254 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 387 us MUL_MAT_ID(ffn_moe_down-42): 263 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 414 us MUL_MAT_ID(ffn_moe_down-43): 311 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 406 us MUL_MAT_ID(ffn_moe_down-44): 252 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 409 us MUL_MAT_ID(ffn_moe_down-45): 194 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 328 us MUL_MAT_ID(ffn_moe_down-46): 245 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 416 us MUL_MAT_ID(ffn_moe_down-47): 341 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 402 us MUL_MAT_ID(ffn_moe_down-48): 203 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 412 us MUL_MAT_ID(ffn_moe_down-49): 203 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 409 us MUL_MAT_ID(ffn_moe_down-50): 198 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 376 us MUL_MAT_ID(ffn_moe_down-51): 180 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 398 us MUL_MAT_ID(ffn_moe_down-52): 273 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 365 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 415 us MUL_MAT_ID(ffn_moe_down-54): 429 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 387 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 435 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 379 us MUL_MAT_ID(ffn_moe_down-57): 484 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 536 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 421 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 460 us MUL_MAT_ID(ffn_moe_down-60): 244 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 405 us MUL_MAT_ID(ffn_moe_down-61): 247 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 371 us MUL_MAT_ID(ffn_moe_down-62): 283 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 438 us MUL_MAT_ID(ffn_moe_down-63): 302 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 642 us MUL_MAT_ID(ffn_moe_down-64): 210 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 410 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 432 us MUL_MAT_ID(ffn_moe_down-66): 199 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 408 us MUL_MAT_ID(ffn_moe_down-67): 177 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 372 us MUL_MAT_ID(ffn_moe_down-68): 197 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 379 us MUL_MAT_ID(ffn_moe_down-69): 223 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 388 us MUL_MAT_ID(ffn_moe_down-70): 205 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 386 us MUL_MAT_ID(ffn_moe_down-71): 199 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 386 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 423 us MUL_MAT_ID(ffn_moe_down-73): 197 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 378 us MUL_MAT_ID(ffn_moe_down-74): 288 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 400 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 403 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 383 us MUL_MAT_ID(ffn_moe_down-77): 300 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 395 us MUL_MAT_ID(ffn_moe_down-78): 202 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 411 us MUL_MAT_ID(ffn_moe_down-79): 290 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 377 us MUL_MAT_ID(ffn_moe_down-80): 251 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 423 us MUL_MAT_ID(ffn_moe_down-81): 303 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 407 us MUL_MAT_ID(ffn_moe_down-82): 214 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 412 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 410 us MUL_MAT_ID(ffn_moe_down-84): 296 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 393 us MUL_MAT_ID(ffn_moe_down-85): 224 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 377 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 451 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 394 us MUL_MAT_ID(ffn_moe_down-88): 189 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 453 us MUL_MAT_ID(ffn_moe_down-89): 243 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 425 us MUL_MAT_ID(ffn_moe_down-90): 240 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 377 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 48 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 513 us MUL_MAT_ID(ffn_moe_down-25): 205 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 473 us MUL_MAT_ID(ffn_moe_down-26): 188 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 381 us MUL_MAT_ID(ffn_moe_down-27): 212 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 488 us MUL_MAT_ID(ffn_moe_down-28): 190 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 412 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 410 us MUL_MAT_ID(ffn_moe_down-30): 240 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 395 us MUL_MAT_ID(ffn_moe_down-31): 228 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 352 us MUL_MAT_ID(ffn_moe_down-32): 205 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 395 us MUL_MAT_ID(ffn_moe_down-33): 610 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 360 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 400 us MUL_MAT_ID(ffn_moe_down-35): 218 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 425 us MUL_MAT_ID(ffn_moe_down-36): 178 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 400 us MUL_MAT_ID(ffn_moe_down-37): 283 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 428 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 363 us MUL_MAT_ID(ffn_moe_down-39): 244 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 356 us MUL_MAT_ID(ffn_moe_down-40): 238 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 357 us MUL_MAT_ID(ffn_moe_down-41): 363 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 420 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 369 us MUL_MAT_ID(ffn_moe_down-43): 216 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 367 us MUL_MAT_ID(ffn_moe_down-44): 254 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 384 us MUL_MAT_ID(ffn_moe_down-45): 194 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 365 us MUL_MAT_ID(ffn_moe_down-46): 210 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 482 us MUL_MAT_ID(ffn_moe_down-47): 176 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 408 us MUL_MAT_ID(ffn_moe_down-48): 216 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 349 us MUL_MAT_ID(ffn_moe_down-49): 220 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 403 us MUL_MAT_ID(ffn_moe_down-50): 187 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 392 us MUL_MAT_ID(ffn_moe_down-51): 189 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 371 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 660 us MUL_MAT_ID(ffn_moe_down-53): 224 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 545 us MUL_MAT_ID(ffn_moe_down-54): 248 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 394 us MUL_MAT_ID(ffn_moe_down-55): 189 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 218 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 404 us MUL_MAT_ID(ffn_moe_down-57): 222 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 395 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 376 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 424 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 417 us MUL_MAT_ID(ffn_moe_down-61): 200 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 412 us MUL_MAT_ID(ffn_moe_down-62): 206 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 343 us MUL_MAT_ID(ffn_moe_down-63): 322 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 411 us MUL_MAT_ID(ffn_moe_down-64): 233 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 403 us MUL_MAT_ID(ffn_moe_down-65): 263 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 402 us MUL_MAT_ID(ffn_moe_down-66): 297 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 358 us MUL_MAT_ID(ffn_moe_down-67): 239 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 409 us MUL_MAT_ID(ffn_moe_down-68): 208 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 377 us MUL_MAT_ID(ffn_moe_down-69): 500 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 389 us MUL_MAT_ID(ffn_moe_down-70): 370 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 597 us MUL_MAT_ID(ffn_moe_down-71): 240 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 768 us MUL_MAT_ID(ffn_moe_down-72): 359 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 357 us MUL_MAT_ID(ffn_moe_down-73): 205 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 388 us MUL_MAT_ID(ffn_moe_down-74): 318 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 435 us MUL_MAT_ID(ffn_moe_down-75): 206 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 408 us MUL_MAT_ID(ffn_moe_down-76): 211 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 423 us MUL_MAT_ID(ffn_moe_down-77): 233 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 407 us MUL_MAT_ID(ffn_moe_down-78): 206 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 358 us MUL_MAT_ID(ffn_moe_down-79): 222 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 411 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 238 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 444 us MUL_MAT_ID(ffn_moe_down-82): 224 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 413 us MUL_MAT_ID(ffn_moe_down-83): 211 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 456 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 382 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 414 us MUL_MAT_ID(ffn_moe_down-86): 183 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 438 us MUL_MAT_ID(ffn_moe_down-87): 366 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 420 us MUL_MAT_ID(ffn_moe_down-88): 251 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 375 us MUL_MAT_ID(ffn_moe_down-89): 196 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 441 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 425 us MUL_MAT_ID(ffn_moe_down-91): 245 us ggml_barrier(...): 20 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 384 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 405 us MUL_MAT_ID(ffn_moe_down-26): 214 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 465 us MUL_MAT_ID(ffn_moe_down-27): 214 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 428 us MUL_MAT_ID(ffn_moe_down-28): 214 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 449 us MUL_MAT_ID(ffn_moe_down-29): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 351 us MUL_MAT_ID(ffn_moe_down-30): 185 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 371 us MUL_MAT_ID(ffn_moe_down-31): 205 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 391 us MUL_MAT_ID(ffn_moe_down-32): 181 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 386 us MUL_MAT_ID(ffn_moe_down-33): 184 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 486 us MUL_MAT_ID(ffn_moe_down-34): 391 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 400 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 387 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 428 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 435 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 420 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 437 us MUL_MAT_ID(ffn_moe_down-40): 242 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 405 us MUL_MAT_ID(ffn_moe_down-41): 204 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 421 us MUL_MAT_ID(ffn_moe_down-42): 385 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 547 us MUL_MAT_ID(ffn_moe_down-43): 213 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 411 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 393 us MUL_MAT_ID(ffn_moe_down-45): 542 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 415 us MUL_MAT_ID(ffn_moe_down-46): 186 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 425 us MUL_MAT_ID(ffn_moe_down-47): 531 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 404 us MUL_MAT_ID(ffn_moe_down-48): 526 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 493 us MUL_MAT_ID(ffn_moe_down-49): 192 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 423 us MUL_MAT_ID(ffn_moe_down-50): 215 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 386 us MUL_MAT_ID(ffn_moe_down-51): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 444 us MUL_MAT_ID(ffn_moe_down-52): 219 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 403 us MUL_MAT_ID(ffn_moe_down-53): 181 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 398 us MUL_MAT_ID(ffn_moe_down-54): 301 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 401 us MUL_MAT_ID(ffn_moe_down-55): 186 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 432 us MUL_MAT_ID(ffn_moe_down-56): 212 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 430 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 459 us MUL_MAT_ID(ffn_moe_down-58): 210 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 414 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 500 us MUL_MAT_ID(ffn_moe_down-60): 209 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 409 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 424 us MUL_MAT_ID(ffn_moe_down-62): 317 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 477 us MUL_MAT_ID(ffn_moe_down-63): 181 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 387 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 469 us MUL_MAT_ID(ffn_moe_down-65): 203 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 389 us MUL_MAT_ID(ffn_moe_down-66): 173 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 433 us MUL_MAT_ID(ffn_moe_down-67): 196 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 388 us MUL_MAT_ID(ffn_moe_down-68): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 481 us MUL_MAT_ID(ffn_moe_down-69): 187 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 400 us MUL_MAT_ID(ffn_moe_down-70): 229 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 419 us MUL_MAT_ID(ffn_moe_down-71): 201 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 571 us MUL_MAT_ID(ffn_moe_down-72): 219 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 447 us MUL_MAT_ID(ffn_moe_down-73): 500 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 364 us MUL_MAT_ID(ffn_moe_down-74): 501 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 498 us MUL_MAT_ID(ffn_moe_down-75): 342 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 390 us MUL_MAT_ID(ffn_moe_down-76): 507 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 406 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 392 us MUL_MAT_ID(ffn_moe_down-78): 209 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 387 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 405 us MUL_MAT_ID(ffn_moe_down-80): 218 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 694 us MUL_MAT_ID(ffn_moe_down-81): 355 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 367 us MUL_MAT_ID(ffn_moe_down-82): 205 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 507 us MUL_MAT_ID(ffn_moe_down-83): 226 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 407 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 451 us MUL_MAT_ID(ffn_moe_down-85): 275 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 470 us MUL_MAT_ID(ffn_moe_down-86): 393 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 366 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 428 us MUL_MAT_ID(ffn_moe_down-88): 418 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 380 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 406 us MUL_MAT_ID(ffn_moe_down-90): 229 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 436 us MUL_MAT_ID(ffn_moe_down-91): 278 us ggml_barrier(...): 203 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 488 us MUL_MAT_ID(ffn_moe_down-25): 196 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 433 us MUL_MAT_ID(ffn_moe_down-26): 227 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 645 us MUL_MAT_ID(ffn_moe_down-27): 331 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 373 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 469 us MUL_MAT_ID(ffn_moe_down-29): 314 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 409 us MUL_MAT_ID(ffn_moe_down-30): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 469 us MUL_MAT_ID(ffn_moe_down-31): 178 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 413 us MUL_MAT_ID(ffn_moe_down-32): 205 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 357 us MUL_MAT_ID(ffn_moe_down-33): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 384 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 435 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 437 us MUL_MAT_ID(ffn_moe_down-36): 317 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 419 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 384 us MUL_MAT_ID(ffn_moe_down-38): 200 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 408 us MUL_MAT_ID(ffn_moe_down-39): 318 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 350 us MUL_MAT_ID(ffn_moe_down-40): 256 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 402 us MUL_MAT_ID(ffn_moe_down-41): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 353 us MUL_MAT_ID(ffn_moe_down-42): 187 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 387 us MUL_MAT_ID(ffn_moe_down-43): 449 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 399 us MUL_MAT_ID(ffn_moe_down-44): 289 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 438 us MUL_MAT_ID(ffn_moe_down-45): 208 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 476 us MUL_MAT_ID(ffn_moe_down-46): 571 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 527 us MUL_MAT_ID(ffn_moe_down-47): 257 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 392 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 366 us MUL_MAT_ID(ffn_moe_down-49): 204 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 392 us MUL_MAT_ID(ffn_moe_down-50): 241 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 361 us MUL_MAT_ID(ffn_moe_down-51): 171 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 411 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 380 us MUL_MAT_ID(ffn_moe_down-53): 570 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 431 us MUL_MAT_ID(ffn_moe_down-54): 227 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 346 us MUL_MAT_ID(ffn_moe_down-55): 367 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 626 us MUL_MAT_ID(ffn_moe_down-56): 382 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 398 us MUL_MAT_ID(ffn_moe_down-57): 233 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 424 us MUL_MAT_ID(ffn_moe_down-58): 220 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 380 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 482 us MUL_MAT_ID(ffn_moe_down-60): 315 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 406 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 439 us MUL_MAT_ID(ffn_moe_down-62): 420 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 391 us MUL_MAT_ID(ffn_moe_down-63): 214 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 387 us MUL_MAT_ID(ffn_moe_down-64): 247 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 409 us MUL_MAT_ID(ffn_moe_down-65): 202 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 410 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 470 us MUL_MAT_ID(ffn_moe_down-67): 210 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 414 us MUL_MAT_ID(ffn_moe_down-68): 303 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 388 us MUL_MAT_ID(ffn_moe_down-69): 259 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 421 us MUL_MAT_ID(ffn_moe_down-70): 227 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 416 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 498 us MUL_MAT_ID(ffn_moe_down-72): 570 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 422 us MUL_MAT_ID(ffn_moe_down-73): 555 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 379 us MUL_MAT_ID(ffn_moe_down-74): 253 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 446 us MUL_MAT_ID(ffn_moe_down-75): 302 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 407 us MUL_MAT_ID(ffn_moe_down-76): 298 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 446 us MUL_MAT_ID(ffn_moe_down-77): 236 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 387 us MUL_MAT_ID(ffn_moe_down-78): 185 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 447 us MUL_MAT_ID(ffn_moe_down-79): 221 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 485 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 410 us MUL_MAT_ID(ffn_moe_down-81): 288 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 475 us MUL_MAT_ID(ffn_moe_down-82): 422 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 409 us MUL_MAT_ID(ffn_moe_down-83): 271 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 438 us MUL_MAT_ID(ffn_moe_down-84): 245 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 751 us MUL_MAT_ID(ffn_moe_down-85): 404 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 765 us MUL_MAT_ID(ffn_moe_down-86): 341 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 427 us MUL_MAT_ID(ffn_moe_down-87): 217 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 523 us MUL_MAT_ID(ffn_moe_down-88): 208 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 429 us MUL_MAT_ID(ffn_moe_down-89): 410 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 440 us MUL_MAT_ID(ffn_moe_down-90): 189 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 498 us MUL_MAT_ID(ffn_moe_down-91): 249 us ggml_barrier(...): 170 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 378 us MUL_MAT_ID(ffn_moe_down-25): 397 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 402 us MUL_MAT_ID(ffn_moe_down-26): 185 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 638 us MUL_MAT_ID(ffn_moe_down-27): 243 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 464 us MUL_MAT_ID(ffn_moe_down-28): 212 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 411 us MUL_MAT_ID(ffn_moe_down-29): 210 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 430 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 422 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 409 us MUL_MAT_ID(ffn_moe_down-32): 205 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 457 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 535 us MUL_MAT_ID(ffn_moe_down-34): 389 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 419 us MUL_MAT_ID(ffn_moe_down-35): 203 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 555 us MUL_MAT_ID(ffn_moe_down-36): 247 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 423 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 429 us MUL_MAT_ID(ffn_moe_down-38): 203 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 434 us MUL_MAT_ID(ffn_moe_down-39): 424 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 438 us MUL_MAT_ID(ffn_moe_down-40): 285 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 390 us MUL_MAT_ID(ffn_moe_down-41): 567 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 390 us MUL_MAT_ID(ffn_moe_down-42): 266 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 416 us MUL_MAT_ID(ffn_moe_down-43): 244 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 479 us MUL_MAT_ID(ffn_moe_down-44): 269 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 551 us MUL_MAT_ID(ffn_moe_down-45): 223 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 418 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 477 us MUL_MAT_ID(ffn_moe_down-47): 275 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 403 us MUL_MAT_ID(ffn_moe_down-48): 463 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 415 us MUL_MAT_ID(ffn_moe_down-49): 210 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 415 us MUL_MAT_ID(ffn_moe_down-50): 567 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 388 us MUL_MAT_ID(ffn_moe_down-51): 203 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 445 us MUL_MAT_ID(ffn_moe_down-52): 555 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 401 us MUL_MAT_ID(ffn_moe_down-53): 621 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 365 us MUL_MAT_ID(ffn_moe_down-54): 215 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 385 us MUL_MAT_ID(ffn_moe_down-55): 419 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 379 us MUL_MAT_ID(ffn_moe_down-56): 394 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 423 us MUL_MAT_ID(ffn_moe_down-57): 233 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 471 us MUL_MAT_ID(ffn_moe_down-58): 206 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 378 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 393 us MUL_MAT_ID(ffn_moe_down-60): 230 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 426 us MUL_MAT_ID(ffn_moe_down-61): 374 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 400 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 424 us MUL_MAT_ID(ffn_moe_down-63): 649 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 396 us MUL_MAT_ID(ffn_moe_down-64): 260 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 497 us MUL_MAT_ID(ffn_moe_down-65): 366 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 376 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 502 us MUL_MAT_ID(ffn_moe_down-67): 232 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 387 us MUL_MAT_ID(ffn_moe_down-68): 427 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 511 us MUL_MAT_ID(ffn_moe_down-69): 186 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 411 us MUL_MAT_ID(ffn_moe_down-70): 289 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 414 us MUL_MAT_ID(ffn_moe_down-71): 311 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 382 us MUL_MAT_ID(ffn_moe_down-72): 204 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 410 us MUL_MAT_ID(ffn_moe_down-73): 236 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 412 us MUL_MAT_ID(ffn_moe_down-74): 188 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 490 us MUL_MAT_ID(ffn_moe_down-75): 169 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 487 us MUL_MAT_ID(ffn_moe_down-76): 291 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 413 us MUL_MAT_ID(ffn_moe_down-77): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 358 us MUL_MAT_ID(ffn_moe_down-78): 263 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 441 us MUL_MAT_ID(ffn_moe_down-79): 226 us ggml_barrier(...): 335 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 507 us MUL_MAT_ID(ffn_moe_down-80): 219 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 441 us MUL_MAT_ID(ffn_moe_down-81): 300 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 473 us MUL_MAT_ID(ffn_moe_down-82): 236 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 429 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 547 us MUL_MAT_ID(ffn_moe_down-84): 251 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 412 us MUL_MAT_ID(ffn_moe_down-85): 230 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 399 us MUL_MAT_ID(ffn_moe_down-86): 196 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 425 us MUL_MAT_ID(ffn_moe_down-87): 225 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 600 us MUL_MAT_ID(ffn_moe_down-88): 244 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 755 us MUL_MAT_ID(ffn_moe_down-89): 361 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 426 us MUL_MAT_ID(ffn_moe_down-90): 225 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 410 us MUL_MAT_ID(ffn_moe_down-91): 250 us ggml_barrier(...): 211 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 374 us MUL_MAT_ID(ffn_moe_down-25): 181 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 396 us MUL_MAT_ID(ffn_moe_down-26): 236 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 436 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 413 us MUL_MAT_ID(ffn_moe_down-28): 188 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 395 us MUL_MAT_ID(ffn_moe_down-29): 217 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 432 us MUL_MAT_ID(ffn_moe_down-30): 213 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 419 us MUL_MAT_ID(ffn_moe_down-31): 217 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 441 us MUL_MAT_ID(ffn_moe_down-32): 396 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 370 us MUL_MAT_ID(ffn_moe_down-33): 230 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 465 us MUL_MAT_ID(ffn_moe_down-34): 237 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 408 us MUL_MAT_ID(ffn_moe_down-35): 213 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 577 us MUL_MAT_ID(ffn_moe_down-36): 241 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 436 us MUL_MAT_ID(ffn_moe_down-37): 260 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 409 us MUL_MAT_ID(ffn_moe_down-38): 206 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 443 us MUL_MAT_ID(ffn_moe_down-39): 266 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 426 us MUL_MAT_ID(ffn_moe_down-40): 483 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 402 us MUL_MAT_ID(ffn_moe_down-41): 208 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 397 us MUL_MAT_ID(ffn_moe_down-42): 208 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 413 us MUL_MAT_ID(ffn_moe_down-43): 516 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 402 us MUL_MAT_ID(ffn_moe_down-44): 207 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 391 us MUL_MAT_ID(ffn_moe_down-45): 619 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 395 us MUL_MAT_ID(ffn_moe_down-46): 209 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 418 us MUL_MAT_ID(ffn_moe_down-47): 585 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 399 us MUL_MAT_ID(ffn_moe_down-48): 202 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 410 us MUL_MAT_ID(ffn_moe_down-49): 349 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 419 us MUL_MAT_ID(ffn_moe_down-50): 209 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 395 us MUL_MAT_ID(ffn_moe_down-51): 235 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 428 us MUL_MAT_ID(ffn_moe_down-52): 286 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 413 us MUL_MAT_ID(ffn_moe_down-53): 207 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 373 us MUL_MAT_ID(ffn_moe_down-54): 209 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 433 us MUL_MAT_ID(ffn_moe_down-55): 221 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 411 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 457 us MUL_MAT_ID(ffn_moe_down-58): 217 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 399 us MUL_MAT_ID(ffn_moe_down-59): 206 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 426 us MUL_MAT_ID(ffn_moe_down-60): 219 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 420 us MUL_MAT_ID(ffn_moe_down-61): 237 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 425 us MUL_MAT_ID(ffn_moe_down-62): 238 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 443 us MUL_MAT_ID(ffn_moe_down-63): 244 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 440 us MUL_MAT_ID(ffn_moe_down-64): 246 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 441 us MUL_MAT_ID(ffn_moe_down-65): 225 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 437 us MUL_MAT_ID(ffn_moe_down-66): 588 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 557 us MUL_MAT_ID(ffn_moe_down-67): 267 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 446 us MUL_MAT_ID(ffn_moe_down-68): 235 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 437 us MUL_MAT_ID(ffn_moe_down-69): 626 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 477 us MUL_MAT_ID(ffn_moe_down-70): 226 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 428 us MUL_MAT_ID(ffn_moe_down-71): 234 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 448 us MUL_MAT_ID(ffn_moe_down-72): 220 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 435 us MUL_MAT_ID(ffn_moe_down-73): 227 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 467 us MUL_MAT_ID(ffn_moe_down-74): 323 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 433 us MUL_MAT_ID(ffn_moe_down-75): 229 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 460 us MUL_MAT_ID(ffn_moe_down-76): 230 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 442 us MUL_MAT_ID(ffn_moe_down-77): 232 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 436 us MUL_MAT_ID(ffn_moe_down-78): 225 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 436 us MUL_MAT_ID(ffn_moe_down-79): 233 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 437 us MUL_MAT_ID(ffn_moe_down-80): 231 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 512 us MUL_MAT_ID(ffn_moe_down-81): 250 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 530 us MUL_MAT_ID(ffn_moe_down-82): 220 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 453 us MUL_MAT_ID(ffn_moe_down-83): 228 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 440 us MUL_MAT_ID(ffn_moe_down-84): 251 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 420 us MUL_MAT_ID(ffn_moe_down-85): 311 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 450 us MUL_MAT_ID(ffn_moe_down-86): 222 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 497 us MUL_MAT_ID(ffn_moe_down-87): 639 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 430 us MUL_MAT_ID(ffn_moe_down-88): 226 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 434 us MUL_MAT_ID(ffn_moe_down-89): 232 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 575 us MUL_MAT_ID(ffn_moe_down-90): 478 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 440 us MUL_MAT_ID(ffn_moe_down-91): 252 us ggml_barrier(...): 5 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 484 us MUL_MAT_ID(ffn_moe_down-25): 591 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 409 us MUL_MAT_ID(ffn_moe_down-26): 256 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 415 us MUL_MAT_ID(ffn_moe_down-27): 218 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 410 us MUL_MAT_ID(ffn_moe_down-28): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 403 us MUL_MAT_ID(ffn_moe_down-29): 211 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 436 us MUL_MAT_ID(ffn_moe_down-30): 227 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 501 us MUL_MAT_ID(ffn_moe_down-31): 340 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 416 us MUL_MAT_ID(ffn_moe_down-32): 210 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 414 us MUL_MAT_ID(ffn_moe_down-33): 486 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 409 us MUL_MAT_ID(ffn_moe_down-34): 233 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 396 us MUL_MAT_ID(ffn_moe_down-35): 237 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 425 us MUL_MAT_ID(ffn_moe_down-36): 213 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 385 us MUL_MAT_ID(ffn_moe_down-37): 246 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 540 us MUL_MAT_ID(ffn_moe_down-38): 395 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 411 us MUL_MAT_ID(ffn_moe_down-39): 735 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 385 us MUL_MAT_ID(ffn_moe_down-40): 242 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 448 us MUL_MAT_ID(ffn_moe_down-41): 232 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 406 us MUL_MAT_ID(ffn_moe_down-42): 214 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 411 us MUL_MAT_ID(ffn_moe_down-43): 229 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 407 us MUL_MAT_ID(ffn_moe_down-44): 205 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 459 us MUL_MAT_ID(ffn_moe_down-45): 240 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 417 us MUL_MAT_ID(ffn_moe_down-46): 213 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 418 us MUL_MAT_ID(ffn_moe_down-47): 211 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 383 us MUL_MAT_ID(ffn_moe_down-48): 478 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 440 us MUL_MAT_ID(ffn_moe_down-49): 233 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 414 us MUL_MAT_ID(ffn_moe_down-50): 617 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 483 us MUL_MAT_ID(ffn_moe_down-51): 285 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 397 us MUL_MAT_ID(ffn_moe_down-52): 214 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 408 us MUL_MAT_ID(ffn_moe_down-53): 359 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 400 us MUL_MAT_ID(ffn_moe_down-54): 205 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 418 us MUL_MAT_ID(ffn_moe_down-55): 237 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 411 us MUL_MAT_ID(ffn_moe_down-56): 263 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 409 us MUL_MAT_ID(ffn_moe_down-57): 213 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 399 us MUL_MAT_ID(ffn_moe_down-58): 244 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 397 us MUL_MAT_ID(ffn_moe_down-59): 212 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 430 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 500 us MUL_MAT_ID(ffn_moe_down-61): 194 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 438 us MUL_MAT_ID(ffn_moe_down-62): 269 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 417 us MUL_MAT_ID(ffn_moe_down-63): 258 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 380 us MUL_MAT_ID(ffn_moe_down-64): 232 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 417 us MUL_MAT_ID(ffn_moe_down-65): 207 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 417 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 378 us MUL_MAT_ID(ffn_moe_down-67): 211 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 445 us MUL_MAT_ID(ffn_moe_down-68): 597 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 409 us MUL_MAT_ID(ffn_moe_down-69): 258 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 388 us MUL_MAT_ID(ffn_moe_down-70): 207 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 414 us MUL_MAT_ID(ffn_moe_down-71): 241 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 219 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 404 us MUL_MAT_ID(ffn_moe_down-73): 205 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 436 us MUL_MAT_ID(ffn_moe_down-74): 224 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 409 us MUL_MAT_ID(ffn_moe_down-75): 208 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 443 us MUL_MAT_ID(ffn_moe_down-76): 226 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 446 us MUL_MAT_ID(ffn_moe_down-77): 277 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 432 us MUL_MAT_ID(ffn_moe_down-78): 239 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 407 us MUL_MAT_ID(ffn_moe_down-79): 211 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 435 us MUL_MAT_ID(ffn_moe_down-80): 228 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 390 us MUL_MAT_ID(ffn_moe_down-81): 333 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 441 us MUL_MAT_ID(ffn_moe_down-82): 575 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 452 us MUL_MAT_ID(ffn_moe_down-83): 302 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 398 us MUL_MAT_ID(ffn_moe_down-84): 242 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 459 us MUL_MAT_ID(ffn_moe_down-85): 301 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 412 us MUL_MAT_ID(ffn_moe_down-86): 591 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 457 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 420 us MUL_MAT_ID(ffn_moe_down-88): 428 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 393 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 391 us MUL_MAT_ID(ffn_moe_down-90): 236 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 408 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 33 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 374 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 510 us MUL_MAT_ID(ffn_moe_down-26): 197 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 438 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 360 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 360 us MUL_MAT_ID(ffn_moe_down-29): 218 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 589 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 374 us MUL_MAT_ID(ffn_moe_down-31): 203 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 413 us MUL_MAT_ID(ffn_moe_down-32): 227 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 342 us MUL_MAT_ID(ffn_moe_down-33): 196 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 397 us MUL_MAT_ID(ffn_moe_down-34): 222 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 385 us MUL_MAT_ID(ffn_moe_down-35): 222 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 413 us MUL_MAT_ID(ffn_moe_down-36): 192 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 406 us MUL_MAT_ID(ffn_moe_down-37): 239 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 393 us MUL_MAT_ID(ffn_moe_down-39): 250 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 412 us MUL_MAT_ID(ffn_moe_down-40): 262 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 388 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 399 us MUL_MAT_ID(ffn_moe_down-42): 188 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 358 us MUL_MAT_ID(ffn_moe_down-43): 323 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 548 us MUL_MAT_ID(ffn_moe_down-44): 305 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 462 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 416 us MUL_MAT_ID(ffn_moe_down-46): 246 us ggml_barrier(...): 372 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 387 us MUL_MAT_ID(ffn_moe_down-47): 212 us ggml_barrier(...): 369 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 403 us MUL_MAT_ID(ffn_moe_down-48): 198 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 405 us MUL_MAT_ID(ffn_moe_down-49): 273 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 430 us MUL_MAT_ID(ffn_moe_down-50): 218 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 399 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 465 us MUL_MAT_ID(ffn_moe_down-52): 210 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 417 us MUL_MAT_ID(ffn_moe_down-53): 301 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 404 us MUL_MAT_ID(ffn_moe_down-54): 373 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 452 us MUL_MAT_ID(ffn_moe_down-55): 240 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 415 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 383 us MUL_MAT_ID(ffn_moe_down-57): 200 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 390 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 380 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 395 us MUL_MAT_ID(ffn_moe_down-60): 197 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 345 us MUL_MAT_ID(ffn_moe_down-61): 180 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 536 us MUL_MAT_ID(ffn_moe_down-62): 219 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 387 us MUL_MAT_ID(ffn_moe_down-63): 248 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 402 us MUL_MAT_ID(ffn_moe_down-64): 295 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 410 us MUL_MAT_ID(ffn_moe_down-65): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 451 us MUL_MAT_ID(ffn_moe_down-66): 171 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 402 us MUL_MAT_ID(ffn_moe_down-67): 189 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 435 us MUL_MAT_ID(ffn_moe_down-68): 184 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 486 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 398 us MUL_MAT_ID(ffn_moe_down-70): 206 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 374 us MUL_MAT_ID(ffn_moe_down-71): 238 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 382 us MUL_MAT_ID(ffn_moe_down-72): 432 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 488 us MUL_MAT_ID(ffn_moe_down-74): 191 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 367 us MUL_MAT_ID(ffn_moe_down-75): 200 us ggml_barrier(...): 356 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 392 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 439 us MUL_MAT_ID(ffn_moe_down-77): 267 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 417 us MUL_MAT_ID(ffn_moe_down-78): 205 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 422 us MUL_MAT_ID(ffn_moe_down-79): 225 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 400 us MUL_MAT_ID(ffn_moe_down-80): 196 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 320 us MUL_MAT_ID(ffn_moe_down-81): 249 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 380 us MUL_MAT_ID(ffn_moe_down-82): 472 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 390 us MUL_MAT_ID(ffn_moe_down-83): 523 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 356 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 363 us MUL_MAT_ID(ffn_moe_down-85): 245 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 359 us MUL_MAT_ID(ffn_moe_down-86): 204 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 394 us MUL_MAT_ID(ffn_moe_down-87): 329 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 410 us MUL_MAT_ID(ffn_moe_down-88): 397 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 415 us MUL_MAT_ID(ffn_moe_down-89): 217 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 387 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 433 us MUL_MAT_ID(ffn_moe_down-91): 229 us ggml_barrier(...): 27 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 373 us MUL_MAT_ID(ffn_moe_down-25): 208 us ggml_barrier(...): 587 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 419 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 386 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 432 us MUL_MAT_ID(ffn_moe_down-27): 204 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 383 us MUL_MAT_ID(ffn_moe_down-28): 239 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 410 us MUL_MAT_ID(ffn_moe_down-29): 200 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 406 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 352 us MUL_MAT_ID(ffn_moe_down-31): 245 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 514 us MUL_MAT_ID(ffn_moe_down-32): 309 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 348 us MUL_MAT_ID(ffn_moe_down-33): 202 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 439 us MUL_MAT_ID(ffn_moe_down-34): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 534 us MUL_MAT_ID(ffn_moe_down-35): 208 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 395 us MUL_MAT_ID(ffn_moe_down-36): 189 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 388 us MUL_MAT_ID(ffn_moe_down-37): 637 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 390 us MUL_MAT_ID(ffn_moe_down-38): 185 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 403 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 429 us MUL_MAT_ID(ffn_moe_down-40): 229 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 405 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 459 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 412 us MUL_MAT_ID(ffn_moe_down-43): 197 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 387 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 351 us MUL_MAT_ID(ffn_moe_down-45): 204 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 451 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 382 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 428 us MUL_MAT_ID(ffn_moe_down-48): 182 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 415 us MUL_MAT_ID(ffn_moe_down-49): 200 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 407 us MUL_MAT_ID(ffn_moe_down-50): 197 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 410 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 433 us MUL_MAT_ID(ffn_moe_down-52): 215 us ggml_barrier(...): 360 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 395 us MUL_MAT_ID(ffn_moe_down-53): 267 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 416 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 390 us MUL_MAT_ID(ffn_moe_down-55): 220 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 440 us MUL_MAT_ID(ffn_moe_down-56): 453 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 375 us MUL_MAT_ID(ffn_moe_down-57): 318 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 419 us MUL_MAT_ID(ffn_moe_down-58): 357 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 429 us MUL_MAT_ID(ffn_moe_down-59): 259 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 434 us MUL_MAT_ID(ffn_moe_down-60): 464 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 459 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 394 us MUL_MAT_ID(ffn_moe_down-63): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 455 us MUL_MAT_ID(ffn_moe_down-64): 203 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 417 us MUL_MAT_ID(ffn_moe_down-65): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 387 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 368 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 397 us MUL_MAT_ID(ffn_moe_down-68): 187 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 368 us MUL_MAT_ID(ffn_moe_down-69): 188 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 368 us MUL_MAT_ID(ffn_moe_down-70): 587 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 366 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 410 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 382 us MUL_MAT_ID(ffn_moe_down-73): 213 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 398 us MUL_MAT_ID(ffn_moe_down-74): 211 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 435 us MUL_MAT_ID(ffn_moe_down-75): 180 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 372 us MUL_MAT_ID(ffn_moe_down-76): 206 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 409 us MUL_MAT_ID(ffn_moe_down-77): 223 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 347 us MUL_MAT_ID(ffn_moe_down-78): 204 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 384 us MUL_MAT_ID(ffn_moe_down-79): 203 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 388 us MUL_MAT_ID(ffn_moe_down-80): 229 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 416 us MUL_MAT_ID(ffn_moe_down-81): 213 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 393 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 343 us MUL_MAT_ID(ffn_moe_down-83): 189 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 410 us MUL_MAT_ID(ffn_moe_down-84): 369 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 369 us MUL_MAT_ID(ffn_moe_down-85): 228 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 385 us MUL_MAT_ID(ffn_moe_down-86): 207 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 248 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 408 us MUL_MAT_ID(ffn_moe_down-88): 203 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 382 us MUL_MAT_ID(ffn_moe_down-89): 235 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 409 us MUL_MAT_ID(ffn_moe_down-90): 238 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 387 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 67 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 364 us MUL_MAT_ID(ffn_moe_down-25): 471 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 433 us MUL_MAT_ID(ffn_moe_down-26): 234 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 426 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 440 us MUL_MAT_ID(ffn_moe_down-28): 210 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 539 us MUL_MAT_ID(ffn_moe_down-29): 454 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 435 us MUL_MAT_ID(ffn_moe_down-30): 302 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 365 us MUL_MAT_ID(ffn_moe_down-31): 193 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 455 us MUL_MAT_ID(ffn_moe_down-32): 291 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 416 us MUL_MAT_ID(ffn_moe_down-33): 189 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 388 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 345 us MUL_MAT_ID(ffn_moe_down-35): 201 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 449 us MUL_MAT_ID(ffn_moe_down-36): 526 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 377 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 369 us MUL_MAT_ID(ffn_moe_down-38): 328 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 381 us MUL_MAT_ID(ffn_moe_down-39): 244 us ggml_barrier(...): 372 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 390 us MUL_MAT_ID(ffn_moe_down-40): 248 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 395 us MUL_MAT_ID(ffn_moe_down-41): 195 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 442 us MUL_MAT_ID(ffn_moe_down-42): 180 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 454 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 534 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 381 us MUL_MAT_ID(ffn_moe_down-45): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 394 us MUL_MAT_ID(ffn_moe_down-46): 184 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 368 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 409 us MUL_MAT_ID(ffn_moe_down-48): 205 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 217 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 380 us MUL_MAT_ID(ffn_moe_down-50): 204 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 503 us MUL_MAT_ID(ffn_moe_down-51): 440 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 374 us MUL_MAT_ID(ffn_moe_down-52): 441 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 370 us MUL_MAT_ID(ffn_moe_down-53): 204 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 449 us MUL_MAT_ID(ffn_moe_down-54): 189 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 355 us MUL_MAT_ID(ffn_moe_down-55): 388 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 400 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 384 us MUL_MAT_ID(ffn_moe_down-57): 200 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 388 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 516 us MUL_MAT_ID(ffn_moe_down-59): 184 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 379 us MUL_MAT_ID(ffn_moe_down-60): 599 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 390 us MUL_MAT_ID(ffn_moe_down-61): 333 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 398 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 383 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 372 us MUL_MAT_ID(ffn_moe_down-64): 203 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 341 us MUL_MAT_ID(ffn_moe_down-65): 405 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 418 us MUL_MAT_ID(ffn_moe_down-66): 205 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 350 us MUL_MAT_ID(ffn_moe_down-67): 211 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 418 us MUL_MAT_ID(ffn_moe_down-68): 237 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 401 us MUL_MAT_ID(ffn_moe_down-69): 199 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 438 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 427 us MUL_MAT_ID(ffn_moe_down-71): 197 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 393 us MUL_MAT_ID(ffn_moe_down-72): 223 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 393 us MUL_MAT_ID(ffn_moe_down-73): 329 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 382 us MUL_MAT_ID(ffn_moe_down-74): 293 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 386 us MUL_MAT_ID(ffn_moe_down-75): 209 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 429 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 426 us MUL_MAT_ID(ffn_moe_down-77): 188 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 347 us MUL_MAT_ID(ffn_moe_down-78): 225 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 402 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 355 us MUL_MAT_ID(ffn_moe_down-80): 265 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 208 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 521 us MUL_MAT_ID(ffn_moe_down-82): 191 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 411 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 386 us MUL_MAT_ID(ffn_moe_down-84): 224 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 371 us MUL_MAT_ID(ffn_moe_down-85): 349 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 406 us MUL_MAT_ID(ffn_moe_down-86): 209 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 196 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 357 us MUL_MAT_ID(ffn_moe_down-89): 183 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 478 us MUL_MAT_ID(ffn_moe_down-90): 243 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 387 us MUL_MAT_ID(ffn_moe_down-91): 509 us ggml_barrier(...): 43 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 394 us MUL_MAT_ID(ffn_moe_down-25): 199 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 431 us MUL_MAT_ID(ffn_moe_down-26): 224 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 418 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 408 us MUL_MAT_ID(ffn_moe_down-28): 399 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 477 us MUL_MAT_ID(ffn_moe_down-29): 304 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 423 us MUL_MAT_ID(ffn_moe_down-30): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 449 us MUL_MAT_ID(ffn_moe_down-31): 199 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 426 us MUL_MAT_ID(ffn_moe_down-32): 203 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 410 us MUL_MAT_ID(ffn_moe_down-33): 219 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 377 us MUL_MAT_ID(ffn_moe_down-34): 205 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 365 us MUL_MAT_ID(ffn_moe_down-35): 348 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 407 us MUL_MAT_ID(ffn_moe_down-36): 222 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 436 us MUL_MAT_ID(ffn_moe_down-37): 247 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 378 us MUL_MAT_ID(ffn_moe_down-38): 204 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 428 us MUL_MAT_ID(ffn_moe_down-39): 248 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 507 us MUL_MAT_ID(ffn_moe_down-40): 250 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 419 us MUL_MAT_ID(ffn_moe_down-41): 209 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 428 us MUL_MAT_ID(ffn_moe_down-42): 338 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 537 us MUL_MAT_ID(ffn_moe_down-43): 239 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 393 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 415 us MUL_MAT_ID(ffn_moe_down-45): 594 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 373 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 469 us MUL_MAT_ID(ffn_moe_down-47): 424 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 426 us MUL_MAT_ID(ffn_moe_down-48): 199 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 427 us MUL_MAT_ID(ffn_moe_down-49): 217 us ggml_barrier(...): 522 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 585 us MUL_MAT_ID(ffn_moe_down-50): 272 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 408 us MUL_MAT_ID(ffn_moe_down-51): 194 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 479 us MUL_MAT_ID(ffn_moe_down-52): 201 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 400 us MUL_MAT_ID(ffn_moe_down-53): 208 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 390 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 460 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 385 us MUL_MAT_ID(ffn_moe_down-56): 207 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 468 us MUL_MAT_ID(ffn_moe_down-57): 196 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 400 us MUL_MAT_ID(ffn_moe_down-58): 164 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 378 us MUL_MAT_ID(ffn_moe_down-59): 179 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 180 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 393 us MUL_MAT_ID(ffn_moe_down-61): 212 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 402 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 357 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 401 us MUL_MAT_ID(ffn_moe_down-64): 205 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 387 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 391 us MUL_MAT_ID(ffn_moe_down-66): 186 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 371 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 517 us MUL_MAT_ID(ffn_moe_down-68): 324 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 451 us MUL_MAT_ID(ffn_moe_down-69): 205 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 428 us MUL_MAT_ID(ffn_moe_down-70): 283 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 387 us MUL_MAT_ID(ffn_moe_down-71): 330 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 362 us MUL_MAT_ID(ffn_moe_down-72): 189 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 210 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 373 us MUL_MAT_ID(ffn_moe_down-74): 187 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 404 us MUL_MAT_ID(ffn_moe_down-75): 196 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 422 us MUL_MAT_ID(ffn_moe_down-76): 203 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 476 us MUL_MAT_ID(ffn_moe_down-77): 227 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 418 us MUL_MAT_ID(ffn_moe_down-78): 286 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 374 us MUL_MAT_ID(ffn_moe_down-79): 177 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 369 us MUL_MAT_ID(ffn_moe_down-80): 203 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 393 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 373 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 347 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 430 us MUL_MAT_ID(ffn_moe_down-84): 316 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 390 us MUL_MAT_ID(ffn_moe_down-85): 223 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 336 us MUL_MAT_ID(ffn_moe_down-86): 183 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 446 us MUL_MAT_ID(ffn_moe_down-87): 247 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 436 us MUL_MAT_ID(ffn_moe_down-88): 175 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 469 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 381 us MUL_MAT_ID(ffn_moe_down-90): 218 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 420 us MUL_MAT_ID(ffn_moe_down-91): 241 us ggml_barrier(...): 27 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 471 us MUL_MAT_ID(ffn_moe_down-25): 211 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 405 us MUL_MAT_ID(ffn_moe_down-26): 204 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 420 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 485 us MUL_MAT_ID(ffn_moe_down-28): 275 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 530 us MUL_MAT_ID(ffn_moe_down-29): 212 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 407 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 421 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 359 us MUL_MAT_ID(ffn_moe_down-32): 185 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 373 us MUL_MAT_ID(ffn_moe_down-33): 193 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 440 us MUL_MAT_ID(ffn_moe_down-34): 246 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 402 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 334 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 401 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 456 us MUL_MAT_ID(ffn_moe_down-38): 186 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 397 us MUL_MAT_ID(ffn_moe_down-39): 259 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 414 us MUL_MAT_ID(ffn_moe_down-40): 198 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 404 us MUL_MAT_ID(ffn_moe_down-41): 254 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 493 us MUL_MAT_ID(ffn_moe_down-42): 174 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 422 us MUL_MAT_ID(ffn_moe_down-43): 206 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 436 us MUL_MAT_ID(ffn_moe_down-44): 305 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 440 us MUL_MAT_ID(ffn_moe_down-45): 215 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 395 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 396 us MUL_MAT_ID(ffn_moe_down-47): 266 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 407 us MUL_MAT_ID(ffn_moe_down-48): 206 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 361 us MUL_MAT_ID(ffn_moe_down-49): 192 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 374 us MUL_MAT_ID(ffn_moe_down-50): 226 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 443 us MUL_MAT_ID(ffn_moe_down-51): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 444 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 469 us MUL_MAT_ID(ffn_moe_down-53): 205 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 421 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 439 us MUL_MAT_ID(ffn_moe_down-55): 247 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 442 us MUL_MAT_ID(ffn_moe_down-56): 224 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 383 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 388 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 451 us MUL_MAT_ID(ffn_moe_down-58): 206 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 389 us MUL_MAT_ID(ffn_moe_down-59): 207 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 350 us MUL_MAT_ID(ffn_moe_down-60): 173 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 397 us MUL_MAT_ID(ffn_moe_down-61): 180 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 350 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 403 us MUL_MAT_ID(ffn_moe_down-63): 203 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 539 us MUL_MAT_ID(ffn_moe_down-64): 173 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 465 us MUL_MAT_ID(ffn_moe_down-65): 277 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 369 us MUL_MAT_ID(ffn_moe_down-66): 187 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 430 us MUL_MAT_ID(ffn_moe_down-67): 223 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 386 us MUL_MAT_ID(ffn_moe_down-68): 191 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 431 us MUL_MAT_ID(ffn_moe_down-69): 169 us ggml_barrier(...): 322 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 393 us MUL_MAT_ID(ffn_moe_down-70): 184 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 422 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 337 us MUL_MAT_ID(ffn_moe_down-72): 201 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 169 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 408 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 390 us MUL_MAT_ID(ffn_moe_down-75): 364 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 397 us MUL_MAT_ID(ffn_moe_down-76): 214 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 409 us MUL_MAT_ID(ffn_moe_down-77): 209 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 412 us MUL_MAT_ID(ffn_moe_down-78): 270 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 348 us MUL_MAT_ID(ffn_moe_down-79): 268 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 425 us MUL_MAT_ID(ffn_moe_down-80): 190 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 450 us MUL_MAT_ID(ffn_moe_down-81): 187 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 350 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 374 us MUL_MAT_ID(ffn_moe_down-83): 208 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 484 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 406 us MUL_MAT_ID(ffn_moe_down-85): 219 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 380 us MUL_MAT_ID(ffn_moe_down-86): 202 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 391 us MUL_MAT_ID(ffn_moe_down-87): 267 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 496 us MUL_MAT_ID(ffn_moe_down-88): 266 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 401 us MUL_MAT_ID(ffn_moe_down-89): 180 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 416 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 403 us MUL_MAT_ID(ffn_moe_down-91): 277 us ggml_barrier(...): 17 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 403 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 403 us MUL_MAT_ID(ffn_moe_down-26): 203 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 355 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 401 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 342 us MUL_MAT_ID(ffn_moe_down-29): 386 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 433 us MUL_MAT_ID(ffn_moe_down-30): 260 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 415 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 409 us MUL_MAT_ID(ffn_moe_down-32): 212 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 400 us MUL_MAT_ID(ffn_moe_down-33): 245 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 399 us MUL_MAT_ID(ffn_moe_down-34): 251 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 438 us MUL_MAT_ID(ffn_moe_down-35): 200 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 376 us MUL_MAT_ID(ffn_moe_down-36): 203 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 348 us MUL_MAT_ID(ffn_moe_down-37): 309 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 355 us MUL_MAT_ID(ffn_moe_down-38): 435 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 373 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 399 us MUL_MAT_ID(ffn_moe_down-40): 220 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 445 us MUL_MAT_ID(ffn_moe_down-41): 252 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 380 us MUL_MAT_ID(ffn_moe_down-42): 217 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 429 us MUL_MAT_ID(ffn_moe_down-43): 210 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 419 us MUL_MAT_ID(ffn_moe_down-44): 187 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 438 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 438 us MUL_MAT_ID(ffn_moe_down-46): 210 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 383 us MUL_MAT_ID(ffn_moe_down-47): 331 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 378 us MUL_MAT_ID(ffn_moe_down-48): 498 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 396 us MUL_MAT_ID(ffn_moe_down-49): 199 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 370 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 421 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 404 us MUL_MAT_ID(ffn_moe_down-52): 374 us ggml_barrier(...): 344 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 440 us MUL_MAT_ID(ffn_moe_down-53): 572 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 442 us MUL_MAT_ID(ffn_moe_down-54): 443 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 374 us MUL_MAT_ID(ffn_moe_down-55): 204 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 414 us MUL_MAT_ID(ffn_moe_down-56): 422 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 453 us MUL_MAT_ID(ffn_moe_down-57): 196 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 396 us MUL_MAT_ID(ffn_moe_down-58): 187 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 386 us MUL_MAT_ID(ffn_moe_down-59): 189 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 367 us MUL_MAT_ID(ffn_moe_down-60): 178 us ggml_barrier(...): 334 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 414 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 422 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 413 us MUL_MAT_ID(ffn_moe_down-63): 234 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 420 us MUL_MAT_ID(ffn_moe_down-64): 187 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 521 us MUL_MAT_ID(ffn_moe_down-65): 169 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 388 us MUL_MAT_ID(ffn_moe_down-66): 351 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 452 us MUL_MAT_ID(ffn_moe_down-67): 417 us ggml_barrier(...): 490 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 481 us MUL_MAT_ID(ffn_moe_down-68): 222 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 400 us MUL_MAT_ID(ffn_moe_down-69): 186 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 437 us MUL_MAT_ID(ffn_moe_down-70): 343 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 390 us MUL_MAT_ID(ffn_moe_down-71): 216 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 401 us MUL_MAT_ID(ffn_moe_down-72): 571 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 368 us MUL_MAT_ID(ffn_moe_down-73): 182 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 388 us MUL_MAT_ID(ffn_moe_down-74): 233 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 427 us MUL_MAT_ID(ffn_moe_down-75): 203 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 433 us MUL_MAT_ID(ffn_moe_down-76): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 527 us MUL_MAT_ID(ffn_moe_down-77): 383 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 416 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 429 us MUL_MAT_ID(ffn_moe_down-79): 212 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 452 us MUL_MAT_ID(ffn_moe_down-80): 217 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 405 us MUL_MAT_ID(ffn_moe_down-81): 297 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 417 us MUL_MAT_ID(ffn_moe_down-82): 406 us ggml_barrier(...): 379 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 399 us MUL_MAT_ID(ffn_moe_down-83): 213 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 359 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 400 us MUL_MAT_ID(ffn_moe_down-85): 327 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 372 us MUL_MAT_ID(ffn_moe_down-86): 210 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 428 us MUL_MAT_ID(ffn_moe_down-87): 204 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 421 us MUL_MAT_ID(ffn_moe_down-88): 231 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 420 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 412 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 403 us MUL_MAT_ID(ffn_moe_down-91): 246 us ggml_barrier(...): 230 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 362 us MUL_MAT_ID(ffn_moe_down-25): 177 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 418 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 487 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 408 us MUL_MAT_ID(ffn_moe_down-28): 206 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 598 us MUL_MAT_ID(ffn_moe_down-29): 188 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 447 us MUL_MAT_ID(ffn_moe_down-30): 198 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 413 us MUL_MAT_ID(ffn_moe_down-31): 209 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 385 us MUL_MAT_ID(ffn_moe_down-32): 199 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 410 us MUL_MAT_ID(ffn_moe_down-33): 207 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 389 us MUL_MAT_ID(ffn_moe_down-34): 320 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 366 us MUL_MAT_ID(ffn_moe_down-35): 218 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 419 us MUL_MAT_ID(ffn_moe_down-36): 214 us ggml_barrier(...): 391 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 409 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 413 us MUL_MAT_ID(ffn_moe_down-38): 202 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 392 us MUL_MAT_ID(ffn_moe_down-39): 633 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 353 us MUL_MAT_ID(ffn_moe_down-40): 251 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 470 us MUL_MAT_ID(ffn_moe_down-41): 222 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 376 us MUL_MAT_ID(ffn_moe_down-42): 170 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 355 us MUL_MAT_ID(ffn_moe_down-43): 206 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 390 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 449 us MUL_MAT_ID(ffn_moe_down-45): 339 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 348 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 430 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 408 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 357 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 435 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 427 us MUL_MAT_ID(ffn_moe_down-51): 187 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 409 us MUL_MAT_ID(ffn_moe_down-52): 200 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 408 us MUL_MAT_ID(ffn_moe_down-53): 315 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 435 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 466 us MUL_MAT_ID(ffn_moe_down-55): 448 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 412 us MUL_MAT_ID(ffn_moe_down-56): 234 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 425 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 459 us MUL_MAT_ID(ffn_moe_down-58): 206 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 412 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 479 us MUL_MAT_ID(ffn_moe_down-60): 208 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 416 us MUL_MAT_ID(ffn_moe_down-61): 185 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 454 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 426 us MUL_MAT_ID(ffn_moe_down-63): 595 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 397 us MUL_MAT_ID(ffn_moe_down-64): 204 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 406 us MUL_MAT_ID(ffn_moe_down-65): 200 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 385 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 421 us MUL_MAT_ID(ffn_moe_down-67): 180 us ggml_barrier(...): 341 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 475 us MUL_MAT_ID(ffn_moe_down-68): 316 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 365 us MUL_MAT_ID(ffn_moe_down-69): 198 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 440 us MUL_MAT_ID(ffn_moe_down-70): 386 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 393 us MUL_MAT_ID(ffn_moe_down-71): 364 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 439 us MUL_MAT_ID(ffn_moe_down-72): 220 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 459 us MUL_MAT_ID(ffn_moe_down-73): 211 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 419 us MUL_MAT_ID(ffn_moe_down-74): 206 us ggml_barrier(...): 508 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 406 us MUL_MAT_ID(ffn_moe_down-75): 609 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 458 us MUL_MAT_ID(ffn_moe_down-76): 173 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 389 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 434 us MUL_MAT_ID(ffn_moe_down-78): 371 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 427 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 388 us MUL_MAT_ID(ffn_moe_down-80): 184 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 394 us MUL_MAT_ID(ffn_moe_down-81): 180 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 407 us MUL_MAT_ID(ffn_moe_down-82): 184 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 359 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 430 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 440 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 377 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 399 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 459 us MUL_MAT_ID(ffn_moe_down-88): 193 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 425 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 406 us MUL_MAT_ID(ffn_moe_down-90): 242 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 378 us MUL_MAT_ID(ffn_moe_down-91): 257 us ggml_barrier(...): 29 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 395 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 372 us MUL_MAT_ID(ffn_moe_down-26): 261 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 449 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 442 us MUL_MAT_ID(ffn_moe_down-28): 200 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 461 us MUL_MAT_ID(ffn_moe_down-29): 301 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 418 us MUL_MAT_ID(ffn_moe_down-30): 175 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 422 us MUL_MAT_ID(ffn_moe_down-31): 208 us ggml_barrier(...): 342 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 409 us MUL_MAT_ID(ffn_moe_down-32): 216 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 334 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 425 us MUL_MAT_ID(ffn_moe_down-34): 209 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 440 us MUL_MAT_ID(ffn_moe_down-35): 272 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 406 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 407 us MUL_MAT_ID(ffn_moe_down-37): 246 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 404 us MUL_MAT_ID(ffn_moe_down-38): 261 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 450 us MUL_MAT_ID(ffn_moe_down-39): 206 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 397 us MUL_MAT_ID(ffn_moe_down-40): 244 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 377 us MUL_MAT_ID(ffn_moe_down-41): 191 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 392 us MUL_MAT_ID(ffn_moe_down-42): 187 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 540 us MUL_MAT_ID(ffn_moe_down-43): 218 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 409 us MUL_MAT_ID(ffn_moe_down-44): 246 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 445 us MUL_MAT_ID(ffn_moe_down-45): 196 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 414 us MUL_MAT_ID(ffn_moe_down-46): 296 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 434 us MUL_MAT_ID(ffn_moe_down-47): 226 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 457 us MUL_MAT_ID(ffn_moe_down-48): 200 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 405 us MUL_MAT_ID(ffn_moe_down-49): 498 us ggml_barrier(...): 357 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 420 us MUL_MAT_ID(ffn_moe_down-50): 378 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 431 us MUL_MAT_ID(ffn_moe_down-51): 515 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 425 us MUL_MAT_ID(ffn_moe_down-52): 222 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 502 us MUL_MAT_ID(ffn_moe_down-53): 277 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 350 us MUL_MAT_ID(ffn_moe_down-54): 540 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 385 us MUL_MAT_ID(ffn_moe_down-55): 180 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 408 us MUL_MAT_ID(ffn_moe_down-56): 191 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 433 us MUL_MAT_ID(ffn_moe_down-57): 207 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 408 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 479 us MUL_MAT_ID(ffn_moe_down-59): 181 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 440 us MUL_MAT_ID(ffn_moe_down-60): 441 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 476 us MUL_MAT_ID(ffn_moe_down-61): 314 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 391 us MUL_MAT_ID(ffn_moe_down-62): 225 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 395 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 443 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 417 us MUL_MAT_ID(ffn_moe_down-65): 208 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 403 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 387 us MUL_MAT_ID(ffn_moe_down-67): 187 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 430 us MUL_MAT_ID(ffn_moe_down-68): 229 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 377 us MUL_MAT_ID(ffn_moe_down-69): 183 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 408 us MUL_MAT_ID(ffn_moe_down-70): 193 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 418 us MUL_MAT_ID(ffn_moe_down-71): 231 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 408 us MUL_MAT_ID(ffn_moe_down-72): 193 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 435 us MUL_MAT_ID(ffn_moe_down-73): 436 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 545 us MUL_MAT_ID(ffn_moe_down-74): 235 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 378 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 386 us MUL_MAT_ID(ffn_moe_down-76): 255 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 385 us MUL_MAT_ID(ffn_moe_down-77): 207 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 393 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 405 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 386 us MUL_MAT_ID(ffn_moe_down-80): 211 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 382 us MUL_MAT_ID(ffn_moe_down-81): 227 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 489 us MUL_MAT_ID(ffn_moe_down-82): 204 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 494 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 384 us MUL_MAT_ID(ffn_moe_down-84): 242 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 445 us MUL_MAT_ID(ffn_moe_down-85): 206 us ggml_barrier(...): 386 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 431 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 440 us MUL_MAT_ID(ffn_moe_down-87): 217 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 412 us MUL_MAT_ID(ffn_moe_down-88): 356 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 383 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 450 us MUL_MAT_ID(ffn_moe_down-90): 267 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 380 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 48 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 362 us MUL_MAT_ID(ffn_moe_down-25): 221 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 429 us MUL_MAT_ID(ffn_moe_down-26): 344 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 427 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 479 us MUL_MAT_ID(ffn_moe_down-28): 208 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 398 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 445 us MUL_MAT_ID(ffn_moe_down-30): 221 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 415 us MUL_MAT_ID(ffn_moe_down-31): 208 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 430 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 383 us MUL_MAT_ID(ffn_moe_down-33): 442 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 412 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 384 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 391 us MUL_MAT_ID(ffn_moe_down-36): 208 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 364 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 426 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 401 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 354 us MUL_MAT_ID(ffn_moe_down-40): 240 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 369 us MUL_MAT_ID(ffn_moe_down-41): 203 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 369 us MUL_MAT_ID(ffn_moe_down-42): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 406 us MUL_MAT_ID(ffn_moe_down-43): 186 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 327 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 335 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 378 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 387 us MUL_MAT_ID(ffn_moe_down-47): 312 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 378 us MUL_MAT_ID(ffn_moe_down-48): 184 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 376 us MUL_MAT_ID(ffn_moe_down-49): 641 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 363 us MUL_MAT_ID(ffn_moe_down-50): 199 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 413 us MUL_MAT_ID(ffn_moe_down-51): 404 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 402 us MUL_MAT_ID(ffn_moe_down-52): 299 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 379 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 401 us MUL_MAT_ID(ffn_moe_down-54): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 397 us MUL_MAT_ID(ffn_moe_down-55): 183 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 377 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 394 us MUL_MAT_ID(ffn_moe_down-57): 246 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 374 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 414 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 414 us MUL_MAT_ID(ffn_moe_down-60): 206 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 386 us MUL_MAT_ID(ffn_moe_down-61): 201 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 373 us MUL_MAT_ID(ffn_moe_down-62): 185 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 366 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 479 us MUL_MAT_ID(ffn_moe_down-64): 187 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 406 us MUL_MAT_ID(ffn_moe_down-65): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 377 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 369 us MUL_MAT_ID(ffn_moe_down-67): 196 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 389 us MUL_MAT_ID(ffn_moe_down-68): 188 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 381 us MUL_MAT_ID(ffn_moe_down-69): 204 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 389 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 381 us MUL_MAT_ID(ffn_moe_down-71): 244 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 453 us MUL_MAT_ID(ffn_moe_down-72): 229 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 410 us MUL_MAT_ID(ffn_moe_down-73): 210 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 384 us MUL_MAT_ID(ffn_moe_down-74): 566 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 486 us MUL_MAT_ID(ffn_moe_down-75): 298 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 413 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 446 us MUL_MAT_ID(ffn_moe_down-77): 258 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 453 us MUL_MAT_ID(ffn_moe_down-78): 281 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 396 us MUL_MAT_ID(ffn_moe_down-79): 218 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 426 us MUL_MAT_ID(ffn_moe_down-80): 178 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 379 us MUL_MAT_ID(ffn_moe_down-81): 209 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 389 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 392 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 402 us MUL_MAT_ID(ffn_moe_down-84): 469 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 381 us MUL_MAT_ID(ffn_moe_down-85): 226 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 372 us MUL_MAT_ID(ffn_moe_down-86): 312 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 392 us MUL_MAT_ID(ffn_moe_down-87): 208 us ggml_barrier(...): 372 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 509 us MUL_MAT_ID(ffn_moe_down-88): 187 us ggml_barrier(...): 365 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 435 us MUL_MAT_ID(ffn_moe_down-89): 206 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 540 us MUL_MAT_ID(ffn_moe_down-90): 266 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 404 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 87 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 390 us MUL_MAT_ID(ffn_moe_down-25): 178 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 404 us MUL_MAT_ID(ffn_moe_down-26): 301 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 412 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 361 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 358 us MUL_MAT_ID(ffn_moe_down-29): 480 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 362 us MUL_MAT_ID(ffn_moe_down-30): 212 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 402 us MUL_MAT_ID(ffn_moe_down-31): 529 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 364 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 499 us MUL_MAT_ID(ffn_moe_down-33): 219 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 370 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 382 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 376 us MUL_MAT_ID(ffn_moe_down-36): 259 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 399 us MUL_MAT_ID(ffn_moe_down-37): 221 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 413 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 444 us MUL_MAT_ID(ffn_moe_down-39): 770 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 361 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 363 us MUL_MAT_ID(ffn_moe_down-41): 198 us ggml_barrier(...): 456 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 393 us MUL_MAT_ID(ffn_moe_down-42): 186 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 414 us MUL_MAT_ID(ffn_moe_down-43): 198 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 410 us MUL_MAT_ID(ffn_moe_down-44): 214 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 428 us MUL_MAT_ID(ffn_moe_down-45): 188 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 402 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 391 us MUL_MAT_ID(ffn_moe_down-47): 215 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 406 us MUL_MAT_ID(ffn_moe_down-48): 187 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 335 us MUL_MAT_ID(ffn_moe_down-49): 465 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 408 us MUL_MAT_ID(ffn_moe_down-50): 533 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 405 us MUL_MAT_ID(ffn_moe_down-51): 559 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 391 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 376 us MUL_MAT_ID(ffn_moe_down-53): 198 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 381 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 378 us MUL_MAT_ID(ffn_moe_down-55): 209 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 387 us MUL_MAT_ID(ffn_moe_down-56): 204 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 395 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 423 us MUL_MAT_ID(ffn_moe_down-58): 295 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 397 us MUL_MAT_ID(ffn_moe_down-59): 214 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 378 us MUL_MAT_ID(ffn_moe_down-60): 322 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 408 us MUL_MAT_ID(ffn_moe_down-61): 213 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 384 us MUL_MAT_ID(ffn_moe_down-62): 217 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 400 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 348 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 440 us MUL_MAT_ID(ffn_moe_down-65): 184 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 467 us MUL_MAT_ID(ffn_moe_down-66): 294 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 417 us MUL_MAT_ID(ffn_moe_down-67): 228 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 415 us MUL_MAT_ID(ffn_moe_down-68): 421 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 368 us MUL_MAT_ID(ffn_moe_down-69): 200 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 447 us MUL_MAT_ID(ffn_moe_down-70): 208 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 416 us MUL_MAT_ID(ffn_moe_down-71): 207 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 370 us MUL_MAT_ID(ffn_moe_down-72): 229 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 430 us MUL_MAT_ID(ffn_moe_down-73): 267 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 388 us MUL_MAT_ID(ffn_moe_down-74): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 418 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 447 us MUL_MAT_ID(ffn_moe_down-76): 221 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 377 us MUL_MAT_ID(ffn_moe_down-77): 236 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 408 us MUL_MAT_ID(ffn_moe_down-78): 236 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 397 us MUL_MAT_ID(ffn_moe_down-79): 326 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 385 us MUL_MAT_ID(ffn_moe_down-80): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 402 us MUL_MAT_ID(ffn_moe_down-81): 704 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 366 us MUL_MAT_ID(ffn_moe_down-82): 196 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 372 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 358 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 385 us MUL_MAT_ID(ffn_moe_down-85): 228 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 348 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 412 us MUL_MAT_ID(ffn_moe_down-87): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 393 us MUL_MAT_ID(ffn_moe_down-88): 198 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 397 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 455 us MUL_MAT_ID(ffn_moe_down-90): 285 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 391 us MUL_MAT_ID(ffn_moe_down-91): 678 us ggml_barrier(...): 36 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 426 us MUL_MAT_ID(ffn_moe_down-26): 186 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 364 us MUL_MAT_ID(ffn_moe_down-27): 198 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 379 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 221 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 374 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 405 us MUL_MAT_ID(ffn_moe_down-31): 580 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 381 us MUL_MAT_ID(ffn_moe_down-32): 205 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 413 us MUL_MAT_ID(ffn_moe_down-33): 193 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 403 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 389 us MUL_MAT_ID(ffn_moe_down-35): 213 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 418 us MUL_MAT_ID(ffn_moe_down-36): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 379 us MUL_MAT_ID(ffn_moe_down-37): 250 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 401 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 387 us MUL_MAT_ID(ffn_moe_down-39): 688 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 370 us MUL_MAT_ID(ffn_moe_down-40): 206 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 464 us MUL_MAT_ID(ffn_moe_down-41): 276 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 382 us MUL_MAT_ID(ffn_moe_down-42): 193 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 399 us MUL_MAT_ID(ffn_moe_down-43): 204 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 425 us MUL_MAT_ID(ffn_moe_down-44): 213 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 385 us MUL_MAT_ID(ffn_moe_down-45): 199 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 344 us MUL_MAT_ID(ffn_moe_down-46): 212 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 375 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 397 us MUL_MAT_ID(ffn_moe_down-48): 214 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 394 us MUL_MAT_ID(ffn_moe_down-49): 203 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 378 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 411 us MUL_MAT_ID(ffn_moe_down-51): 672 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 441 us MUL_MAT_ID(ffn_moe_down-52): 177 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 422 us MUL_MAT_ID(ffn_moe_down-53): 432 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 372 us MUL_MAT_ID(ffn_moe_down-54): 266 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 386 us MUL_MAT_ID(ffn_moe_down-55): 585 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 357 us MUL_MAT_ID(ffn_moe_down-56): 396 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 377 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 405 us MUL_MAT_ID(ffn_moe_down-58): 202 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 406 us MUL_MAT_ID(ffn_moe_down-59): 206 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 395 us MUL_MAT_ID(ffn_moe_down-60): 228 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 424 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 377 us MUL_MAT_ID(ffn_moe_down-62): 205 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 381 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 351 us MUL_MAT_ID(ffn_moe_down-64): 568 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 416 us MUL_MAT_ID(ffn_moe_down-65): 228 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 445 us MUL_MAT_ID(ffn_moe_down-66): 225 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 397 us MUL_MAT_ID(ffn_moe_down-67): 397 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 428 us MUL_MAT_ID(ffn_moe_down-68): 206 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 392 us MUL_MAT_ID(ffn_moe_down-69): 402 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 457 us MUL_MAT_ID(ffn_moe_down-70): 212 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 397 us MUL_MAT_ID(ffn_moe_down-71): 211 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 403 us MUL_MAT_ID(ffn_moe_down-72): 211 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 350 us MUL_MAT_ID(ffn_moe_down-73): 221 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 433 us MUL_MAT_ID(ffn_moe_down-74): 213 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 412 us MUL_MAT_ID(ffn_moe_down-75): 204 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 400 us MUL_MAT_ID(ffn_moe_down-76): 245 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 411 us MUL_MAT_ID(ffn_moe_down-78): 191 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 384 us MUL_MAT_ID(ffn_moe_down-79): 205 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 395 us MUL_MAT_ID(ffn_moe_down-80): 224 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 433 us MUL_MAT_ID(ffn_moe_down-81): 203 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 413 us MUL_MAT_ID(ffn_moe_down-82): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 354 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 376 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 362 us MUL_MAT_ID(ffn_moe_down-85): 240 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 407 us MUL_MAT_ID(ffn_moe_down-86): 184 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 398 us MUL_MAT_ID(ffn_moe_down-87): 591 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 380 us MUL_MAT_ID(ffn_moe_down-88): 201 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 385 us MUL_MAT_ID(ffn_moe_down-89): 468 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 371 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 411 us MUL_MAT_ID(ffn_moe_down-91): 229 us ggml_barrier(...): 42 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 356 us MUL_MAT_ID(ffn_moe_down-25): 179 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 416 us MUL_MAT_ID(ffn_moe_down-26): 188 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 369 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 384 us MUL_MAT_ID(ffn_moe_down-28): 262 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 450 us MUL_MAT_ID(ffn_moe_down-29): 188 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 397 us MUL_MAT_ID(ffn_moe_down-30): 198 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 361 us MUL_MAT_ID(ffn_moe_down-31): 186 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 386 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 410 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 401 us MUL_MAT_ID(ffn_moe_down-33): 189 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 385 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 401 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 452 us MUL_MAT_ID(ffn_moe_down-36): 356 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 372 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 368 us MUL_MAT_ID(ffn_moe_down-38): 211 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 382 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 403 us MUL_MAT_ID(ffn_moe_down-40): 522 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 367 us MUL_MAT_ID(ffn_moe_down-41): 339 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 380 us MUL_MAT_ID(ffn_moe_down-42): 402 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 404 us MUL_MAT_ID(ffn_moe_down-43): 214 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 387 us MUL_MAT_ID(ffn_moe_down-44): 187 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 404 us MUL_MAT_ID(ffn_moe_down-45): 388 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 351 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 408 us MUL_MAT_ID(ffn_moe_down-47): 380 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 387 us MUL_MAT_ID(ffn_moe_down-48): 318 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 392 us MUL_MAT_ID(ffn_moe_down-49): 184 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 391 us MUL_MAT_ID(ffn_moe_down-50): 192 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 505 us MUL_MAT_ID(ffn_moe_down-51): 559 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 407 us MUL_MAT_ID(ffn_moe_down-52): 266 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 435 us MUL_MAT_ID(ffn_moe_down-53): 573 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 419 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 459 us MUL_MAT_ID(ffn_moe_down-55): 175 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 406 us MUL_MAT_ID(ffn_moe_down-56): 552 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 393 us MUL_MAT_ID(ffn_moe_down-57): 186 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 403 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 375 us MUL_MAT_ID(ffn_moe_down-59): 211 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 407 us MUL_MAT_ID(ffn_moe_down-60): 195 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 479 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 415 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 365 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 397 us MUL_MAT_ID(ffn_moe_down-64): 184 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 413 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 363 us MUL_MAT_ID(ffn_moe_down-66): 197 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 390 us MUL_MAT_ID(ffn_moe_down-67): 216 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 398 us MUL_MAT_ID(ffn_moe_down-68): 555 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 418 us MUL_MAT_ID(ffn_moe_down-69): 200 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 418 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 374 us MUL_MAT_ID(ffn_moe_down-71): 225 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 441 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 449 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 373 us MUL_MAT_ID(ffn_moe_down-74): 213 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 444 us MUL_MAT_ID(ffn_moe_down-75): 198 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 399 us MUL_MAT_ID(ffn_moe_down-76): 231 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 430 us MUL_MAT_ID(ffn_moe_down-77): 257 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 430 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 388 us MUL_MAT_ID(ffn_moe_down-79): 184 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 384 us MUL_MAT_ID(ffn_moe_down-80): 410 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 420 us MUL_MAT_ID(ffn_moe_down-81): 292 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 493 us MUL_MAT_ID(ffn_moe_down-82): 250 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 402 us MUL_MAT_ID(ffn_moe_down-83): 422 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 399 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 434 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 431 us MUL_MAT_ID(ffn_moe_down-86): 178 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 417 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 408 us MUL_MAT_ID(ffn_moe_down-88): 206 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 413 us MUL_MAT_ID(ffn_moe_down-89): 187 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 375 us MUL_MAT_ID(ffn_moe_down-90): 237 us ggml_barrier(...): 382 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 363 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 54 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 374 us MUL_MAT_ID(ffn_moe_down-25): 168 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 406 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 429 us MUL_MAT_ID(ffn_moe_down-27): 230 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 417 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 384 us MUL_MAT_ID(ffn_moe_down-29): 207 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 415 us MUL_MAT_ID(ffn_moe_down-30): 299 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 398 us MUL_MAT_ID(ffn_moe_down-31): 192 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 378 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 384 us MUL_MAT_ID(ffn_moe_down-33): 197 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 403 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 349 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 379 us MUL_MAT_ID(ffn_moe_down-36): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 391 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 385 us MUL_MAT_ID(ffn_moe_down-38): 233 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 391 us MUL_MAT_ID(ffn_moe_down-39): 239 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 395 us MUL_MAT_ID(ffn_moe_down-40): 708 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 404 us MUL_MAT_ID(ffn_moe_down-41): 432 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 392 us MUL_MAT_ID(ffn_moe_down-42): 192 us ggml_barrier(...): 411 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 502 us MUL_MAT_ID(ffn_moe_down-43): 274 us ggml_barrier(...): 444 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 407 us MUL_MAT_ID(ffn_moe_down-44): 187 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 407 us MUL_MAT_ID(ffn_moe_down-45): 345 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 359 us MUL_MAT_ID(ffn_moe_down-46): 168 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 445 us MUL_MAT_ID(ffn_moe_down-47): 216 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 390 us MUL_MAT_ID(ffn_moe_down-48): 301 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 349 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 380 us MUL_MAT_ID(ffn_moe_down-50): 211 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 388 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 373 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 377 us MUL_MAT_ID(ffn_moe_down-53): 213 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 378 us MUL_MAT_ID(ffn_moe_down-54): 300 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 463 us MUL_MAT_ID(ffn_moe_down-55): 235 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 405 us MUL_MAT_ID(ffn_moe_down-56): 476 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 370 us MUL_MAT_ID(ffn_moe_down-57): 391 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 348 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 394 us MUL_MAT_ID(ffn_moe_down-59): 207 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 562 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 385 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 364 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 359 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 375 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 346 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 401 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 345 us MUL_MAT_ID(ffn_moe_down-67): 209 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 448 us MUL_MAT_ID(ffn_moe_down-68): 227 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 423 us MUL_MAT_ID(ffn_moe_down-69): 218 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 353 us MUL_MAT_ID(ffn_moe_down-70): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 468 us MUL_MAT_ID(ffn_moe_down-71): 191 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 373 us MUL_MAT_ID(ffn_moe_down-73): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 369 us MUL_MAT_ID(ffn_moe_down-74): 203 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 415 us MUL_MAT_ID(ffn_moe_down-75): 207 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 396 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 390 us MUL_MAT_ID(ffn_moe_down-77): 175 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 386 us MUL_MAT_ID(ffn_moe_down-78): 198 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 392 us MUL_MAT_ID(ffn_moe_down-79): 209 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 351 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 395 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 462 us MUL_MAT_ID(ffn_moe_down-82): 179 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 442 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 403 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 410 us MUL_MAT_ID(ffn_moe_down-85): 248 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 369 us MUL_MAT_ID(ffn_moe_down-86): 196 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 463 us MUL_MAT_ID(ffn_moe_down-87): 285 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 385 us MUL_MAT_ID(ffn_moe_down-88): 211 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 379 us MUL_MAT_ID(ffn_moe_down-89): 283 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 402 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 536 us MUL_MAT_ID(ffn_moe_down-91): 418 us ggml_barrier(...): 93 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 392 us MUL_MAT_ID(ffn_moe_down-25): 246 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 414 us MUL_MAT_ID(ffn_moe_down-26): 296 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 404 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 377 us MUL_MAT_ID(ffn_moe_down-28): 221 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 342 us MUL_MAT_ID(ffn_moe_down-29): 271 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 375 us MUL_MAT_ID(ffn_moe_down-30): 190 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 467 us MUL_MAT_ID(ffn_moe_down-31): 207 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 415 us MUL_MAT_ID(ffn_moe_down-32): 276 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 401 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 347 us MUL_MAT_ID(ffn_moe_down-34): 221 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 357 us MUL_MAT_ID(ffn_moe_down-35): 183 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 382 us MUL_MAT_ID(ffn_moe_down-36): 205 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 545 us MUL_MAT_ID(ffn_moe_down-37): 213 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 401 us MUL_MAT_ID(ffn_moe_down-38): 221 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 440 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 501 us MUL_MAT_ID(ffn_moe_down-40): 211 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 392 us MUL_MAT_ID(ffn_moe_down-41): 223 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 373 us MUL_MAT_ID(ffn_moe_down-42): 194 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 376 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 402 us MUL_MAT_ID(ffn_moe_down-44): 218 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 536 us MUL_MAT_ID(ffn_moe_down-45): 305 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 370 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 446 us MUL_MAT_ID(ffn_moe_down-47): 372 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 391 us MUL_MAT_ID(ffn_moe_down-48): 191 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 376 us MUL_MAT_ID(ffn_moe_down-49): 328 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 402 us MUL_MAT_ID(ffn_moe_down-50): 439 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 385 us MUL_MAT_ID(ffn_moe_down-51): 212 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 432 us MUL_MAT_ID(ffn_moe_down-52): 199 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 450 us MUL_MAT_ID(ffn_moe_down-53): 446 us ggml_barrier(...): 544 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 378 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 453 us MUL_MAT_ID(ffn_moe_down-55): 228 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 402 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 446 us MUL_MAT_ID(ffn_moe_down-57): 413 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 395 us MUL_MAT_ID(ffn_moe_down-58): 189 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 408 us MUL_MAT_ID(ffn_moe_down-59): 181 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 444 us MUL_MAT_ID(ffn_moe_down-60): 223 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 454 us MUL_MAT_ID(ffn_moe_down-61): 183 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 492 us MUL_MAT_ID(ffn_moe_down-62): 384 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 396 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 370 us MUL_MAT_ID(ffn_moe_down-64): 226 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 422 us MUL_MAT_ID(ffn_moe_down-65): 477 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 411 us MUL_MAT_ID(ffn_moe_down-66): 176 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 376 us MUL_MAT_ID(ffn_moe_down-67): 265 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 403 us MUL_MAT_ID(ffn_moe_down-68): 219 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 334 us MUL_MAT_ID(ffn_moe_down-69): 243 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 380 us MUL_MAT_ID(ffn_moe_down-70): 377 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 449 us MUL_MAT_ID(ffn_moe_down-71): 202 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 411 us MUL_MAT_ID(ffn_moe_down-72): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 425 us MUL_MAT_ID(ffn_moe_down-73): 202 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 420 us MUL_MAT_ID(ffn_moe_down-74): 191 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 389 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 396 us MUL_MAT_ID(ffn_moe_down-76): 256 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 394 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 456 us MUL_MAT_ID(ffn_moe_down-78): 305 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 419 us MUL_MAT_ID(ffn_moe_down-79): 205 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 357 us MUL_MAT_ID(ffn_moe_down-80): 356 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 485 us MUL_MAT_ID(ffn_moe_down-81): 349 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 405 us MUL_MAT_ID(ffn_moe_down-82): 210 us ggml_barrier(...): 334 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 414 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 442 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 510 us MUL_MAT_ID(ffn_moe_down-85): 242 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 422 us MUL_MAT_ID(ffn_moe_down-86): 556 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 469 us MUL_MAT_ID(ffn_moe_down-87): 209 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 433 us MUL_MAT_ID(ffn_moe_down-88): 187 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 347 us MUL_MAT_ID(ffn_moe_down-89): 206 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 421 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 426 us MUL_MAT_ID(ffn_moe_down-91): 247 us ggml_barrier(...): 15 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 396 us MUL_MAT_ID(ffn_moe_down-25): 619 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 421 us MUL_MAT_ID(ffn_moe_down-26): 309 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 401 us MUL_MAT_ID(ffn_moe_down-27): 300 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 403 us MUL_MAT_ID(ffn_moe_down-28): 218 us ggml_barrier(...): 399 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 404 us MUL_MAT_ID(ffn_moe_down-29): 200 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 516 us MUL_MAT_ID(ffn_moe_down-30): 305 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 417 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 377 us MUL_MAT_ID(ffn_moe_down-32): 251 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 486 us MUL_MAT_ID(ffn_moe_down-33): 208 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 388 us MUL_MAT_ID(ffn_moe_down-34): 219 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 397 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 404 us MUL_MAT_ID(ffn_moe_down-36): 195 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 398 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 389 us MUL_MAT_ID(ffn_moe_down-38): 209 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 449 us MUL_MAT_ID(ffn_moe_down-39): 253 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 390 us MUL_MAT_ID(ffn_moe_down-40): 300 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 366 us MUL_MAT_ID(ffn_moe_down-41): 543 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 391 us MUL_MAT_ID(ffn_moe_down-42): 237 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 430 us MUL_MAT_ID(ffn_moe_down-43): 224 us ggml_barrier(...): 380 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 384 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 433 us MUL_MAT_ID(ffn_moe_down-45): 298 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 410 us MUL_MAT_ID(ffn_moe_down-46): 212 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 446 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 468 us MUL_MAT_ID(ffn_moe_down-48): 234 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 405 us MUL_MAT_ID(ffn_moe_down-49): 355 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 363 us MUL_MAT_ID(ffn_moe_down-50): 197 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 385 us MUL_MAT_ID(ffn_moe_down-51): 300 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 399 us MUL_MAT_ID(ffn_moe_down-52): 374 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 480 us MUL_MAT_ID(ffn_moe_down-53): 241 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 370 us MUL_MAT_ID(ffn_moe_down-54): 229 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 424 us MUL_MAT_ID(ffn_moe_down-55): 186 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 201 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 439 us MUL_MAT_ID(ffn_moe_down-57): 183 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 492 us MUL_MAT_ID(ffn_moe_down-58): 293 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 326 us MUL_MAT_ID(ffn_moe_down-59): 176 us ggml_barrier(...): 507 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 346 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 406 us MUL_MAT_ID(ffn_moe_down-61): 205 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 441 us MUL_MAT_ID(ffn_moe_down-62): 192 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 430 us MUL_MAT_ID(ffn_moe_down-63): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 513 us MUL_MAT_ID(ffn_moe_down-64): 180 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 399 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 363 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 449 us MUL_MAT_ID(ffn_moe_down-67): 226 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 400 us MUL_MAT_ID(ffn_moe_down-68): 214 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 424 us MUL_MAT_ID(ffn_moe_down-69): 207 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 380 us MUL_MAT_ID(ffn_moe_down-70): 197 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 377 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 379 us MUL_MAT_ID(ffn_moe_down-72): 246 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 421 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 362 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 370 us MUL_MAT_ID(ffn_moe_down-75): 208 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 415 us MUL_MAT_ID(ffn_moe_down-76): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 390 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 395 us MUL_MAT_ID(ffn_moe_down-78): 283 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 386 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 384 us MUL_MAT_ID(ffn_moe_down-80): 194 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 418 us MUL_MAT_ID(ffn_moe_down-81): 211 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 354 us MUL_MAT_ID(ffn_moe_down-82): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 369 us MUL_MAT_ID(ffn_moe_down-83): 236 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 416 us MUL_MAT_ID(ffn_moe_down-84): 601 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 397 us MUL_MAT_ID(ffn_moe_down-85): 692 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 412 us MUL_MAT_ID(ffn_moe_down-86): 294 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 396 us MUL_MAT_ID(ffn_moe_down-87): 414 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 358 us MUL_MAT_ID(ffn_moe_down-88): 341 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 383 us MUL_MAT_ID(ffn_moe_down-89): 179 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 371 us MUL_MAT_ID(ffn_moe_down-90): 325 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 402 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 12 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 396 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 372 us MUL_MAT_ID(ffn_moe_down-26): 177 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 479 us MUL_MAT_ID(ffn_moe_down-27): 173 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 424 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 405 us MUL_MAT_ID(ffn_moe_down-29): 202 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 386 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 424 us MUL_MAT_ID(ffn_moe_down-31): 212 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 444 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 432 us MUL_MAT_ID(ffn_moe_down-33): 184 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 371 us MUL_MAT_ID(ffn_moe_down-34): 225 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 410 us MUL_MAT_ID(ffn_moe_down-35): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 386 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 390 us MUL_MAT_ID(ffn_moe_down-38): 204 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 369 us MUL_MAT_ID(ffn_moe_down-39): 361 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 368 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 526 us MUL_MAT_ID(ffn_moe_down-41): 214 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 413 us MUL_MAT_ID(ffn_moe_down-42): 217 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 441 us MUL_MAT_ID(ffn_moe_down-43): 576 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 391 us MUL_MAT_ID(ffn_moe_down-44): 222 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 437 us MUL_MAT_ID(ffn_moe_down-45): 214 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 392 us MUL_MAT_ID(ffn_moe_down-46): 239 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 422 us MUL_MAT_ID(ffn_moe_down-47): 353 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 412 us MUL_MAT_ID(ffn_moe_down-48): 183 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 385 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 407 us MUL_MAT_ID(ffn_moe_down-50): 244 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 383 us MUL_MAT_ID(ffn_moe_down-51): 178 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 455 us MUL_MAT_ID(ffn_moe_down-52): 216 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 387 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 432 us MUL_MAT_ID(ffn_moe_down-54): 213 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 383 us MUL_MAT_ID(ffn_moe_down-55): 649 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 251 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 438 us MUL_MAT_ID(ffn_moe_down-57): 378 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 399 us MUL_MAT_ID(ffn_moe_down-58): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 455 us MUL_MAT_ID(ffn_moe_down-59): 202 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 446 us MUL_MAT_ID(ffn_moe_down-60): 185 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 394 us MUL_MAT_ID(ffn_moe_down-61): 204 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 367 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 439 us MUL_MAT_ID(ffn_moe_down-63): 204 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 515 us MUL_MAT_ID(ffn_moe_down-64): 210 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 371 us MUL_MAT_ID(ffn_moe_down-65): 196 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 436 us MUL_MAT_ID(ffn_moe_down-66): 278 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 364 us MUL_MAT_ID(ffn_moe_down-67): 218 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 389 us MUL_MAT_ID(ffn_moe_down-68): 230 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 438 us MUL_MAT_ID(ffn_moe_down-69): 178 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 411 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 408 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 408 us MUL_MAT_ID(ffn_moe_down-71): 216 us ggml_barrier(...): 360 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 419 us MUL_MAT_ID(ffn_moe_down-72): 311 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 412 us MUL_MAT_ID(ffn_moe_down-73): 254 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 431 us MUL_MAT_ID(ffn_moe_down-74): 224 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 371 us MUL_MAT_ID(ffn_moe_down-75): 194 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 434 us MUL_MAT_ID(ffn_moe_down-76): 187 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 438 us MUL_MAT_ID(ffn_moe_down-77): 174 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 519 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 373 us MUL_MAT_ID(ffn_moe_down-79): 267 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 435 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 417 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 473 us MUL_MAT_ID(ffn_moe_down-82): 238 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 419 us MUL_MAT_ID(ffn_moe_down-83): 250 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 410 us MUL_MAT_ID(ffn_moe_down-84): 248 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 384 us MUL_MAT_ID(ffn_moe_down-85): 248 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 431 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 460 us MUL_MAT_ID(ffn_moe_down-87): 185 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 412 us MUL_MAT_ID(ffn_moe_down-88): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 390 us MUL_MAT_ID(ffn_moe_down-89): 182 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 439 us MUL_MAT_ID(ffn_moe_down-90): 275 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 400 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 39 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 373 us MUL_MAT_ID(ffn_moe_down-25): 183 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 401 us MUL_MAT_ID(ffn_moe_down-26): 280 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 420 us MUL_MAT_ID(ffn_moe_down-27): 227 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 403 us MUL_MAT_ID(ffn_moe_down-28): 389 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 414 us MUL_MAT_ID(ffn_moe_down-29): 186 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 413 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 407 us MUL_MAT_ID(ffn_moe_down-31): 179 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 408 us MUL_MAT_ID(ffn_moe_down-32): 219 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 384 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 493 us MUL_MAT_ID(ffn_moe_down-34): 202 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 464 us MUL_MAT_ID(ffn_moe_down-35): 189 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 370 us MUL_MAT_ID(ffn_moe_down-36): 443 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 371 us MUL_MAT_ID(ffn_moe_down-37): 397 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 377 us MUL_MAT_ID(ffn_moe_down-38): 202 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 422 us MUL_MAT_ID(ffn_moe_down-39): 661 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 386 us MUL_MAT_ID(ffn_moe_down-40): 240 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 437 us MUL_MAT_ID(ffn_moe_down-41): 199 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 436 us MUL_MAT_ID(ffn_moe_down-42): 182 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 481 us MUL_MAT_ID(ffn_moe_down-43): 203 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 417 us MUL_MAT_ID(ffn_moe_down-44): 365 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 420 us MUL_MAT_ID(ffn_moe_down-45): 208 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 483 us MUL_MAT_ID(ffn_moe_down-46): 171 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 444 us MUL_MAT_ID(ffn_moe_down-47): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 359 us MUL_MAT_ID(ffn_moe_down-48): 190 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 399 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 403 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 412 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 395 us MUL_MAT_ID(ffn_moe_down-52): 238 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 393 us MUL_MAT_ID(ffn_moe_down-53): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 379 us MUL_MAT_ID(ffn_moe_down-54): 188 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 401 us MUL_MAT_ID(ffn_moe_down-55): 227 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 396 us MUL_MAT_ID(ffn_moe_down-56): 199 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 397 us MUL_MAT_ID(ffn_moe_down-57): 256 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 414 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 434 us MUL_MAT_ID(ffn_moe_down-59): 225 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 435 us MUL_MAT_ID(ffn_moe_down-60): 252 us ggml_barrier(...): 377 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 403 us MUL_MAT_ID(ffn_moe_down-61): 336 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 400 us MUL_MAT_ID(ffn_moe_down-62): 201 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 389 us MUL_MAT_ID(ffn_moe_down-63): 201 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 392 us MUL_MAT_ID(ffn_moe_down-64): 172 us ggml_barrier(...): 368 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 397 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 383 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 415 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 415 us MUL_MAT_ID(ffn_moe_down-68): 288 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 514 us MUL_MAT_ID(ffn_moe_down-69): 197 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 388 us MUL_MAT_ID(ffn_moe_down-70): 330 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 418 us MUL_MAT_ID(ffn_moe_down-71): 367 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 352 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 397 us MUL_MAT_ID(ffn_moe_down-73): 227 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 453 us MUL_MAT_ID(ffn_moe_down-74): 223 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 442 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 433 us MUL_MAT_ID(ffn_moe_down-76): 312 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 395 us MUL_MAT_ID(ffn_moe_down-77): 204 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 420 us MUL_MAT_ID(ffn_moe_down-78): 305 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 395 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 368 us MUL_MAT_ID(ffn_moe_down-80): 216 us ggml_barrier(...): 438 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 534 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 398 us MUL_MAT_ID(ffn_moe_down-82): 253 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 463 us MUL_MAT_ID(ffn_moe_down-83): 217 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 397 us MUL_MAT_ID(ffn_moe_down-84): 272 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 402 us MUL_MAT_ID(ffn_moe_down-85): 240 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 377 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 430 us MUL_MAT_ID(ffn_moe_down-87): 200 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 392 us MUL_MAT_ID(ffn_moe_down-88): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 379 us MUL_MAT_ID(ffn_moe_down-89): 202 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 415 us MUL_MAT_ID(ffn_moe_down-90): 293 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 419 us MUL_MAT_ID(ffn_moe_down-91): 241 us ggml_barrier(...): 63 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 414 us MUL_MAT_ID(ffn_moe_down-26): 498 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 374 us MUL_MAT_ID(ffn_moe_down-27): 253 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 423 us MUL_MAT_ID(ffn_moe_down-28): 348 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 186 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 336 us MUL_MAT_ID(ffn_moe_down-30): 187 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 403 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 345 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 391 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 370 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 371 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 404 us MUL_MAT_ID(ffn_moe_down-36): 202 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 341 us MUL_MAT_ID(ffn_moe_down-37): 225 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 430 us MUL_MAT_ID(ffn_moe_down-38): 244 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 381 us MUL_MAT_ID(ffn_moe_down-39): 684 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 359 us MUL_MAT_ID(ffn_moe_down-40): 719 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 361 us MUL_MAT_ID(ffn_moe_down-41): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 408 us MUL_MAT_ID(ffn_moe_down-42): 204 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 358 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 344 us MUL_MAT_ID(ffn_moe_down-44): 200 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 425 us MUL_MAT_ID(ffn_moe_down-45): 212 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 371 us MUL_MAT_ID(ffn_moe_down-46): 198 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 390 us MUL_MAT_ID(ffn_moe_down-47): 352 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 385 us MUL_MAT_ID(ffn_moe_down-48): 214 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 399 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 385 us MUL_MAT_ID(ffn_moe_down-50): 197 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 480 us MUL_MAT_ID(ffn_moe_down-51): 210 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 411 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 445 us MUL_MAT_ID(ffn_moe_down-53): 170 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 412 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 431 us MUL_MAT_ID(ffn_moe_down-55): 185 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 444 us MUL_MAT_ID(ffn_moe_down-56): 304 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 380 us MUL_MAT_ID(ffn_moe_down-57): 171 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 414 us MUL_MAT_ID(ffn_moe_down-58): 175 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 341 us MUL_MAT_ID(ffn_moe_down-59): 277 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 436 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 506 us MUL_MAT_ID(ffn_moe_down-61): 408 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 390 us MUL_MAT_ID(ffn_moe_down-62): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 401 us MUL_MAT_ID(ffn_moe_down-63): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 425 us MUL_MAT_ID(ffn_moe_down-64): 296 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 354 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 400 us MUL_MAT_ID(ffn_moe_down-66): 197 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 394 us MUL_MAT_ID(ffn_moe_down-67): 361 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 226 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 421 us MUL_MAT_ID(ffn_moe_down-69): 200 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 460 us MUL_MAT_ID(ffn_moe_down-70): 205 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 449 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 375 us MUL_MAT_ID(ffn_moe_down-72): 469 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 438 us MUL_MAT_ID(ffn_moe_down-73): 203 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 463 us MUL_MAT_ID(ffn_moe_down-74): 229 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 401 us MUL_MAT_ID(ffn_moe_down-75): 288 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 426 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 459 us MUL_MAT_ID(ffn_moe_down-77): 169 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 361 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 430 us MUL_MAT_ID(ffn_moe_down-79): 213 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 430 us MUL_MAT_ID(ffn_moe_down-80): 587 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 443 us MUL_MAT_ID(ffn_moe_down-81): 173 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 438 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 451 us MUL_MAT_ID(ffn_moe_down-83): 188 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 512 us MUL_MAT_ID(ffn_moe_down-84): 248 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 405 us MUL_MAT_ID(ffn_moe_down-85): 219 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 469 us MUL_MAT_ID(ffn_moe_down-86): 590 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 401 us MUL_MAT_ID(ffn_moe_down-87): 170 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 516 us MUL_MAT_ID(ffn_moe_down-88): 185 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 416 us MUL_MAT_ID(ffn_moe_down-89): 444 us ggml_barrier(...): 468 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 383 us MUL_MAT_ID(ffn_moe_down-90): 404 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 412 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 171 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 213 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 456 us MUL_MAT_ID(ffn_moe_down-26): 290 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 392 us MUL_MAT_ID(ffn_moe_down-27): 206 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 393 us MUL_MAT_ID(ffn_moe_down-28): 533 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 403 us MUL_MAT_ID(ffn_moe_down-29): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 349 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 399 us MUL_MAT_ID(ffn_moe_down-31): 197 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 352 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 451 us MUL_MAT_ID(ffn_moe_down-33): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 374 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 361 us MUL_MAT_ID(ffn_moe_down-35): 280 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 411 us MUL_MAT_ID(ffn_moe_down-36): 225 us ggml_barrier(...): 334 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 408 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 384 us MUL_MAT_ID(ffn_moe_down-38): 187 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 387 us MUL_MAT_ID(ffn_moe_down-39): 312 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 424 us MUL_MAT_ID(ffn_moe_down-40): 234 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 379 us MUL_MAT_ID(ffn_moe_down-41): 211 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 406 us MUL_MAT_ID(ffn_moe_down-42): 353 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 392 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 387 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 370 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 359 us MUL_MAT_ID(ffn_moe_down-46): 267 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 427 us MUL_MAT_ID(ffn_moe_down-47): 185 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 383 us MUL_MAT_ID(ffn_moe_down-48): 191 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 371 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 356 us MUL_MAT_ID(ffn_moe_down-50): 189 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 396 us MUL_MAT_ID(ffn_moe_down-51): 207 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 404 us MUL_MAT_ID(ffn_moe_down-52): 274 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 409 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 433 us MUL_MAT_ID(ffn_moe_down-54): 188 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 370 us MUL_MAT_ID(ffn_moe_down-55): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 386 us MUL_MAT_ID(ffn_moe_down-56): 209 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 384 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 407 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 403 us MUL_MAT_ID(ffn_moe_down-59): 208 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 371 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 372 us MUL_MAT_ID(ffn_moe_down-61): 296 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 375 us MUL_MAT_ID(ffn_moe_down-62): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 386 us MUL_MAT_ID(ffn_moe_down-63): 479 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 418 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 379 us MUL_MAT_ID(ffn_moe_down-65): 203 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 370 us MUL_MAT_ID(ffn_moe_down-66): 503 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 417 us MUL_MAT_ID(ffn_moe_down-67): 383 us ggml_barrier(...): 373 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 433 us MUL_MAT_ID(ffn_moe_down-68): 189 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 399 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 537 us MUL_MAT_ID(ffn_moe_down-70): 194 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 401 us MUL_MAT_ID(ffn_moe_down-71): 432 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 365 us MUL_MAT_ID(ffn_moe_down-72): 189 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 385 us MUL_MAT_ID(ffn_moe_down-73): 185 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 342 us MUL_MAT_ID(ffn_moe_down-74): 214 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 420 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 409 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 380 us MUL_MAT_ID(ffn_moe_down-77): 200 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 358 us MUL_MAT_ID(ffn_moe_down-78): 191 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 411 us MUL_MAT_ID(ffn_moe_down-79): 203 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 418 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 437 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 416 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 383 us MUL_MAT_ID(ffn_moe_down-83): 182 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 391 us MUL_MAT_ID(ffn_moe_down-84): 246 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 423 us MUL_MAT_ID(ffn_moe_down-85): 250 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 429 us MUL_MAT_ID(ffn_moe_down-86): 262 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 434 us MUL_MAT_ID(ffn_moe_down-87): 215 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 453 us MUL_MAT_ID(ffn_moe_down-88): 188 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 464 us MUL_MAT_ID(ffn_moe_down-89): 180 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 349 us MUL_MAT_ID(ffn_moe_down-90): 243 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 430 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 44 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 422 us MUL_MAT_ID(ffn_moe_down-25): 233 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 468 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 436 us MUL_MAT_ID(ffn_moe_down-27): 211 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 445 us MUL_MAT_ID(ffn_moe_down-28): 190 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 503 us MUL_MAT_ID(ffn_moe_down-29): 264 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 408 us MUL_MAT_ID(ffn_moe_down-30): 189 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 397 us MUL_MAT_ID(ffn_moe_down-31): 199 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 444 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 405 us MUL_MAT_ID(ffn_moe_down-33): 396 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 399 us MUL_MAT_ID(ffn_moe_down-34): 207 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 400 us MUL_MAT_ID(ffn_moe_down-35): 242 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 424 us MUL_MAT_ID(ffn_moe_down-36): 184 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 402 us MUL_MAT_ID(ffn_moe_down-37): 439 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 391 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 414 us MUL_MAT_ID(ffn_moe_down-39): 419 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 352 us MUL_MAT_ID(ffn_moe_down-40): 557 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 395 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 427 us MUL_MAT_ID(ffn_moe_down-42): 199 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 462 us MUL_MAT_ID(ffn_moe_down-43): 220 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 369 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 363 us MUL_MAT_ID(ffn_moe_down-45): 193 us ggml_barrier(...): 447 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 373 us MUL_MAT_ID(ffn_moe_down-46): 213 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 427 us MUL_MAT_ID(ffn_moe_down-47): 170 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 352 us MUL_MAT_ID(ffn_moe_down-48): 312 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 407 us MUL_MAT_ID(ffn_moe_down-49): 457 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 503 us MUL_MAT_ID(ffn_moe_down-50): 218 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 416 us MUL_MAT_ID(ffn_moe_down-51): 563 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 399 us MUL_MAT_ID(ffn_moe_down-52): 204 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 399 us MUL_MAT_ID(ffn_moe_down-53): 184 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 396 us MUL_MAT_ID(ffn_moe_down-54): 216 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 382 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 445 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 441 us MUL_MAT_ID(ffn_moe_down-57): 184 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 441 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 432 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 499 us MUL_MAT_ID(ffn_moe_down-60): 221 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 403 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 402 us MUL_MAT_ID(ffn_moe_down-62): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 374 us MUL_MAT_ID(ffn_moe_down-63): 202 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 387 us MUL_MAT_ID(ffn_moe_down-64): 550 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 404 us MUL_MAT_ID(ffn_moe_down-65): 206 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 457 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 455 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 460 us MUL_MAT_ID(ffn_moe_down-68): 185 us ggml_barrier(...): 358 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 379 us MUL_MAT_ID(ffn_moe_down-69): 221 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 441 us MUL_MAT_ID(ffn_moe_down-70): 262 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 400 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 360 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 359 us MUL_MAT_ID(ffn_moe_down-74): 203 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 391 us MUL_MAT_ID(ffn_moe_down-75): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 439 us MUL_MAT_ID(ffn_moe_down-76): 263 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 406 us MUL_MAT_ID(ffn_moe_down-77): 209 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 425 us MUL_MAT_ID(ffn_moe_down-78): 226 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 475 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 395 us MUL_MAT_ID(ffn_moe_down-80): 186 us ggml_barrier(...): 371 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 531 us MUL_MAT_ID(ffn_moe_down-81): 266 us ggml_barrier(...): 348 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 417 us MUL_MAT_ID(ffn_moe_down-82): 201 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 413 us MUL_MAT_ID(ffn_moe_down-83): 397 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 375 us MUL_MAT_ID(ffn_moe_down-84): 281 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 387 us MUL_MAT_ID(ffn_moe_down-85): 251 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 473 us MUL_MAT_ID(ffn_moe_down-86): 451 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 416 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 399 us MUL_MAT_ID(ffn_moe_down-88): 201 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 460 us MUL_MAT_ID(ffn_moe_down-89): 407 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 428 us MUL_MAT_ID(ffn_moe_down-90): 261 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 425 us MUL_MAT_ID(ffn_moe_down-91): 187 us ggml_barrier(...): 328 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 374 us MUL_MAT_ID(ffn_moe_down-25): 243 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 475 us MUL_MAT_ID(ffn_moe_down-26): 211 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 502 us MUL_MAT_ID(ffn_moe_down-27): 237 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 441 us MUL_MAT_ID(ffn_moe_down-28): 264 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 511 us MUL_MAT_ID(ffn_moe_down-29): 187 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 433 us MUL_MAT_ID(ffn_moe_down-30): 197 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 411 us MUL_MAT_ID(ffn_moe_down-31): 209 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 364 us MUL_MAT_ID(ffn_moe_down-32): 199 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 546 us MUL_MAT_ID(ffn_moe_down-33): 176 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 420 us MUL_MAT_ID(ffn_moe_down-34): 240 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 406 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 409 us MUL_MAT_ID(ffn_moe_down-36): 240 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 418 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 367 us MUL_MAT_ID(ffn_moe_down-38): 187 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 331 us MUL_MAT_ID(ffn_moe_down-39): 224 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 371 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 388 us MUL_MAT_ID(ffn_moe_down-41): 404 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 399 us MUL_MAT_ID(ffn_moe_down-42): 199 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 363 us MUL_MAT_ID(ffn_moe_down-43): 202 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 341 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 374 us MUL_MAT_ID(ffn_moe_down-45): 212 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 389 us MUL_MAT_ID(ffn_moe_down-46): 256 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 377 us MUL_MAT_ID(ffn_moe_down-47): 337 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 380 us MUL_MAT_ID(ffn_moe_down-48): 209 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 362 us MUL_MAT_ID(ffn_moe_down-49): 431 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 402 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 498 us MUL_MAT_ID(ffn_moe_down-51): 548 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 435 us MUL_MAT_ID(ffn_moe_down-52): 405 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 353 us MUL_MAT_ID(ffn_moe_down-53): 380 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 431 us MUL_MAT_ID(ffn_moe_down-54): 368 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 412 us MUL_MAT_ID(ffn_moe_down-55): 358 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 420 us MUL_MAT_ID(ffn_moe_down-56): 189 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 434 us MUL_MAT_ID(ffn_moe_down-57): 202 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 439 us MUL_MAT_ID(ffn_moe_down-58): 211 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 448 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 356 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 377 us MUL_MAT_ID(ffn_moe_down-60): 189 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 416 us MUL_MAT_ID(ffn_moe_down-61): 199 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 446 us MUL_MAT_ID(ffn_moe_down-62): 263 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 397 us MUL_MAT_ID(ffn_moe_down-63): 199 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 435 us MUL_MAT_ID(ffn_moe_down-64): 369 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 432 us MUL_MAT_ID(ffn_moe_down-65): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 446 us MUL_MAT_ID(ffn_moe_down-66): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 421 us MUL_MAT_ID(ffn_moe_down-67): 202 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 507 us MUL_MAT_ID(ffn_moe_down-68): 434 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 401 us MUL_MAT_ID(ffn_moe_down-69): 405 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 394 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 362 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 387 us MUL_MAT_ID(ffn_moe_down-72): 233 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 418 us MUL_MAT_ID(ffn_moe_down-73): 285 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 550 us MUL_MAT_ID(ffn_moe_down-74): 224 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 392 us MUL_MAT_ID(ffn_moe_down-75): 191 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 441 us MUL_MAT_ID(ffn_moe_down-76): 197 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 534 us MUL_MAT_ID(ffn_moe_down-77): 203 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 402 us MUL_MAT_ID(ffn_moe_down-78): 190 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 418 us MUL_MAT_ID(ffn_moe_down-79): 203 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 404 us MUL_MAT_ID(ffn_moe_down-80): 247 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 439 us MUL_MAT_ID(ffn_moe_down-81): 185 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 423 us MUL_MAT_ID(ffn_moe_down-82): 189 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 471 us MUL_MAT_ID(ffn_moe_down-83): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 393 us MUL_MAT_ID(ffn_moe_down-84): 224 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 387 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 383 us MUL_MAT_ID(ffn_moe_down-86): 181 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 386 us MUL_MAT_ID(ffn_moe_down-87): 228 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 521 us MUL_MAT_ID(ffn_moe_down-88): 215 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 400 us MUL_MAT_ID(ffn_moe_down-89): 196 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 456 us MUL_MAT_ID(ffn_moe_down-90): 240 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 425 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 23 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 387 us MUL_MAT_ID(ffn_moe_down-25): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 433 us MUL_MAT_ID(ffn_moe_down-26): 186 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 397 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 347 us MUL_MAT_ID(ffn_moe_down-28): 205 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 390 us MUL_MAT_ID(ffn_moe_down-29): 212 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 388 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 421 us MUL_MAT_ID(ffn_moe_down-31): 186 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 402 us MUL_MAT_ID(ffn_moe_down-32): 185 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 405 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 445 us MUL_MAT_ID(ffn_moe_down-34): 213 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 400 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 453 us MUL_MAT_ID(ffn_moe_down-36): 203 us ggml_barrier(...): 525 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 393 us MUL_MAT_ID(ffn_moe_down-37): 221 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 405 us MUL_MAT_ID(ffn_moe_down-38): 197 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 421 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 366 us MUL_MAT_ID(ffn_moe_down-40): 240 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 414 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 373 us MUL_MAT_ID(ffn_moe_down-42): 212 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 369 us MUL_MAT_ID(ffn_moe_down-43): 588 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 419 us MUL_MAT_ID(ffn_moe_down-44): 186 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 452 us MUL_MAT_ID(ffn_moe_down-45): 202 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 371 us MUL_MAT_ID(ffn_moe_down-46): 208 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 391 us MUL_MAT_ID(ffn_moe_down-47): 261 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 388 us MUL_MAT_ID(ffn_moe_down-48): 197 us ggml_barrier(...): 407 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 427 us MUL_MAT_ID(ffn_moe_down-49): 202 us ggml_barrier(...): 345 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 380 us MUL_MAT_ID(ffn_moe_down-50): 205 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 453 us MUL_MAT_ID(ffn_moe_down-51): 187 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 419 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 381 us MUL_MAT_ID(ffn_moe_down-53): 198 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 466 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 365 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 414 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 439 us MUL_MAT_ID(ffn_moe_down-57): 202 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 430 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 426 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 387 us MUL_MAT_ID(ffn_moe_down-60): 162 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 477 us MUL_MAT_ID(ffn_moe_down-61): 237 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 374 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 416 us MUL_MAT_ID(ffn_moe_down-63): 224 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 357 us MUL_MAT_ID(ffn_moe_down-64): 182 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 398 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 425 us MUL_MAT_ID(ffn_moe_down-66): 209 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 420 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 402 us MUL_MAT_ID(ffn_moe_down-68): 229 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 475 us MUL_MAT_ID(ffn_moe_down-69): 221 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 384 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 474 us MUL_MAT_ID(ffn_moe_down-71): 406 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 504 us MUL_MAT_ID(ffn_moe_down-72): 187 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 440 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 418 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 338 us MUL_MAT_ID(ffn_moe_down-75): 530 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 343 us MUL_MAT_ID(ffn_moe_down-76): 231 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 427 us MUL_MAT_ID(ffn_moe_down-77): 196 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 406 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 376 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 387 us MUL_MAT_ID(ffn_moe_down-80): 189 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 218 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 467 us MUL_MAT_ID(ffn_moe_down-82): 225 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 406 us MUL_MAT_ID(ffn_moe_down-83): 600 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 360 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 481 us MUL_MAT_ID(ffn_moe_down-85): 258 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 410 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 389 us MUL_MAT_ID(ffn_moe_down-87): 199 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 393 us MUL_MAT_ID(ffn_moe_down-88): 443 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 402 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 413 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 411 us MUL_MAT_ID(ffn_moe_down-91): 247 us ggml_barrier(...): 35 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 391 us MUL_MAT_ID(ffn_moe_down-25): 216 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 424 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 442 us MUL_MAT_ID(ffn_moe_down-27): 213 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 406 us MUL_MAT_ID(ffn_moe_down-28): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 250 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 418 us MUL_MAT_ID(ffn_moe_down-30): 561 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 363 us MUL_MAT_ID(ffn_moe_down-31): 172 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 412 us MUL_MAT_ID(ffn_moe_down-32): 235 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 389 us MUL_MAT_ID(ffn_moe_down-33): 352 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 417 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 412 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 359 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 398 us MUL_MAT_ID(ffn_moe_down-37): 269 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 427 us MUL_MAT_ID(ffn_moe_down-39): 638 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 399 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 343 us MUL_MAT_ID(ffn_moe_down-41): 211 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 395 us MUL_MAT_ID(ffn_moe_down-42): 179 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 407 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 394 us MUL_MAT_ID(ffn_moe_down-44): 200 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 336 us MUL_MAT_ID(ffn_moe_down-45): 179 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 471 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 390 us MUL_MAT_ID(ffn_moe_down-47): 197 us ggml_barrier(...): 529 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 401 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 405 us MUL_MAT_ID(ffn_moe_down-49): 182 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 389 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 400 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 435 us MUL_MAT_ID(ffn_moe_down-52): 202 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 396 us MUL_MAT_ID(ffn_moe_down-53): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 373 us MUL_MAT_ID(ffn_moe_down-54): 244 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 409 us MUL_MAT_ID(ffn_moe_down-55): 204 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 405 us MUL_MAT_ID(ffn_moe_down-56): 220 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 424 us MUL_MAT_ID(ffn_moe_down-57): 210 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 400 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 393 us MUL_MAT_ID(ffn_moe_down-59): 214 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 424 us MUL_MAT_ID(ffn_moe_down-60): 189 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 436 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 361 us MUL_MAT_ID(ffn_moe_down-62): 201 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 385 us MUL_MAT_ID(ffn_moe_down-63): 374 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 361 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 410 us MUL_MAT_ID(ffn_moe_down-65): 389 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 440 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 478 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 446 us MUL_MAT_ID(ffn_moe_down-68): 215 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 408 us MUL_MAT_ID(ffn_moe_down-69): 254 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 392 us MUL_MAT_ID(ffn_moe_down-70): 211 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 411 us MUL_MAT_ID(ffn_moe_down-71): 264 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 433 us MUL_MAT_ID(ffn_moe_down-72): 178 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 423 us MUL_MAT_ID(ffn_moe_down-73): 196 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 386 us MUL_MAT_ID(ffn_moe_down-74): 297 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 380 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 382 us MUL_MAT_ID(ffn_moe_down-76): 389 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 443 us MUL_MAT_ID(ffn_moe_down-77): 222 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 393 us MUL_MAT_ID(ffn_moe_down-78): 214 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 428 us MUL_MAT_ID(ffn_moe_down-79): 182 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 429 us MUL_MAT_ID(ffn_moe_down-80): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 404 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 345 us MUL_MAT_ID(ffn_moe_down-82): 190 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 410 us MUL_MAT_ID(ffn_moe_down-83): 208 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 390 us MUL_MAT_ID(ffn_moe_down-84): 365 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 433 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 403 us MUL_MAT_ID(ffn_moe_down-86): 581 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 392 us MUL_MAT_ID(ffn_moe_down-87): 213 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 417 us MUL_MAT_ID(ffn_moe_down-88): 198 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 397 us MUL_MAT_ID(ffn_moe_down-89): 191 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 404 us MUL_MAT_ID(ffn_moe_down-90): 237 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 511 us MUL_MAT_ID(ffn_moe_down-91): 258 us ggml_barrier(...): 159 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 399 us MUL_MAT_ID(ffn_moe_down-25): 219 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 506 us MUL_MAT_ID(ffn_moe_down-26): 169 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 439 us MUL_MAT_ID(ffn_moe_down-27): 182 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 462 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 386 us MUL_MAT_ID(ffn_moe_down-29): 186 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 392 us MUL_MAT_ID(ffn_moe_down-30): 174 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 484 us MUL_MAT_ID(ffn_moe_down-31): 533 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 404 us MUL_MAT_ID(ffn_moe_down-32): 201 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 465 us MUL_MAT_ID(ffn_moe_down-33): 210 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 417 us MUL_MAT_ID(ffn_moe_down-34): 219 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 424 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 385 us MUL_MAT_ID(ffn_moe_down-36): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 433 us MUL_MAT_ID(ffn_moe_down-37): 239 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 392 us MUL_MAT_ID(ffn_moe_down-38): 196 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 437 us MUL_MAT_ID(ffn_moe_down-39): 239 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 424 us MUL_MAT_ID(ffn_moe_down-40): 248 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 407 us MUL_MAT_ID(ffn_moe_down-41): 198 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 458 us MUL_MAT_ID(ffn_moe_down-42): 285 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 424 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 414 us MUL_MAT_ID(ffn_moe_down-44): 366 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 380 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 414 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 421 us MUL_MAT_ID(ffn_moe_down-47): 416 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 401 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 482 us MUL_MAT_ID(ffn_moe_down-49): 340 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 420 us MUL_MAT_ID(ffn_moe_down-50): 208 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 463 us MUL_MAT_ID(ffn_moe_down-51): 266 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 441 us MUL_MAT_ID(ffn_moe_down-52): 240 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 406 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 423 us MUL_MAT_ID(ffn_moe_down-54): 192 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 345 us MUL_MAT_ID(ffn_moe_down-55): 203 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 466 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 398 us MUL_MAT_ID(ffn_moe_down-57): 196 us ggml_barrier(...): 369 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 426 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 399 us MUL_MAT_ID(ffn_moe_down-59): 226 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 388 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 484 us MUL_MAT_ID(ffn_moe_down-61): 201 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 419 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 403 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 413 us MUL_MAT_ID(ffn_moe_down-64): 206 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 390 us MUL_MAT_ID(ffn_moe_down-65): 202 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 429 us MUL_MAT_ID(ffn_moe_down-66): 202 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 415 us MUL_MAT_ID(ffn_moe_down-67): 188 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 220 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 417 us MUL_MAT_ID(ffn_moe_down-69): 227 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 402 us MUL_MAT_ID(ffn_moe_down-70): 321 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 426 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 259 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 366 us MUL_MAT_ID(ffn_moe_down-74): 218 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 447 us MUL_MAT_ID(ffn_moe_down-75): 212 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 397 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 398 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 364 us MUL_MAT_ID(ffn_moe_down-78): 191 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 386 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 339 us MUL_MAT_ID(ffn_moe_down-80): 182 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 389 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 375 us MUL_MAT_ID(ffn_moe_down-82): 205 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 411 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 428 us MUL_MAT_ID(ffn_moe_down-84): 242 us ggml_barrier(...): 368 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 371 us MUL_MAT_ID(ffn_moe_down-85): 240 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 397 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 396 us MUL_MAT_ID(ffn_moe_down-87): 358 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 400 us MUL_MAT_ID(ffn_moe_down-88): 197 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 408 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 430 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 402 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 36 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 376 us MUL_MAT_ID(ffn_moe_down-25): 181 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 453 us MUL_MAT_ID(ffn_moe_down-26): 338 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 418 us MUL_MAT_ID(ffn_moe_down-27): 285 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 375 us MUL_MAT_ID(ffn_moe_down-28): 207 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 447 us MUL_MAT_ID(ffn_moe_down-29): 181 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 369 us MUL_MAT_ID(ffn_moe_down-30): 290 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 434 us MUL_MAT_ID(ffn_moe_down-31): 190 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 418 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 428 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 365 us MUL_MAT_ID(ffn_moe_down-34): 214 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 435 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 416 us MUL_MAT_ID(ffn_moe_down-36): 189 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 537 us MUL_MAT_ID(ffn_moe_down-37): 290 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 404 us MUL_MAT_ID(ffn_moe_down-38): 182 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 517 us MUL_MAT_ID(ffn_moe_down-39): 384 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 395 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 365 us MUL_MAT_ID(ffn_moe_down-41): 191 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 408 us MUL_MAT_ID(ffn_moe_down-42): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 404 us MUL_MAT_ID(ffn_moe_down-43): 197 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 433 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 407 us MUL_MAT_ID(ffn_moe_down-45): 417 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 353 us MUL_MAT_ID(ffn_moe_down-46): 205 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 379 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 399 us MUL_MAT_ID(ffn_moe_down-48): 208 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 366 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 339 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 420 us MUL_MAT_ID(ffn_moe_down-50): 216 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 472 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 518 us MUL_MAT_ID(ffn_moe_down-52): 264 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 431 us MUL_MAT_ID(ffn_moe_down-53): 200 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 397 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 390 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 187 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 374 us MUL_MAT_ID(ffn_moe_down-57): 216 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 365 us MUL_MAT_ID(ffn_moe_down-58): 233 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 362 us MUL_MAT_ID(ffn_moe_down-59): 302 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 514 us MUL_MAT_ID(ffn_moe_down-60): 224 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 476 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 382 us MUL_MAT_ID(ffn_moe_down-62): 204 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 448 us MUL_MAT_ID(ffn_moe_down-63): 300 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 463 us MUL_MAT_ID(ffn_moe_down-64): 209 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 428 us MUL_MAT_ID(ffn_moe_down-65): 204 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 393 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 349 us MUL_MAT_ID(ffn_moe_down-67): 431 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 534 us MUL_MAT_ID(ffn_moe_down-68): 220 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 384 us MUL_MAT_ID(ffn_moe_down-69): 197 us ggml_barrier(...): 346 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 388 us MUL_MAT_ID(ffn_moe_down-70): 297 us ggml_barrier(...): 372 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 451 us MUL_MAT_ID(ffn_moe_down-71): 312 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 381 us MUL_MAT_ID(ffn_moe_down-72): 182 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 473 us MUL_MAT_ID(ffn_moe_down-73): 208 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 405 us MUL_MAT_ID(ffn_moe_down-74): 187 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 483 us MUL_MAT_ID(ffn_moe_down-75): 304 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 409 us MUL_MAT_ID(ffn_moe_down-76): 198 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 424 us MUL_MAT_ID(ffn_moe_down-77): 205 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 451 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 365 us MUL_MAT_ID(ffn_moe_down-79): 189 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 386 us MUL_MAT_ID(ffn_moe_down-80): 202 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 435 us MUL_MAT_ID(ffn_moe_down-81): 210 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 433 us MUL_MAT_ID(ffn_moe_down-82): 278 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 386 us MUL_MAT_ID(ffn_moe_down-83): 321 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 506 us MUL_MAT_ID(ffn_moe_down-84): 270 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 373 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 398 us MUL_MAT_ID(ffn_moe_down-86): 259 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 412 us MUL_MAT_ID(ffn_moe_down-87): 201 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 377 us MUL_MAT_ID(ffn_moe_down-88): 216 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 452 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 400 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 395 us MUL_MAT_ID(ffn_moe_down-91): 664 us ggml_barrier(...): 43 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 397 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 401 us MUL_MAT_ID(ffn_moe_down-26): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 348 us MUL_MAT_ID(ffn_moe_down-27): 287 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 413 us MUL_MAT_ID(ffn_moe_down-28): 196 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 413 us MUL_MAT_ID(ffn_moe_down-29): 207 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 413 us MUL_MAT_ID(ffn_moe_down-30): 481 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 385 us MUL_MAT_ID(ffn_moe_down-31): 228 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 399 us MUL_MAT_ID(ffn_moe_down-32): 201 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 402 us MUL_MAT_ID(ffn_moe_down-33): 184 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 420 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 382 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 364 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 391 us MUL_MAT_ID(ffn_moe_down-37): 251 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 405 us MUL_MAT_ID(ffn_moe_down-38): 610 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 394 us MUL_MAT_ID(ffn_moe_down-39): 656 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 353 us MUL_MAT_ID(ffn_moe_down-40): 225 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 403 us MUL_MAT_ID(ffn_moe_down-41): 187 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 415 us MUL_MAT_ID(ffn_moe_down-42): 214 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 388 us MUL_MAT_ID(ffn_moe_down-43): 198 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 353 us MUL_MAT_ID(ffn_moe_down-44): 186 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 376 us MUL_MAT_ID(ffn_moe_down-45): 190 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 412 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 451 us MUL_MAT_ID(ffn_moe_down-47): 186 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 400 us MUL_MAT_ID(ffn_moe_down-48): 217 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 394 us MUL_MAT_ID(ffn_moe_down-49): 214 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 421 us MUL_MAT_ID(ffn_moe_down-50): 366 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 411 us MUL_MAT_ID(ffn_moe_down-51): 204 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 449 us MUL_MAT_ID(ffn_moe_down-52): 368 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 408 us MUL_MAT_ID(ffn_moe_down-53): 204 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 409 us MUL_MAT_ID(ffn_moe_down-54): 202 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 434 us MUL_MAT_ID(ffn_moe_down-55): 316 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 423 us MUL_MAT_ID(ffn_moe_down-56): 205 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 454 us MUL_MAT_ID(ffn_moe_down-57): 218 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 455 us MUL_MAT_ID(ffn_moe_down-58): 207 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 442 us MUL_MAT_ID(ffn_moe_down-59): 318 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 400 us MUL_MAT_ID(ffn_moe_down-60): 187 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 443 us MUL_MAT_ID(ffn_moe_down-61): 196 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 336 us MUL_MAT_ID(ffn_moe_down-62): 207 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 403 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 442 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 378 us MUL_MAT_ID(ffn_moe_down-65): 249 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 386 us MUL_MAT_ID(ffn_moe_down-66): 204 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 420 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 456 us MUL_MAT_ID(ffn_moe_down-68): 292 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 432 us MUL_MAT_ID(ffn_moe_down-69): 205 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 394 us MUL_MAT_ID(ffn_moe_down-70): 194 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 432 us MUL_MAT_ID(ffn_moe_down-71): 558 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 375 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 465 us MUL_MAT_ID(ffn_moe_down-73): 418 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 389 us MUL_MAT_ID(ffn_moe_down-74): 409 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 389 us MUL_MAT_ID(ffn_moe_down-75): 212 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 466 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 448 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 381 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 460 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 361 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 430 us MUL_MAT_ID(ffn_moe_down-80): 280 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 366 us MUL_MAT_ID(ffn_moe_down-81): 220 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 397 us MUL_MAT_ID(ffn_moe_down-82): 194 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 395 us MUL_MAT_ID(ffn_moe_down-83): 201 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 418 us MUL_MAT_ID(ffn_moe_down-84): 299 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 420 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 395 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 466 us MUL_MAT_ID(ffn_moe_down-87): 220 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 435 us MUL_MAT_ID(ffn_moe_down-88): 247 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 407 us MUL_MAT_ID(ffn_moe_down-89): 185 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 450 us MUL_MAT_ID(ffn_moe_down-90): 357 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 394 us MUL_MAT_ID(ffn_moe_down-91): 457 us ggml_barrier(...): 20 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 519 us MUL_MAT_ID(ffn_moe_down-26): 221 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 401 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 348 us MUL_MAT_ID(ffn_moe_down-28): 561 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 365 us MUL_MAT_ID(ffn_moe_down-29): 183 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 418 us MUL_MAT_ID(ffn_moe_down-30): 200 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 489 us MUL_MAT_ID(ffn_moe_down-31): 203 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 408 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 390 us MUL_MAT_ID(ffn_moe_down-33): 260 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 467 us MUL_MAT_ID(ffn_moe_down-34): 243 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 373 us MUL_MAT_ID(ffn_moe_down-35): 391 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 479 us MUL_MAT_ID(ffn_moe_down-36): 245 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 394 us MUL_MAT_ID(ffn_moe_down-37): 284 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 431 us MUL_MAT_ID(ffn_moe_down-38): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 347 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 427 us MUL_MAT_ID(ffn_moe_down-40): 241 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 381 us MUL_MAT_ID(ffn_moe_down-41): 190 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 383 us MUL_MAT_ID(ffn_moe_down-42): 233 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 396 us MUL_MAT_ID(ffn_moe_down-43): 210 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 451 us MUL_MAT_ID(ffn_moe_down-44): 305 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 402 us MUL_MAT_ID(ffn_moe_down-45): 198 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 379 us MUL_MAT_ID(ffn_moe_down-46): 238 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 373 us MUL_MAT_ID(ffn_moe_down-47): 197 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 360 us MUL_MAT_ID(ffn_moe_down-48): 187 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 439 us MUL_MAT_ID(ffn_moe_down-49): 206 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 379 us MUL_MAT_ID(ffn_moe_down-50): 187 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 466 us MUL_MAT_ID(ffn_moe_down-51): 238 us ggml_barrier(...): 338 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 384 us MUL_MAT_ID(ffn_moe_down-52): 210 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 447 us MUL_MAT_ID(ffn_moe_down-53): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 426 us MUL_MAT_ID(ffn_moe_down-54): 204 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 352 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 520 us MUL_MAT_ID(ffn_moe_down-56): 295 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 408 us MUL_MAT_ID(ffn_moe_down-57): 189 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 535 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 416 us MUL_MAT_ID(ffn_moe_down-59): 228 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 387 us MUL_MAT_ID(ffn_moe_down-60): 201 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 359 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 374 us MUL_MAT_ID(ffn_moe_down-62): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 354 us MUL_MAT_ID(ffn_moe_down-63): 206 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 408 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 447 us MUL_MAT_ID(ffn_moe_down-65): 299 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 384 us MUL_MAT_ID(ffn_moe_down-66): 238 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 389 us MUL_MAT_ID(ffn_moe_down-67): 177 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 390 us MUL_MAT_ID(ffn_moe_down-68): 211 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 420 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 357 us MUL_MAT_ID(ffn_moe_down-70): 214 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 424 us MUL_MAT_ID(ffn_moe_down-71): 240 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 441 us MUL_MAT_ID(ffn_moe_down-72): 295 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 417 us MUL_MAT_ID(ffn_moe_down-73): 208 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 351 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 363 us MUL_MAT_ID(ffn_moe_down-75): 213 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 412 us MUL_MAT_ID(ffn_moe_down-76): 218 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 420 us MUL_MAT_ID(ffn_moe_down-77): 196 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 396 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 418 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 406 us MUL_MAT_ID(ffn_moe_down-80): 201 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 401 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 410 us MUL_MAT_ID(ffn_moe_down-82): 190 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 462 us MUL_MAT_ID(ffn_moe_down-83): 296 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 396 us MUL_MAT_ID(ffn_moe_down-84): 240 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 383 us MUL_MAT_ID(ffn_moe_down-85): 708 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 401 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 384 us MUL_MAT_ID(ffn_moe_down-87): 184 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 405 us MUL_MAT_ID(ffn_moe_down-88): 214 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 369 us MUL_MAT_ID(ffn_moe_down-89): 443 us ggml_barrier(...): 461 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 368 us MUL_MAT_ID(ffn_moe_down-90): 229 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 399 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 148 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 392 us MUL_MAT_ID(ffn_moe_down-25): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 394 us MUL_MAT_ID(ffn_moe_down-26): 218 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 369 us MUL_MAT_ID(ffn_moe_down-27): 216 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 191 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 366 us MUL_MAT_ID(ffn_moe_down-29): 380 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 391 us MUL_MAT_ID(ffn_moe_down-30): 234 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 417 us MUL_MAT_ID(ffn_moe_down-31): 224 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 356 us MUL_MAT_ID(ffn_moe_down-32): 199 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 430 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 440 us MUL_MAT_ID(ffn_moe_down-34): 177 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 392 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 347 us MUL_MAT_ID(ffn_moe_down-36): 188 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 369 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 388 us MUL_MAT_ID(ffn_moe_down-38): 186 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 352 us MUL_MAT_ID(ffn_moe_down-39): 243 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 352 us MUL_MAT_ID(ffn_moe_down-40): 235 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 386 us MUL_MAT_ID(ffn_moe_down-41): 581 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 346 us MUL_MAT_ID(ffn_moe_down-42): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 386 us MUL_MAT_ID(ffn_moe_down-43): 380 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 412 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 483 us MUL_MAT_ID(ffn_moe_down-45): 316 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 431 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 425 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 420 us MUL_MAT_ID(ffn_moe_down-48): 207 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 391 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 422 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 421 us MUL_MAT_ID(ffn_moe_down-51): 202 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 418 us MUL_MAT_ID(ffn_moe_down-52): 496 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 369 us MUL_MAT_ID(ffn_moe_down-53): 456 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 335 us MUL_MAT_ID(ffn_moe_down-54): 204 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 377 us MUL_MAT_ID(ffn_moe_down-55): 295 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 183 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 376 us MUL_MAT_ID(ffn_moe_down-57): 178 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 391 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 349 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 389 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 392 us MUL_MAT_ID(ffn_moe_down-61): 216 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 399 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 382 us MUL_MAT_ID(ffn_moe_down-63): 215 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 383 us MUL_MAT_ID(ffn_moe_down-64): 507 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 388 us MUL_MAT_ID(ffn_moe_down-65): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 400 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 346 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 376 us MUL_MAT_ID(ffn_moe_down-68): 263 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 369 us MUL_MAT_ID(ffn_moe_down-69): 619 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 368 us MUL_MAT_ID(ffn_moe_down-70): 401 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 372 us MUL_MAT_ID(ffn_moe_down-71): 207 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 387 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 347 us MUL_MAT_ID(ffn_moe_down-73): 203 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 358 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 438 us MUL_MAT_ID(ffn_moe_down-75): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 347 us MUL_MAT_ID(ffn_moe_down-76): 434 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 367 us MUL_MAT_ID(ffn_moe_down-77): 197 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 450 us MUL_MAT_ID(ffn_moe_down-78): 253 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 403 us MUL_MAT_ID(ffn_moe_down-79): 340 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 370 us MUL_MAT_ID(ffn_moe_down-80): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 403 us MUL_MAT_ID(ffn_moe_down-81): 209 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 393 us MUL_MAT_ID(ffn_moe_down-82): 194 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 410 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 358 us MUL_MAT_ID(ffn_moe_down-84): 459 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 363 us MUL_MAT_ID(ffn_moe_down-85): 355 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 357 us MUL_MAT_ID(ffn_moe_down-86): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 379 us MUL_MAT_ID(ffn_moe_down-87): 206 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 408 us MUL_MAT_ID(ffn_moe_down-88): 448 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 409 us MUL_MAT_ID(ffn_moe_down-89): 427 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 348 us MUL_MAT_ID(ffn_moe_down-90): 656 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 383 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 335 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 435 us MUL_MAT_ID(ffn_moe_down-25): 450 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 371 us MUL_MAT_ID(ffn_moe_down-26): 197 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 402 us MUL_MAT_ID(ffn_moe_down-27): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 345 us MUL_MAT_ID(ffn_moe_down-28): 196 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 521 us MUL_MAT_ID(ffn_moe_down-29): 322 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 425 us MUL_MAT_ID(ffn_moe_down-30): 344 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 374 us MUL_MAT_ID(ffn_moe_down-31): 227 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 425 us MUL_MAT_ID(ffn_moe_down-32): 201 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 389 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 458 us MUL_MAT_ID(ffn_moe_down-35): 325 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 376 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 467 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 377 us MUL_MAT_ID(ffn_moe_down-38): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 364 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 363 us MUL_MAT_ID(ffn_moe_down-40): 305 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 394 us MUL_MAT_ID(ffn_moe_down-41): 426 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 393 us MUL_MAT_ID(ffn_moe_down-42): 465 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 338 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 399 us MUL_MAT_ID(ffn_moe_down-44): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 353 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 377 us MUL_MAT_ID(ffn_moe_down-46): 567 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 394 us MUL_MAT_ID(ffn_moe_down-47): 210 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 393 us MUL_MAT_ID(ffn_moe_down-48): 211 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 372 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 461 us MUL_MAT_ID(ffn_moe_down-50): 334 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 360 us MUL_MAT_ID(ffn_moe_down-51): 441 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 388 us MUL_MAT_ID(ffn_moe_down-52): 202 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 390 us MUL_MAT_ID(ffn_moe_down-53): 343 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 421 us MUL_MAT_ID(ffn_moe_down-54): 231 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 403 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 420 us MUL_MAT_ID(ffn_moe_down-56): 228 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 411 us MUL_MAT_ID(ffn_moe_down-57): 186 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 382 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 478 us MUL_MAT_ID(ffn_moe_down-59): 175 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 457 us MUL_MAT_ID(ffn_moe_down-60): 193 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 397 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 400 us MUL_MAT_ID(ffn_moe_down-62): 233 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 405 us MUL_MAT_ID(ffn_moe_down-63): 198 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 384 us MUL_MAT_ID(ffn_moe_down-64): 206 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 366 us MUL_MAT_ID(ffn_moe_down-65): 432 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 409 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 358 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 387 us MUL_MAT_ID(ffn_moe_down-68): 204 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 397 us MUL_MAT_ID(ffn_moe_down-69): 210 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 367 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 392 us MUL_MAT_ID(ffn_moe_down-71): 201 us ggml_barrier(...): 398 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 415 us MUL_MAT_ID(ffn_moe_down-72): 223 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 372 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 329 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 398 us MUL_MAT_ID(ffn_moe_down-75): 217 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 448 us MUL_MAT_ID(ffn_moe_down-76): 534 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 385 us MUL_MAT_ID(ffn_moe_down-77): 528 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 379 us MUL_MAT_ID(ffn_moe_down-78): 279 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 407 us MUL_MAT_ID(ffn_moe_down-79): 212 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 364 us MUL_MAT_ID(ffn_moe_down-80): 191 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 411 us MUL_MAT_ID(ffn_moe_down-81): 456 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 362 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 500 us MUL_MAT_ID(ffn_moe_down-83): 318 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 415 us MUL_MAT_ID(ffn_moe_down-84): 224 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 386 us MUL_MAT_ID(ffn_moe_down-85): 431 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 361 us MUL_MAT_ID(ffn_moe_down-86): 200 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 413 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 401 us MUL_MAT_ID(ffn_moe_down-88): 201 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 443 us MUL_MAT_ID(ffn_moe_down-89): 266 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 411 us MUL_MAT_ID(ffn_moe_down-90): 743 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 372 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 28 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 402 us MUL_MAT_ID(ffn_moe_down-25): 205 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 408 us MUL_MAT_ID(ffn_moe_down-26): 218 us ggml_barrier(...): 399 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 398 us MUL_MAT_ID(ffn_moe_down-27): 213 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 407 us MUL_MAT_ID(ffn_moe_down-28): 193 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 434 us MUL_MAT_ID(ffn_moe_down-29): 219 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 463 us MUL_MAT_ID(ffn_moe_down-30): 663 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 385 us MUL_MAT_ID(ffn_moe_down-31): 209 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 415 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 340 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 404 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 362 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 378 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 342 us MUL_MAT_ID(ffn_moe_down-37): 239 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 409 us MUL_MAT_ID(ffn_moe_down-38): 371 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 384 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 370 us MUL_MAT_ID(ffn_moe_down-40): 227 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 426 us MUL_MAT_ID(ffn_moe_down-41): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 353 us MUL_MAT_ID(ffn_moe_down-42): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 367 us MUL_MAT_ID(ffn_moe_down-43): 186 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 428 us MUL_MAT_ID(ffn_moe_down-44): 255 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 376 us MUL_MAT_ID(ffn_moe_down-45): 229 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 398 us MUL_MAT_ID(ffn_moe_down-46): 195 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 366 us MUL_MAT_ID(ffn_moe_down-47): 213 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 417 us MUL_MAT_ID(ffn_moe_down-48): 197 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 383 us MUL_MAT_ID(ffn_moe_down-49): 285 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 374 us MUL_MAT_ID(ffn_moe_down-50): 370 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 368 us MUL_MAT_ID(ffn_moe_down-51): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 395 us MUL_MAT_ID(ffn_moe_down-52): 209 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 359 us MUL_MAT_ID(ffn_moe_down-53): 425 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 407 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 372 us MUL_MAT_ID(ffn_moe_down-55): 220 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 434 us MUL_MAT_ID(ffn_moe_down-56): 294 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 398 us MUL_MAT_ID(ffn_moe_down-57): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 378 us MUL_MAT_ID(ffn_moe_down-58): 207 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 419 us MUL_MAT_ID(ffn_moe_down-59): 211 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 434 us MUL_MAT_ID(ffn_moe_down-60): 208 us ggml_barrier(...): 345 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 383 us MUL_MAT_ID(ffn_moe_down-61): 247 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 360 us MUL_MAT_ID(ffn_moe_down-62): 198 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 372 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 397 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 353 us MUL_MAT_ID(ffn_moe_down-65): 195 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 387 us MUL_MAT_ID(ffn_moe_down-66): 461 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 415 us MUL_MAT_ID(ffn_moe_down-67): 186 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 399 us MUL_MAT_ID(ffn_moe_down-68): 202 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 348 us MUL_MAT_ID(ffn_moe_down-69): 302 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 373 us MUL_MAT_ID(ffn_moe_down-70): 329 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 404 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 349 us MUL_MAT_ID(ffn_moe_down-72): 320 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 372 us MUL_MAT_ID(ffn_moe_down-73): 220 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 401 us MUL_MAT_ID(ffn_moe_down-74): 230 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 427 us MUL_MAT_ID(ffn_moe_down-75): 202 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 358 us MUL_MAT_ID(ffn_moe_down-76): 203 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 406 us MUL_MAT_ID(ffn_moe_down-77): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 344 us MUL_MAT_ID(ffn_moe_down-78): 208 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 403 us MUL_MAT_ID(ffn_moe_down-79): 207 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 377 us MUL_MAT_ID(ffn_moe_down-80): 201 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 337 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 340 us MUL_MAT_ID(ffn_moe_down-82): 200 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 396 us MUL_MAT_ID(ffn_moe_down-83): 201 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 338 us MUL_MAT_ID(ffn_moe_down-84): 225 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 408 us MUL_MAT_ID(ffn_moe_down-85): 226 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 389 us MUL_MAT_ID(ffn_moe_down-86): 195 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 402 us MUL_MAT_ID(ffn_moe_down-87): 219 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 529 us MUL_MAT_ID(ffn_moe_down-88): 187 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 391 us MUL_MAT_ID(ffn_moe_down-89): 204 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 403 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 422 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 86 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 372 us MUL_MAT_ID(ffn_moe_down-25): 205 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 422 us MUL_MAT_ID(ffn_moe_down-26): 210 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 407 us MUL_MAT_ID(ffn_moe_down-27): 226 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 380 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 364 us MUL_MAT_ID(ffn_moe_down-29): 188 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 378 us MUL_MAT_ID(ffn_moe_down-30): 183 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 409 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 429 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 364 us MUL_MAT_ID(ffn_moe_down-33): 209 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 419 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 402 us MUL_MAT_ID(ffn_moe_down-35): 204 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 374 us MUL_MAT_ID(ffn_moe_down-36): 369 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 372 us MUL_MAT_ID(ffn_moe_down-37): 223 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 405 us MUL_MAT_ID(ffn_moe_down-38): 208 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 416 us MUL_MAT_ID(ffn_moe_down-39): 224 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 385 us MUL_MAT_ID(ffn_moe_down-40): 425 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 359 us MUL_MAT_ID(ffn_moe_down-41): 536 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 410 us MUL_MAT_ID(ffn_moe_down-42): 525 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 372 us MUL_MAT_ID(ffn_moe_down-43): 183 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 435 us MUL_MAT_ID(ffn_moe_down-44): 202 us ggml_barrier(...): 518 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 404 us MUL_MAT_ID(ffn_moe_down-45): 465 us ggml_barrier(...): 419 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 381 us MUL_MAT_ID(ffn_moe_down-46): 213 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 406 us MUL_MAT_ID(ffn_moe_down-47): 208 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 403 us MUL_MAT_ID(ffn_moe_down-48): 200 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 383 us MUL_MAT_ID(ffn_moe_down-49): 185 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 421 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 399 us MUL_MAT_ID(ffn_moe_down-51): 222 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 438 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 360 us MUL_MAT_ID(ffn_moe_down-53): 370 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 351 us MUL_MAT_ID(ffn_moe_down-54): 451 us ggml_barrier(...): 345 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 407 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 421 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 426 us MUL_MAT_ID(ffn_moe_down-57): 370 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 460 us MUL_MAT_ID(ffn_moe_down-58): 358 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 415 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 376 us MUL_MAT_ID(ffn_moe_down-60): 515 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 535 us MUL_MAT_ID(ffn_moe_down-61): 203 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 527 us MUL_MAT_ID(ffn_moe_down-62): 468 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 358 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 409 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 498 us MUL_MAT_ID(ffn_moe_down-65): 284 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 482 us MUL_MAT_ID(ffn_moe_down-66): 222 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 432 us MUL_MAT_ID(ffn_moe_down-67): 257 us ggml_barrier(...): 369 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 401 us MUL_MAT_ID(ffn_moe_down-68): 178 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 455 us MUL_MAT_ID(ffn_moe_down-69): 480 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 394 us MUL_MAT_ID(ffn_moe_down-70): 208 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 405 us MUL_MAT_ID(ffn_moe_down-71): 279 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 423 us MUL_MAT_ID(ffn_moe_down-72): 199 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 396 us MUL_MAT_ID(ffn_moe_down-73): 252 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 400 us MUL_MAT_ID(ffn_moe_down-74): 540 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 370 us MUL_MAT_ID(ffn_moe_down-75): 395 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 486 us MUL_MAT_ID(ffn_moe_down-76): 209 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 215 us ggml_barrier(...): 369 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 428 us MUL_MAT_ID(ffn_moe_down-78): 188 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 505 us MUL_MAT_ID(ffn_moe_down-79): 317 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 431 us MUL_MAT_ID(ffn_moe_down-80): 206 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 421 us MUL_MAT_ID(ffn_moe_down-81): 199 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 496 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 408 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 453 us MUL_MAT_ID(ffn_moe_down-84): 248 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 379 us MUL_MAT_ID(ffn_moe_down-85): 255 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 381 us MUL_MAT_ID(ffn_moe_down-86): 176 us ggml_barrier(...): 346 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 412 us MUL_MAT_ID(ffn_moe_down-87): 206 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 429 us MUL_MAT_ID(ffn_moe_down-88): 184 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 372 us MUL_MAT_ID(ffn_moe_down-89): 392 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 416 us MUL_MAT_ID(ffn_moe_down-90): 245 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 436 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 7 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 365 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 423 us MUL_MAT_ID(ffn_moe_down-26): 170 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 406 us MUL_MAT_ID(ffn_moe_down-27): 181 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 399 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 382 us MUL_MAT_ID(ffn_moe_down-29): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 373 us MUL_MAT_ID(ffn_moe_down-30): 204 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 411 us MUL_MAT_ID(ffn_moe_down-31): 199 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 342 us MUL_MAT_ID(ffn_moe_down-32): 332 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 349 us MUL_MAT_ID(ffn_moe_down-33): 465 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 360 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 372 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 408 us MUL_MAT_ID(ffn_moe_down-36): 207 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 369 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 394 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 395 us MUL_MAT_ID(ffn_moe_down-39): 646 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 486 us MUL_MAT_ID(ffn_moe_down-40): 456 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 386 us MUL_MAT_ID(ffn_moe_down-41): 208 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 419 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 377 us MUL_MAT_ID(ffn_moe_down-43): 539 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 371 us MUL_MAT_ID(ffn_moe_down-44): 206 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 427 us MUL_MAT_ID(ffn_moe_down-45): 417 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 362 us MUL_MAT_ID(ffn_moe_down-46): 309 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 374 us MUL_MAT_ID(ffn_moe_down-47): 207 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 426 us MUL_MAT_ID(ffn_moe_down-48): 198 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 387 us MUL_MAT_ID(ffn_moe_down-49): 177 us ggml_barrier(...): 347 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 377 us MUL_MAT_ID(ffn_moe_down-50): 312 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 408 us MUL_MAT_ID(ffn_moe_down-51): 432 us ggml_barrier(...): 387 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 388 us MUL_MAT_ID(ffn_moe_down-52): 217 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 367 us MUL_MAT_ID(ffn_moe_down-53): 195 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 369 us MUL_MAT_ID(ffn_moe_down-54): 202 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 391 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 434 us MUL_MAT_ID(ffn_moe_down-56): 187 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 409 us MUL_MAT_ID(ffn_moe_down-57): 206 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 385 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 398 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 381 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 404 us MUL_MAT_ID(ffn_moe_down-61): 389 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 496 us MUL_MAT_ID(ffn_moe_down-62): 189 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 404 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 368 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 349 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 347 us MUL_MAT_ID(ffn_moe_down-66): 184 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 402 us MUL_MAT_ID(ffn_moe_down-67): 274 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 407 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 484 us MUL_MAT_ID(ffn_moe_down-69): 225 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 372 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 366 us MUL_MAT_ID(ffn_moe_down-71): 325 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 411 us MUL_MAT_ID(ffn_moe_down-72): 204 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 430 us MUL_MAT_ID(ffn_moe_down-73): 302 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 405 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 405 us MUL_MAT_ID(ffn_moe_down-75): 222 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 395 us MUL_MAT_ID(ffn_moe_down-76): 181 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 417 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 379 us MUL_MAT_ID(ffn_moe_down-78): 225 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 404 us MUL_MAT_ID(ffn_moe_down-79): 208 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 379 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 381 us MUL_MAT_ID(ffn_moe_down-81): 350 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 368 us MUL_MAT_ID(ffn_moe_down-82): 244 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 462 us MUL_MAT_ID(ffn_moe_down-83): 185 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 454 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 394 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 417 us MUL_MAT_ID(ffn_moe_down-86): 442 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 371 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 417 us MUL_MAT_ID(ffn_moe_down-88): 196 us ggml_barrier(...): 368 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 351 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 371 us MUL_MAT_ID(ffn_moe_down-90): 197 us ggml_barrier(...): 356 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 376 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 44 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 364 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 388 us MUL_MAT_ID(ffn_moe_down-26): 282 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 452 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 393 us MUL_MAT_ID(ffn_moe_down-28): 205 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 465 us MUL_MAT_ID(ffn_moe_down-30): 650 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 397 us MUL_MAT_ID(ffn_moe_down-31): 225 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 401 us MUL_MAT_ID(ffn_moe_down-32): 217 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 412 us MUL_MAT_ID(ffn_moe_down-33): 207 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 363 us MUL_MAT_ID(ffn_moe_down-34): 207 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 446 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 359 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 414 us MUL_MAT_ID(ffn_moe_down-37): 238 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 388 us MUL_MAT_ID(ffn_moe_down-38): 202 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 359 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 381 us MUL_MAT_ID(ffn_moe_down-40): 242 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 555 us MUL_MAT_ID(ffn_moe_down-41): 208 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 421 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 450 us MUL_MAT_ID(ffn_moe_down-43): 246 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 413 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 389 us MUL_MAT_ID(ffn_moe_down-45): 246 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 417 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 355 us MUL_MAT_ID(ffn_moe_down-47): 366 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 374 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 397 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 411 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 351 us MUL_MAT_ID(ffn_moe_down-51): 208 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 401 us MUL_MAT_ID(ffn_moe_down-52): 206 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 417 us MUL_MAT_ID(ffn_moe_down-53): 308 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 391 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 393 us MUL_MAT_ID(ffn_moe_down-55): 412 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 390 us MUL_MAT_ID(ffn_moe_down-56): 218 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 446 us MUL_MAT_ID(ffn_moe_down-57): 187 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 382 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 410 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 348 us MUL_MAT_ID(ffn_moe_down-60): 362 us ggml_barrier(...): 344 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 404 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 413 us MUL_MAT_ID(ffn_moe_down-62): 212 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 417 us MUL_MAT_ID(ffn_moe_down-63): 210 us ggml_barrier(...): 339 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 416 us MUL_MAT_ID(ffn_moe_down-64): 304 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 385 us MUL_MAT_ID(ffn_moe_down-65): 210 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 424 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 434 us MUL_MAT_ID(ffn_moe_down-67): 202 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 387 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 391 us MUL_MAT_ID(ffn_moe_down-69): 278 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 366 us MUL_MAT_ID(ffn_moe_down-70): 296 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 397 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 429 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 187 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 358 us MUL_MAT_ID(ffn_moe_down-74): 380 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 364 us MUL_MAT_ID(ffn_moe_down-75): 202 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 429 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 441 us MUL_MAT_ID(ffn_moe_down-77): 187 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 403 us MUL_MAT_ID(ffn_moe_down-78): 201 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 415 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 424 us MUL_MAT_ID(ffn_moe_down-80): 180 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 481 us MUL_MAT_ID(ffn_moe_down-81): 401 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 461 us MUL_MAT_ID(ffn_moe_down-82): 597 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 394 us MUL_MAT_ID(ffn_moe_down-83): 189 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 409 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 377 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 372 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 421 us MUL_MAT_ID(ffn_moe_down-87): 199 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 433 us MUL_MAT_ID(ffn_moe_down-88): 279 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 455 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 467 us MUL_MAT_ID(ffn_moe_down-90): 270 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 422 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 365 us MUL_MAT_ID(ffn_moe_down-25): 214 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 415 us MUL_MAT_ID(ffn_moe_down-26): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 378 us MUL_MAT_ID(ffn_moe_down-27): 215 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 413 us MUL_MAT_ID(ffn_moe_down-28): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 434 us MUL_MAT_ID(ffn_moe_down-29): 210 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 553 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 402 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 410 us MUL_MAT_ID(ffn_moe_down-32): 253 us ggml_barrier(...): 360 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 403 us MUL_MAT_ID(ffn_moe_down-33): 277 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 352 us MUL_MAT_ID(ffn_moe_down-34): 312 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 391 us MUL_MAT_ID(ffn_moe_down-35): 217 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 405 us MUL_MAT_ID(ffn_moe_down-36): 248 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 535 us MUL_MAT_ID(ffn_moe_down-37): 652 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 509 us MUL_MAT_ID(ffn_moe_down-38): 456 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 514 us MUL_MAT_ID(ffn_moe_down-39): 229 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 382 us MUL_MAT_ID(ffn_moe_down-40): 239 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 414 us MUL_MAT_ID(ffn_moe_down-41): 198 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 424 us MUL_MAT_ID(ffn_moe_down-42): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 399 us MUL_MAT_ID(ffn_moe_down-43): 220 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 400 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 420 us MUL_MAT_ID(ffn_moe_down-45): 212 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 383 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 400 us MUL_MAT_ID(ffn_moe_down-47): 181 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 441 us MUL_MAT_ID(ffn_moe_down-48): 262 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 382 us MUL_MAT_ID(ffn_moe_down-49): 345 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 468 us MUL_MAT_ID(ffn_moe_down-50): 246 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 390 us MUL_MAT_ID(ffn_moe_down-51): 322 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 372 us MUL_MAT_ID(ffn_moe_down-52): 180 us ggml_barrier(...): 361 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 422 us MUL_MAT_ID(ffn_moe_down-53): 282 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 413 us MUL_MAT_ID(ffn_moe_down-54): 201 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 357 us MUL_MAT_ID(ffn_moe_down-55): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 422 us MUL_MAT_ID(ffn_moe_down-56): 201 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 366 us MUL_MAT_ID(ffn_moe_down-57): 176 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 394 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 398 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 416 us MUL_MAT_ID(ffn_moe_down-60): 384 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 466 us MUL_MAT_ID(ffn_moe_down-61): 224 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 383 us MUL_MAT_ID(ffn_moe_down-62): 192 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 409 us MUL_MAT_ID(ffn_moe_down-63): 185 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 422 us MUL_MAT_ID(ffn_moe_down-64): 553 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 391 us MUL_MAT_ID(ffn_moe_down-65): 260 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 407 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 386 us MUL_MAT_ID(ffn_moe_down-67): 294 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 404 us MUL_MAT_ID(ffn_moe_down-68): 187 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 428 us MUL_MAT_ID(ffn_moe_down-69): 185 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 440 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 448 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 473 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 350 us MUL_MAT_ID(ffn_moe_down-72): 174 us ggml_barrier(...): 456 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 389 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 409 us MUL_MAT_ID(ffn_moe_down-74): 191 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 377 us MUL_MAT_ID(ffn_moe_down-75): 211 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 424 us MUL_MAT_ID(ffn_moe_down-76): 410 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 448 us MUL_MAT_ID(ffn_moe_down-77): 178 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 426 us MUL_MAT_ID(ffn_moe_down-78): 187 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 431 us MUL_MAT_ID(ffn_moe_down-79): 278 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 379 us MUL_MAT_ID(ffn_moe_down-80): 224 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 409 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 433 us MUL_MAT_ID(ffn_moe_down-82): 212 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 495 us MUL_MAT_ID(ffn_moe_down-83): 308 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 409 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 427 us MUL_MAT_ID(ffn_moe_down-85): 223 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 426 us MUL_MAT_ID(ffn_moe_down-86): 394 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 386 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 376 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 361 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 403 us MUL_MAT_ID(ffn_moe_down-90): 293 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 391 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 165 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 396 us MUL_MAT_ID(ffn_moe_down-25): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 413 us MUL_MAT_ID(ffn_moe_down-26): 287 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 419 us MUL_MAT_ID(ffn_moe_down-27): 252 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 395 us MUL_MAT_ID(ffn_moe_down-28): 394 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 392 us MUL_MAT_ID(ffn_moe_down-29): 192 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 441 us MUL_MAT_ID(ffn_moe_down-30): 163 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 377 us MUL_MAT_ID(ffn_moe_down-31): 197 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 528 us MUL_MAT_ID(ffn_moe_down-32): 226 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 434 us MUL_MAT_ID(ffn_moe_down-33): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 388 us MUL_MAT_ID(ffn_moe_down-34): 203 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 413 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 422 us MUL_MAT_ID(ffn_moe_down-36): 208 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 416 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 374 us MUL_MAT_ID(ffn_moe_down-38): 202 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 403 us MUL_MAT_ID(ffn_moe_down-39): 724 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 414 us MUL_MAT_ID(ffn_moe_down-40): 239 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 519 us MUL_MAT_ID(ffn_moe_down-41): 188 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 391 us MUL_MAT_ID(ffn_moe_down-42): 252 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 408 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 481 us MUL_MAT_ID(ffn_moe_down-44): 209 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 396 us MUL_MAT_ID(ffn_moe_down-45): 200 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 402 us MUL_MAT_ID(ffn_moe_down-46): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 449 us MUL_MAT_ID(ffn_moe_down-47): 202 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 451 us MUL_MAT_ID(ffn_moe_down-48): 202 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 410 us MUL_MAT_ID(ffn_moe_down-49): 179 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 442 us MUL_MAT_ID(ffn_moe_down-50): 209 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 404 us MUL_MAT_ID(ffn_moe_down-51): 183 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 394 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 426 us MUL_MAT_ID(ffn_moe_down-53): 184 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 379 us MUL_MAT_ID(ffn_moe_down-54): 205 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 430 us MUL_MAT_ID(ffn_moe_down-55): 203 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 417 us MUL_MAT_ID(ffn_moe_down-56): 286 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 415 us MUL_MAT_ID(ffn_moe_down-57): 198 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 408 us MUL_MAT_ID(ffn_moe_down-58): 208 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 429 us MUL_MAT_ID(ffn_moe_down-59): 208 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 403 us MUL_MAT_ID(ffn_moe_down-60): 214 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 437 us MUL_MAT_ID(ffn_moe_down-61): 207 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 413 us MUL_MAT_ID(ffn_moe_down-62): 198 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 352 us MUL_MAT_ID(ffn_moe_down-63): 195 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 384 us MUL_MAT_ID(ffn_moe_down-64): 189 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 430 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 388 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 415 us MUL_MAT_ID(ffn_moe_down-67): 195 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 438 us MUL_MAT_ID(ffn_moe_down-68): 228 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 524 us MUL_MAT_ID(ffn_moe_down-69): 243 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 399 us MUL_MAT_ID(ffn_moe_down-70): 397 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 359 us MUL_MAT_ID(ffn_moe_down-71): 197 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 373 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 418 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 428 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 475 us MUL_MAT_ID(ffn_moe_down-75): 218 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 399 us MUL_MAT_ID(ffn_moe_down-76): 187 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 453 us MUL_MAT_ID(ffn_moe_down-77): 324 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 390 us MUL_MAT_ID(ffn_moe_down-78): 216 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 356 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 403 us MUL_MAT_ID(ffn_moe_down-80): 196 us ggml_barrier(...): 418 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 385 us MUL_MAT_ID(ffn_moe_down-81): 209 us ggml_barrier(...): 412 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 395 us MUL_MAT_ID(ffn_moe_down-82): 183 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 381 us MUL_MAT_ID(ffn_moe_down-83): 209 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 423 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 463 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 362 us MUL_MAT_ID(ffn_moe_down-86): 313 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 380 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 547 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 395 us MUL_MAT_ID(ffn_moe_down-89): 207 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 391 us MUL_MAT_ID(ffn_moe_down-90): 238 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 393 us MUL_MAT_ID(ffn_moe_down-91): 257 us ggml_barrier(...): 359 us GET_ROWS(inp_embd): 22 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 423 us MUL_MAT_ID(ffn_moe_down-25): 217 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 404 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 380 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 383 us MUL_MAT_ID(ffn_moe_down-28): 460 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 402 us MUL_MAT_ID(ffn_moe_down-29): 186 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 356 us MUL_MAT_ID(ffn_moe_down-30): 188 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 388 us MUL_MAT_ID(ffn_moe_down-31): 203 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 402 us MUL_MAT_ID(ffn_moe_down-32): 398 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 403 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 366 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 342 us MUL_MAT_ID(ffn_moe_down-35): 204 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 369 us MUL_MAT_ID(ffn_moe_down-36): 444 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 377 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 457 us MUL_MAT_ID(ffn_moe_down-38): 191 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 381 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 405 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 380 us MUL_MAT_ID(ffn_moe_down-41): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 399 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 341 us MUL_MAT_ID(ffn_moe_down-43): 222 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 527 us MUL_MAT_ID(ffn_moe_down-44): 330 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 376 us MUL_MAT_ID(ffn_moe_down-45): 354 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 390 us MUL_MAT_ID(ffn_moe_down-46): 188 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 376 us MUL_MAT_ID(ffn_moe_down-47): 189 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 449 us MUL_MAT_ID(ffn_moe_down-48): 292 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 361 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 450 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 507 us MUL_MAT_ID(ffn_moe_down-51): 206 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 408 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 354 us MUL_MAT_ID(ffn_moe_down-53): 183 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 421 us MUL_MAT_ID(ffn_moe_down-54): 195 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 415 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 384 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 461 us MUL_MAT_ID(ffn_moe_down-57): 179 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 377 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 391 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 422 us MUL_MAT_ID(ffn_moe_down-60): 487 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 419 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 348 us MUL_MAT_ID(ffn_moe_down-62): 266 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 511 us MUL_MAT_ID(ffn_moe_down-63): 205 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 394 us MUL_MAT_ID(ffn_moe_down-64): 243 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 440 us MUL_MAT_ID(ffn_moe_down-65): 510 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 427 us MUL_MAT_ID(ffn_moe_down-66): 241 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 425 us MUL_MAT_ID(ffn_moe_down-67): 210 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 432 us MUL_MAT_ID(ffn_moe_down-68): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 437 us MUL_MAT_ID(ffn_moe_down-69): 213 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 448 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 434 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 404 us MUL_MAT_ID(ffn_moe_down-72): 216 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 512 us MUL_MAT_ID(ffn_moe_down-73): 279 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 398 us MUL_MAT_ID(ffn_moe_down-74): 208 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 396 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 436 us MUL_MAT_ID(ffn_moe_down-76): 204 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 409 us MUL_MAT_ID(ffn_moe_down-77): 184 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 438 us MUL_MAT_ID(ffn_moe_down-78): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 377 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 419 us MUL_MAT_ID(ffn_moe_down-80): 210 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 381 us MUL_MAT_ID(ffn_moe_down-81): 287 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 417 us MUL_MAT_ID(ffn_moe_down-82): 410 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 415 us MUL_MAT_ID(ffn_moe_down-83): 412 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 417 us MUL_MAT_ID(ffn_moe_down-84): 224 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 392 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 374 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 431 us MUL_MAT_ID(ffn_moe_down-87): 177 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 348 us MUL_MAT_ID(ffn_moe_down-88): 185 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 386 us MUL_MAT_ID(ffn_moe_down-89): 322 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 425 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 410 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 133 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 404 us MUL_MAT_ID(ffn_moe_down-25): 206 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 412 us MUL_MAT_ID(ffn_moe_down-26): 306 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 410 us MUL_MAT_ID(ffn_moe_down-27): 224 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 358 us MUL_MAT_ID(ffn_moe_down-28): 442 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 400 us MUL_MAT_ID(ffn_moe_down-29): 296 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 386 us MUL_MAT_ID(ffn_moe_down-30): 181 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 445 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 353 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 360 us MUL_MAT_ID(ffn_moe_down-33): 235 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 395 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 395 us MUL_MAT_ID(ffn_moe_down-35): 267 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 412 us MUL_MAT_ID(ffn_moe_down-36): 207 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 417 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 389 us MUL_MAT_ID(ffn_moe_down-38): 209 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 409 us MUL_MAT_ID(ffn_moe_down-39): 707 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 357 us MUL_MAT_ID(ffn_moe_down-40): 657 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 391 us MUL_MAT_ID(ffn_moe_down-41): 195 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 365 us MUL_MAT_ID(ffn_moe_down-42): 181 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 438 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 431 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 416 us MUL_MAT_ID(ffn_moe_down-45): 420 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 385 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 434 us MUL_MAT_ID(ffn_moe_down-47): 200 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 507 us MUL_MAT_ID(ffn_moe_down-48): 221 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 409 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 405 us MUL_MAT_ID(ffn_moe_down-50): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 401 us MUL_MAT_ID(ffn_moe_down-51): 193 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 417 us MUL_MAT_ID(ffn_moe_down-52): 195 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 476 us MUL_MAT_ID(ffn_moe_down-53): 293 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 435 us MUL_MAT_ID(ffn_moe_down-54): 211 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 412 us MUL_MAT_ID(ffn_moe_down-55): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 453 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 404 us MUL_MAT_ID(ffn_moe_down-57): 203 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 387 us MUL_MAT_ID(ffn_moe_down-58): 342 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 399 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 418 us MUL_MAT_ID(ffn_moe_down-60): 481 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 375 us MUL_MAT_ID(ffn_moe_down-61): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 394 us MUL_MAT_ID(ffn_moe_down-62): 188 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 483 us MUL_MAT_ID(ffn_moe_down-63): 317 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 419 us MUL_MAT_ID(ffn_moe_down-64): 567 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 398 us MUL_MAT_ID(ffn_moe_down-65): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 403 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 402 us MUL_MAT_ID(ffn_moe_down-67): 213 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 338 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 397 us MUL_MAT_ID(ffn_moe_down-69): 198 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 380 us MUL_MAT_ID(ffn_moe_down-70): 228 us ggml_barrier(...): 366 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 437 us MUL_MAT_ID(ffn_moe_down-71): 294 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 403 us MUL_MAT_ID(ffn_moe_down-72): 204 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 372 us MUL_MAT_ID(ffn_moe_down-73): 183 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 349 us MUL_MAT_ID(ffn_moe_down-74): 212 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 409 us MUL_MAT_ID(ffn_moe_down-75): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 354 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 399 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 472 us MUL_MAT_ID(ffn_moe_down-78): 299 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 389 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 380 us MUL_MAT_ID(ffn_moe_down-80): 422 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 397 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 421 us MUL_MAT_ID(ffn_moe_down-82): 187 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 365 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 415 us MUL_MAT_ID(ffn_moe_down-84): 244 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 415 us MUL_MAT_ID(ffn_moe_down-85): 262 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 377 us MUL_MAT_ID(ffn_moe_down-86): 226 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 386 us MUL_MAT_ID(ffn_moe_down-87): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 450 us MUL_MAT_ID(ffn_moe_down-88): 274 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 408 us MUL_MAT_ID(ffn_moe_down-89): 249 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 382 us MUL_MAT_ID(ffn_moe_down-90): 237 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 414 us MUL_MAT_ID(ffn_moe_down-91): 280 us ggml_barrier(...): 48 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 398 us MUL_MAT_ID(ffn_moe_down-25): 236 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 385 us MUL_MAT_ID(ffn_moe_down-26): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 436 us MUL_MAT_ID(ffn_moe_down-27): 265 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 464 us MUL_MAT_ID(ffn_moe_down-28): 225 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 408 us MUL_MAT_ID(ffn_moe_down-29): 453 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 374 us MUL_MAT_ID(ffn_moe_down-30): 187 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 401 us MUL_MAT_ID(ffn_moe_down-31): 596 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 503 us MUL_MAT_ID(ffn_moe_down-32): 201 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 383 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 518 us MUL_MAT_ID(ffn_moe_down-34): 222 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 435 us MUL_MAT_ID(ffn_moe_down-35): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 318 us MUL_MAT_ID(ffn_moe_down-36): 170 us ggml_barrier(...): 338 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 413 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 355 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 409 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 412 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 373 us MUL_MAT_ID(ffn_moe_down-41): 177 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 440 us MUL_MAT_ID(ffn_moe_down-42): 213 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 380 us MUL_MAT_ID(ffn_moe_down-43): 272 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 377 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 444 us MUL_MAT_ID(ffn_moe_down-45): 287 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 374 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 454 us MUL_MAT_ID(ffn_moe_down-47): 236 us ggml_barrier(...): 341 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 413 us MUL_MAT_ID(ffn_moe_down-48): 175 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 345 us MUL_MAT_ID(ffn_moe_down-49): 215 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 355 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 402 us MUL_MAT_ID(ffn_moe_down-51): 223 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 437 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 411 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 405 us MUL_MAT_ID(ffn_moe_down-54): 450 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 401 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 513 us MUL_MAT_ID(ffn_moe_down-56): 241 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 411 us MUL_MAT_ID(ffn_moe_down-57): 217 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 404 us MUL_MAT_ID(ffn_moe_down-58): 314 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 418 us MUL_MAT_ID(ffn_moe_down-59): 187 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 396 us MUL_MAT_ID(ffn_moe_down-60): 168 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 423 us MUL_MAT_ID(ffn_moe_down-61): 227 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 392 us MUL_MAT_ID(ffn_moe_down-62): 307 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 407 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 418 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 375 us MUL_MAT_ID(ffn_moe_down-65): 401 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 373 us MUL_MAT_ID(ffn_moe_down-66): 292 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 360 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 432 us MUL_MAT_ID(ffn_moe_down-68): 283 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 389 us MUL_MAT_ID(ffn_moe_down-69): 284 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 393 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 382 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 414 us MUL_MAT_ID(ffn_moe_down-72): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 400 us MUL_MAT_ID(ffn_moe_down-73): 211 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 395 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 409 us MUL_MAT_ID(ffn_moe_down-75): 410 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 437 us MUL_MAT_ID(ffn_moe_down-76): 188 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 379 us MUL_MAT_ID(ffn_moe_down-77): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 446 us MUL_MAT_ID(ffn_moe_down-78): 272 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 421 us MUL_MAT_ID(ffn_moe_down-79): 202 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 393 us MUL_MAT_ID(ffn_moe_down-80): 254 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 406 us MUL_MAT_ID(ffn_moe_down-81): 233 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 385 us MUL_MAT_ID(ffn_moe_down-82): 581 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 352 us MUL_MAT_ID(ffn_moe_down-83): 212 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 423 us MUL_MAT_ID(ffn_moe_down-84): 387 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 399 us MUL_MAT_ID(ffn_moe_down-85): 230 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 436 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 407 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 380 us MUL_MAT_ID(ffn_moe_down-88): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 384 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 373 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 374 us MUL_MAT_ID(ffn_moe_down-91): 242 us ggml_barrier(...): 30 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 379 us MUL_MAT_ID(ffn_moe_down-25): 219 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 393 us MUL_MAT_ID(ffn_moe_down-26): 212 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 460 us MUL_MAT_ID(ffn_moe_down-27): 378 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 431 us MUL_MAT_ID(ffn_moe_down-28): 206 us ggml_barrier(...): 547 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 495 us MUL_MAT_ID(ffn_moe_down-29): 201 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 202 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 408 us MUL_MAT_ID(ffn_moe_down-31): 182 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 427 us MUL_MAT_ID(ffn_moe_down-32): 380 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 419 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 547 us MUL_MAT_ID(ffn_moe_down-34): 276 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 396 us MUL_MAT_ID(ffn_moe_down-35): 201 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 423 us MUL_MAT_ID(ffn_moe_down-36): 289 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 405 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 445 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 402 us MUL_MAT_ID(ffn_moe_down-39): 231 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 520 us MUL_MAT_ID(ffn_moe_down-40): 212 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 386 us MUL_MAT_ID(ffn_moe_down-41): 249 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 384 us MUL_MAT_ID(ffn_moe_down-42): 323 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 428 us MUL_MAT_ID(ffn_moe_down-43): 229 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 423 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 405 us MUL_MAT_ID(ffn_moe_down-45): 324 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 410 us MUL_MAT_ID(ffn_moe_down-46): 229 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 391 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 396 us MUL_MAT_ID(ffn_moe_down-48): 389 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 378 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 508 us MUL_MAT_ID(ffn_moe_down-50): 224 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 418 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 422 us MUL_MAT_ID(ffn_moe_down-52): 215 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 389 us MUL_MAT_ID(ffn_moe_down-53): 644 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 403 us MUL_MAT_ID(ffn_moe_down-54): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 398 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 412 us MUL_MAT_ID(ffn_moe_down-56): 543 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 428 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 376 us MUL_MAT_ID(ffn_moe_down-58): 269 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 397 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 403 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 435 us MUL_MAT_ID(ffn_moe_down-61): 256 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 432 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 352 us MUL_MAT_ID(ffn_moe_down-63): 524 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 390 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 344 us MUL_MAT_ID(ffn_moe_down-65): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 369 us MUL_MAT_ID(ffn_moe_down-66): 321 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 396 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 391 us MUL_MAT_ID(ffn_moe_down-68): 201 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 391 us MUL_MAT_ID(ffn_moe_down-69): 211 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 561 us MUL_MAT_ID(ffn_moe_down-70): 169 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 394 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 407 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 382 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 370 us MUL_MAT_ID(ffn_moe_down-74): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 350 us MUL_MAT_ID(ffn_moe_down-75): 307 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 414 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 394 us MUL_MAT_ID(ffn_moe_down-77): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 421 us MUL_MAT_ID(ffn_moe_down-78): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 406 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 337 us MUL_MAT_ID(ffn_moe_down-80): 218 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 378 us MUL_MAT_ID(ffn_moe_down-81): 188 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 442 us MUL_MAT_ID(ffn_moe_down-82): 288 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 357 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 375 us MUL_MAT_ID(ffn_moe_down-84): 225 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 391 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 401 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 362 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 365 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 349 us MUL_MAT_ID(ffn_moe_down-88): 418 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 385 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 379 us MUL_MAT_ID(ffn_moe_down-90): 229 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 429 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 8 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 274 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 489 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 365 us MUL_MAT_ID(ffn_moe_down-27): 209 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 462 us MUL_MAT_ID(ffn_moe_down-29): 216 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 457 us MUL_MAT_ID(ffn_moe_down-30): 580 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 409 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 385 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 386 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 408 us MUL_MAT_ID(ffn_moe_down-34): 203 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 400 us MUL_MAT_ID(ffn_moe_down-35): 378 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 370 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 412 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 445 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 401 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 418 us MUL_MAT_ID(ffn_moe_down-40): 733 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 470 us MUL_MAT_ID(ffn_moe_down-41): 211 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 415 us MUL_MAT_ID(ffn_moe_down-42): 203 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 432 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 412 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 479 us MUL_MAT_ID(ffn_moe_down-45): 454 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 395 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 493 us MUL_MAT_ID(ffn_moe_down-47): 183 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 419 us MUL_MAT_ID(ffn_moe_down-48): 351 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 402 us MUL_MAT_ID(ffn_moe_down-49): 204 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 392 us MUL_MAT_ID(ffn_moe_down-50): 378 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 403 us MUL_MAT_ID(ffn_moe_down-51): 205 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 474 us MUL_MAT_ID(ffn_moe_down-52): 246 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 440 us MUL_MAT_ID(ffn_moe_down-53): 181 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 434 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 389 us MUL_MAT_ID(ffn_moe_down-55): 183 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 415 us MUL_MAT_ID(ffn_moe_down-56): 199 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 439 us MUL_MAT_ID(ffn_moe_down-57): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 451 us MUL_MAT_ID(ffn_moe_down-58): 251 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 385 us MUL_MAT_ID(ffn_moe_down-59): 275 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 401 us MUL_MAT_ID(ffn_moe_down-60): 272 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 381 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 422 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 360 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 416 us MUL_MAT_ID(ffn_moe_down-64): 217 us ggml_barrier(...): 343 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 403 us MUL_MAT_ID(ffn_moe_down-65): 205 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 379 us MUL_MAT_ID(ffn_moe_down-66): 198 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 356 us MUL_MAT_ID(ffn_moe_down-67): 236 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 511 us MUL_MAT_ID(ffn_moe_down-68): 186 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 399 us MUL_MAT_ID(ffn_moe_down-69): 186 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 371 us MUL_MAT_ID(ffn_moe_down-70): 213 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 394 us MUL_MAT_ID(ffn_moe_down-71): 283 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 407 us MUL_MAT_ID(ffn_moe_down-72): 280 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 402 us MUL_MAT_ID(ffn_moe_down-73): 207 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 421 us MUL_MAT_ID(ffn_moe_down-74): 206 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 469 us MUL_MAT_ID(ffn_moe_down-75): 296 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 376 us MUL_MAT_ID(ffn_moe_down-76): 311 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 392 us MUL_MAT_ID(ffn_moe_down-77): 245 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 384 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 470 us MUL_MAT_ID(ffn_moe_down-79): 711 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 389 us MUL_MAT_ID(ffn_moe_down-80): 230 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 398 us MUL_MAT_ID(ffn_moe_down-81): 199 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 399 us MUL_MAT_ID(ffn_moe_down-82): 183 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 395 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 436 us MUL_MAT_ID(ffn_moe_down-84): 360 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 398 us MUL_MAT_ID(ffn_moe_down-85): 291 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 380 us MUL_MAT_ID(ffn_moe_down-86): 397 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 458 us MUL_MAT_ID(ffn_moe_down-87): 185 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 400 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 367 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 405 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 402 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 26 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 391 us MUL_MAT_ID(ffn_moe_down-25): 204 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 412 us MUL_MAT_ID(ffn_moe_down-26): 220 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 470 us MUL_MAT_ID(ffn_moe_down-27): 400 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 402 us MUL_MAT_ID(ffn_moe_down-28): 440 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 393 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 226 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 472 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 460 us MUL_MAT_ID(ffn_moe_down-32): 206 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 410 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 407 us MUL_MAT_ID(ffn_moe_down-34): 416 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 406 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 368 us MUL_MAT_ID(ffn_moe_down-36): 190 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 449 us MUL_MAT_ID(ffn_moe_down-37): 238 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 455 us MUL_MAT_ID(ffn_moe_down-38): 588 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 402 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 420 us MUL_MAT_ID(ffn_moe_down-40): 405 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 392 us MUL_MAT_ID(ffn_moe_down-41): 190 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 442 us MUL_MAT_ID(ffn_moe_down-42): 189 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 385 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 424 us MUL_MAT_ID(ffn_moe_down-44): 330 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 366 us MUL_MAT_ID(ffn_moe_down-45): 491 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 417 us MUL_MAT_ID(ffn_moe_down-46): 277 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 423 us MUL_MAT_ID(ffn_moe_down-47): 197 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 429 us MUL_MAT_ID(ffn_moe_down-48): 559 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 384 us MUL_MAT_ID(ffn_moe_down-49): 192 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 401 us MUL_MAT_ID(ffn_moe_down-50): 386 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 403 us MUL_MAT_ID(ffn_moe_down-51): 201 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 375 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 437 us MUL_MAT_ID(ffn_moe_down-53): 426 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 398 us MUL_MAT_ID(ffn_moe_down-54): 426 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 381 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 405 us MUL_MAT_ID(ffn_moe_down-56): 213 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 414 us MUL_MAT_ID(ffn_moe_down-57): 225 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 426 us MUL_MAT_ID(ffn_moe_down-58): 207 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 497 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 214 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 388 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 425 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 365 us MUL_MAT_ID(ffn_moe_down-63): 194 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 377 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 402 us MUL_MAT_ID(ffn_moe_down-65): 190 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 586 us MUL_MAT_ID(ffn_moe_down-66): 230 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 389 us MUL_MAT_ID(ffn_moe_down-67): 211 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 367 us MUL_MAT_ID(ffn_moe_down-68): 205 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 399 us MUL_MAT_ID(ffn_moe_down-69): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 382 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 361 us MUL_MAT_ID(ffn_moe_down-71): 183 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 385 us MUL_MAT_ID(ffn_moe_down-72): 425 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 392 us MUL_MAT_ID(ffn_moe_down-73): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 346 us MUL_MAT_ID(ffn_moe_down-74): 425 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 350 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 397 us MUL_MAT_ID(ffn_moe_down-76): 178 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 392 us MUL_MAT_ID(ffn_moe_down-77): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 386 us MUL_MAT_ID(ffn_moe_down-78): 268 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 376 us MUL_MAT_ID(ffn_moe_down-79): 417 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 379 us MUL_MAT_ID(ffn_moe_down-80): 221 us ggml_barrier(...): 403 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 379 us MUL_MAT_ID(ffn_moe_down-81): 207 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 376 us MUL_MAT_ID(ffn_moe_down-82): 211 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 433 us MUL_MAT_ID(ffn_moe_down-83): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 426 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 429 us MUL_MAT_ID(ffn_moe_down-85): 270 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 354 us MUL_MAT_ID(ffn_moe_down-86): 414 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 428 us MUL_MAT_ID(ffn_moe_down-87): 181 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 406 us MUL_MAT_ID(ffn_moe_down-88): 311 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 389 us MUL_MAT_ID(ffn_moe_down-89): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 431 us MUL_MAT_ID(ffn_moe_down-90): 223 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 395 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 368 us MUL_MAT_ID(ffn_moe_down-25): 203 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 391 us MUL_MAT_ID(ffn_moe_down-26): 219 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 367 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 380 us MUL_MAT_ID(ffn_moe_down-28): 269 us ggml_barrier(...): 426 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 539 us MUL_MAT_ID(ffn_moe_down-29): 191 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 430 us MUL_MAT_ID(ffn_moe_down-30): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 456 us MUL_MAT_ID(ffn_moe_down-31): 185 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 401 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 433 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 360 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 402 us MUL_MAT_ID(ffn_moe_down-35): 182 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 392 us MUL_MAT_ID(ffn_moe_down-36): 180 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 392 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 434 us MUL_MAT_ID(ffn_moe_down-38): 210 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 451 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 491 us MUL_MAT_ID(ffn_moe_down-40): 258 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 431 us MUL_MAT_ID(ffn_moe_down-41): 358 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 418 us MUL_MAT_ID(ffn_moe_down-42): 211 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 360 us MUL_MAT_ID(ffn_moe_down-43): 416 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 428 us MUL_MAT_ID(ffn_moe_down-44): 271 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 372 us MUL_MAT_ID(ffn_moe_down-45): 227 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 405 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 414 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 425 us MUL_MAT_ID(ffn_moe_down-48): 202 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 384 us MUL_MAT_ID(ffn_moe_down-49): 308 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 412 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 386 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 444 us MUL_MAT_ID(ffn_moe_down-52): 236 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 413 us MUL_MAT_ID(ffn_moe_down-53): 225 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 462 us MUL_MAT_ID(ffn_moe_down-54): 408 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 401 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 406 us MUL_MAT_ID(ffn_moe_down-56): 396 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 402 us MUL_MAT_ID(ffn_moe_down-57): 370 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 392 us MUL_MAT_ID(ffn_moe_down-58): 210 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 391 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 415 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 431 us MUL_MAT_ID(ffn_moe_down-61): 209 us ggml_barrier(...): 408 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 431 us MUL_MAT_ID(ffn_moe_down-62): 186 us ggml_barrier(...): 438 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 409 us MUL_MAT_ID(ffn_moe_down-63): 297 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 420 us MUL_MAT_ID(ffn_moe_down-64): 201 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 395 us MUL_MAT_ID(ffn_moe_down-65): 198 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 385 us MUL_MAT_ID(ffn_moe_down-66): 239 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 376 us MUL_MAT_ID(ffn_moe_down-67): 575 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 361 us MUL_MAT_ID(ffn_moe_down-68): 244 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 446 us MUL_MAT_ID(ffn_moe_down-69): 185 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 358 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 350 us MUL_MAT_ID(ffn_moe_down-71): 417 us ggml_barrier(...): 411 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 428 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 398 us MUL_MAT_ID(ffn_moe_down-73): 205 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 380 us MUL_MAT_ID(ffn_moe_down-74): 194 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 565 us MUL_MAT_ID(ffn_moe_down-75): 571 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 481 us MUL_MAT_ID(ffn_moe_down-76): 211 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 491 us MUL_MAT_ID(ffn_moe_down-77): 191 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 398 us MUL_MAT_ID(ffn_moe_down-78): 202 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 363 us MUL_MAT_ID(ffn_moe_down-79): 210 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 416 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 376 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 409 us MUL_MAT_ID(ffn_moe_down-82): 269 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 372 us MUL_MAT_ID(ffn_moe_down-83): 190 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 379 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 512 us MUL_MAT_ID(ffn_moe_down-85): 217 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 402 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 381 us MUL_MAT_ID(ffn_moe_down-87): 212 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 372 us MUL_MAT_ID(ffn_moe_down-88): 204 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 396 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 380 us MUL_MAT_ID(ffn_moe_down-90): 193 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 400 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 22 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 374 us MUL_MAT_ID(ffn_moe_down-25): 202 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 406 us MUL_MAT_ID(ffn_moe_down-26): 239 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 361 us MUL_MAT_ID(ffn_moe_down-27): 189 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 408 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 371 us MUL_MAT_ID(ffn_moe_down-29): 209 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 459 us MUL_MAT_ID(ffn_moe_down-30): 474 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 506 us MUL_MAT_ID(ffn_moe_down-31): 214 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 419 us MUL_MAT_ID(ffn_moe_down-32): 266 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 392 us MUL_MAT_ID(ffn_moe_down-33): 247 us ggml_barrier(...): 385 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 431 us MUL_MAT_ID(ffn_moe_down-34): 189 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 387 us MUL_MAT_ID(ffn_moe_down-35): 304 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 478 us MUL_MAT_ID(ffn_moe_down-36): 374 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 544 us MUL_MAT_ID(ffn_moe_down-37): 194 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 464 us MUL_MAT_ID(ffn_moe_down-38): 188 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 432 us MUL_MAT_ID(ffn_moe_down-39): 394 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 391 us MUL_MAT_ID(ffn_moe_down-40): 241 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 387 us MUL_MAT_ID(ffn_moe_down-41): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 414 us MUL_MAT_ID(ffn_moe_down-42): 186 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 386 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 398 us MUL_MAT_ID(ffn_moe_down-44): 307 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 351 us MUL_MAT_ID(ffn_moe_down-45): 369 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 363 us MUL_MAT_ID(ffn_moe_down-46): 208 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 441 us MUL_MAT_ID(ffn_moe_down-47): 285 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 383 us MUL_MAT_ID(ffn_moe_down-48): 298 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 340 us MUL_MAT_ID(ffn_moe_down-49): 277 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 384 us MUL_MAT_ID(ffn_moe_down-50): 206 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 420 us MUL_MAT_ID(ffn_moe_down-51): 256 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 472 us MUL_MAT_ID(ffn_moe_down-52): 281 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 393 us MUL_MAT_ID(ffn_moe_down-53): 203 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 488 us MUL_MAT_ID(ffn_moe_down-54): 245 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 431 us MUL_MAT_ID(ffn_moe_down-55): 214 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 416 us MUL_MAT_ID(ffn_moe_down-56): 216 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 418 us MUL_MAT_ID(ffn_moe_down-57): 203 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 401 us MUL_MAT_ID(ffn_moe_down-58): 375 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 386 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 475 us MUL_MAT_ID(ffn_moe_down-60): 279 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 413 us MUL_MAT_ID(ffn_moe_down-61): 196 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 355 us MUL_MAT_ID(ffn_moe_down-62): 188 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 391 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 489 us MUL_MAT_ID(ffn_moe_down-64): 247 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 376 us MUL_MAT_ID(ffn_moe_down-65): 206 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 403 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 430 us MUL_MAT_ID(ffn_moe_down-67): 385 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 442 us MUL_MAT_ID(ffn_moe_down-68): 211 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 407 us MUL_MAT_ID(ffn_moe_down-69): 183 us ggml_barrier(...): 363 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 439 us MUL_MAT_ID(ffn_moe_down-70): 248 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 423 us MUL_MAT_ID(ffn_moe_down-71): 206 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 439 us MUL_MAT_ID(ffn_moe_down-72): 239 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 374 us MUL_MAT_ID(ffn_moe_down-73): 177 us ggml_barrier(...): 419 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 382 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 478 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 506 us MUL_MAT_ID(ffn_moe_down-76): 210 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 464 us MUL_MAT_ID(ffn_moe_down-77): 270 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 485 us MUL_MAT_ID(ffn_moe_down-78): 179 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 412 us MUL_MAT_ID(ffn_moe_down-79): 368 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 412 us MUL_MAT_ID(ffn_moe_down-80): 255 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 439 us MUL_MAT_ID(ffn_moe_down-81): 321 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 367 us MUL_MAT_ID(ffn_moe_down-82): 313 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 461 us MUL_MAT_ID(ffn_moe_down-83): 230 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 392 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 408 us MUL_MAT_ID(ffn_moe_down-85): 271 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 408 us MUL_MAT_ID(ffn_moe_down-86): 281 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 358 us MUL_MAT_ID(ffn_moe_down-87): 198 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 433 us MUL_MAT_ID(ffn_moe_down-88): 205 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 480 us MUL_MAT_ID(ffn_moe_down-89): 541 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 399 us MUL_MAT_ID(ffn_moe_down-90): 249 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 456 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 9 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 367 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 382 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 431 us MUL_MAT_ID(ffn_moe_down-27): 383 us ggml_barrier(...): 468 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 395 us MUL_MAT_ID(ffn_moe_down-28): 180 us ggml_barrier(...): 470 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 376 us MUL_MAT_ID(ffn_moe_down-29): 238 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 407 us MUL_MAT_ID(ffn_moe_down-30): 216 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 509 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 416 us MUL_MAT_ID(ffn_moe_down-32): 199 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 380 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 347 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 367 us MUL_MAT_ID(ffn_moe_down-35): 259 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 414 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 410 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 368 us MUL_MAT_ID(ffn_moe_down-38): 220 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 386 us MUL_MAT_ID(ffn_moe_down-39): 715 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 374 us MUL_MAT_ID(ffn_moe_down-40): 234 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 417 us MUL_MAT_ID(ffn_moe_down-41): 200 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 398 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 399 us MUL_MAT_ID(ffn_moe_down-43): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 396 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 472 us MUL_MAT_ID(ffn_moe_down-45): 439 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 372 us MUL_MAT_ID(ffn_moe_down-46): 554 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 386 us MUL_MAT_ID(ffn_moe_down-47): 201 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 350 us MUL_MAT_ID(ffn_moe_down-48): 223 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 403 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 365 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 424 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 356 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 376 us MUL_MAT_ID(ffn_moe_down-53): 225 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 398 us MUL_MAT_ID(ffn_moe_down-54): 190 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 185 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 347 us MUL_MAT_ID(ffn_moe_down-56): 295 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 398 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 426 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 397 us MUL_MAT_ID(ffn_moe_down-59): 218 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 378 us MUL_MAT_ID(ffn_moe_down-60): 214 us ggml_barrier(...): 346 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 406 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 413 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 395 us MUL_MAT_ID(ffn_moe_down-63): 187 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 392 us MUL_MAT_ID(ffn_moe_down-64): 181 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 390 us MUL_MAT_ID(ffn_moe_down-65): 213 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 394 us MUL_MAT_ID(ffn_moe_down-66): 240 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 361 us MUL_MAT_ID(ffn_moe_down-67): 298 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 441 us MUL_MAT_ID(ffn_moe_down-68): 269 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 380 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 382 us MUL_MAT_ID(ffn_moe_down-70): 206 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 403 us MUL_MAT_ID(ffn_moe_down-71): 212 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 433 us MUL_MAT_ID(ffn_moe_down-72): 544 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 381 us MUL_MAT_ID(ffn_moe_down-73): 212 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 383 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 387 us MUL_MAT_ID(ffn_moe_down-75): 462 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 391 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 400 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 425 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 371 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 387 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 348 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 351 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 390 us MUL_MAT_ID(ffn_moe_down-82): 184 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 428 us MUL_MAT_ID(ffn_moe_down-83): 263 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 385 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 412 us MUL_MAT_ID(ffn_moe_down-85): 673 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 352 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 445 us MUL_MAT_ID(ffn_moe_down-87): 593 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 369 us MUL_MAT_ID(ffn_moe_down-88): 208 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 420 us MUL_MAT_ID(ffn_moe_down-89): 212 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 427 us MUL_MAT_ID(ffn_moe_down-90): 631 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 406 us MUL_MAT_ID(ffn_moe_down-91): 723 us ggml_barrier(...): 57 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 370 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 399 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 485 us MUL_MAT_ID(ffn_moe_down-27): 181 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 393 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 391 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 425 us MUL_MAT_ID(ffn_moe_down-30): 237 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 415 us MUL_MAT_ID(ffn_moe_down-31): 202 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 374 us MUL_MAT_ID(ffn_moe_down-32): 211 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 402 us MUL_MAT_ID(ffn_moe_down-33): 220 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 440 us MUL_MAT_ID(ffn_moe_down-34): 202 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 397 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 397 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 416 us MUL_MAT_ID(ffn_moe_down-37): 277 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 397 us MUL_MAT_ID(ffn_moe_down-38): 184 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 376 us MUL_MAT_ID(ffn_moe_down-39): 244 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 371 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 405 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 362 us MUL_MAT_ID(ffn_moe_down-42): 269 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 400 us MUL_MAT_ID(ffn_moe_down-43): 201 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 364 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 425 us MUL_MAT_ID(ffn_moe_down-45): 196 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 385 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 395 us MUL_MAT_ID(ffn_moe_down-47): 324 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 397 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 408 us MUL_MAT_ID(ffn_moe_down-49): 192 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 353 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 526 us MUL_MAT_ID(ffn_moe_down-51): 280 us ggml_barrier(...): 394 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 375 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 322 us MUL_MAT_ID(ffn_moe_down-53): 448 us ggml_barrier(...): 385 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 388 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 402 us MUL_MAT_ID(ffn_moe_down-55): 185 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 422 us MUL_MAT_ID(ffn_moe_down-56): 205 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 408 us MUL_MAT_ID(ffn_moe_down-57): 202 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 390 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 370 us MUL_MAT_ID(ffn_moe_down-59): 230 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 372 us MUL_MAT_ID(ffn_moe_down-60): 207 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 387 us MUL_MAT_ID(ffn_moe_down-61): 208 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 382 us MUL_MAT_ID(ffn_moe_down-62): 226 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 403 us MUL_MAT_ID(ffn_moe_down-63): 386 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 414 us MUL_MAT_ID(ffn_moe_down-64): 207 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 361 us MUL_MAT_ID(ffn_moe_down-65): 391 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 372 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 370 us MUL_MAT_ID(ffn_moe_down-67): 251 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 419 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 401 us MUL_MAT_ID(ffn_moe_down-69): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 343 us MUL_MAT_ID(ffn_moe_down-70): 427 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 390 us MUL_MAT_ID(ffn_moe_down-71): 203 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 409 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 408 us MUL_MAT_ID(ffn_moe_down-73): 205 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 448 us MUL_MAT_ID(ffn_moe_down-74): 216 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 372 us MUL_MAT_ID(ffn_moe_down-75): 198 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 413 us MUL_MAT_ID(ffn_moe_down-76): 289 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 415 us MUL_MAT_ID(ffn_moe_down-77): 196 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 390 us MUL_MAT_ID(ffn_moe_down-78): 208 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 422 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 453 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 342 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 393 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 423 us MUL_MAT_ID(ffn_moe_down-83): 188 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 377 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 402 us MUL_MAT_ID(ffn_moe_down-85): 222 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 541 us MUL_MAT_ID(ffn_moe_down-86): 404 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 385 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 396 us MUL_MAT_ID(ffn_moe_down-88): 389 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 377 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 398 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 394 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 406 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 375 us MUL_MAT_ID(ffn_moe_down-25): 252 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 506 us MUL_MAT_ID(ffn_moe_down-26): 286 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 417 us MUL_MAT_ID(ffn_moe_down-27): 259 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 402 us MUL_MAT_ID(ffn_moe_down-28): 244 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 409 us MUL_MAT_ID(ffn_moe_down-29): 207 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 355 us MUL_MAT_ID(ffn_moe_down-30): 558 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 401 us MUL_MAT_ID(ffn_moe_down-31): 210 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 379 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 378 us MUL_MAT_ID(ffn_moe_down-33): 208 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 404 us MUL_MAT_ID(ffn_moe_down-34): 384 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 386 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 398 us MUL_MAT_ID(ffn_moe_down-36): 188 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 440 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 501 us MUL_MAT_ID(ffn_moe_down-38): 222 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 392 us MUL_MAT_ID(ffn_moe_down-39): 569 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 389 us MUL_MAT_ID(ffn_moe_down-40): 244 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 384 us MUL_MAT_ID(ffn_moe_down-41): 209 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 402 us MUL_MAT_ID(ffn_moe_down-42): 396 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 416 us MUL_MAT_ID(ffn_moe_down-43): 207 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 383 us MUL_MAT_ID(ffn_moe_down-44): 202 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 397 us MUL_MAT_ID(ffn_moe_down-45): 196 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 399 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 395 us MUL_MAT_ID(ffn_moe_down-47): 290 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 394 us MUL_MAT_ID(ffn_moe_down-48): 198 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 495 us MUL_MAT_ID(ffn_moe_down-49): 260 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 378 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 415 us MUL_MAT_ID(ffn_moe_down-51): 289 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 450 us MUL_MAT_ID(ffn_moe_down-52): 218 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 390 us MUL_MAT_ID(ffn_moe_down-53): 201 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 447 us MUL_MAT_ID(ffn_moe_down-54): 421 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 434 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 342 us MUL_MAT_ID(ffn_moe_down-56): 206 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 364 us MUL_MAT_ID(ffn_moe_down-57): 243 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 425 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 429 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 333 us MUL_MAT_ID(ffn_moe_down-60): 209 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 462 us MUL_MAT_ID(ffn_moe_down-61): 205 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 340 us MUL_MAT_ID(ffn_moe_down-62): 218 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 373 us MUL_MAT_ID(ffn_moe_down-63): 344 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 385 us MUL_MAT_ID(ffn_moe_down-64): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 441 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 468 us MUL_MAT_ID(ffn_moe_down-66): 338 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 399 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 411 us MUL_MAT_ID(ffn_moe_down-68): 167 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 436 us MUL_MAT_ID(ffn_moe_down-69): 284 us ggml_barrier(...): 360 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 410 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 368 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 381 us MUL_MAT_ID(ffn_moe_down-72): 189 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 435 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 397 us MUL_MAT_ID(ffn_moe_down-74): 192 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 389 us MUL_MAT_ID(ffn_moe_down-75): 523 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 448 us MUL_MAT_ID(ffn_moe_down-76): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 426 us MUL_MAT_ID(ffn_moe_down-77): 712 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 378 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 382 us MUL_MAT_ID(ffn_moe_down-79): 208 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 456 us MUL_MAT_ID(ffn_moe_down-80): 213 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 412 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 236 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 460 us MUL_MAT_ID(ffn_moe_down-83): 284 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 429 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 415 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 438 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 427 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 418 us MUL_MAT_ID(ffn_moe_down-88): 207 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 405 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 424 us MUL_MAT_ID(ffn_moe_down-90): 622 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 365 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 75 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 388 us MUL_MAT_ID(ffn_moe_down-25): 167 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 400 us MUL_MAT_ID(ffn_moe_down-26): 199 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 476 us MUL_MAT_ID(ffn_moe_down-27): 202 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 415 us MUL_MAT_ID(ffn_moe_down-28): 268 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 472 us MUL_MAT_ID(ffn_moe_down-29): 206 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 364 us MUL_MAT_ID(ffn_moe_down-30): 226 us ggml_barrier(...): 338 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 409 us MUL_MAT_ID(ffn_moe_down-31): 208 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 522 us MUL_MAT_ID(ffn_moe_down-32): 275 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 356 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 364 us MUL_MAT_ID(ffn_moe_down-34): 297 us ggml_barrier(...): 346 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 416 us MUL_MAT_ID(ffn_moe_down-35): 207 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 410 us MUL_MAT_ID(ffn_moe_down-36): 248 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 410 us MUL_MAT_ID(ffn_moe_down-37): 313 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 385 us MUL_MAT_ID(ffn_moe_down-38): 205 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 413 us MUL_MAT_ID(ffn_moe_down-39): 750 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 404 us MUL_MAT_ID(ffn_moe_down-40): 213 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 408 us MUL_MAT_ID(ffn_moe_down-41): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 387 us MUL_MAT_ID(ffn_moe_down-42): 241 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 430 us MUL_MAT_ID(ffn_moe_down-43): 222 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 468 us MUL_MAT_ID(ffn_moe_down-44): 200 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 484 us MUL_MAT_ID(ffn_moe_down-45): 197 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 414 us MUL_MAT_ID(ffn_moe_down-46): 205 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 417 us MUL_MAT_ID(ffn_moe_down-47): 220 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 410 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 439 us MUL_MAT_ID(ffn_moe_down-49): 271 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 404 us MUL_MAT_ID(ffn_moe_down-50): 385 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 354 us MUL_MAT_ID(ffn_moe_down-51): 184 us ggml_barrier(...): 397 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 403 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 405 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 490 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 379 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 332 us MUL_MAT_ID(ffn_moe_down-56): 231 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 398 us MUL_MAT_ID(ffn_moe_down-57): 188 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 453 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 381 us MUL_MAT_ID(ffn_moe_down-59): 335 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 406 us MUL_MAT_ID(ffn_moe_down-60): 174 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 436 us MUL_MAT_ID(ffn_moe_down-61): 268 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 443 us MUL_MAT_ID(ffn_moe_down-62): 301 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 493 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 501 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 389 us MUL_MAT_ID(ffn_moe_down-65): 204 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 386 us MUL_MAT_ID(ffn_moe_down-66): 225 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 412 us MUL_MAT_ID(ffn_moe_down-67): 280 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 395 us MUL_MAT_ID(ffn_moe_down-68): 241 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 414 us MUL_MAT_ID(ffn_moe_down-69): 198 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 405 us MUL_MAT_ID(ffn_moe_down-70): 390 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 387 us MUL_MAT_ID(ffn_moe_down-71): 186 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 343 us MUL_MAT_ID(ffn_moe_down-72): 222 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 449 us MUL_MAT_ID(ffn_moe_down-73): 206 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 487 us MUL_MAT_ID(ffn_moe_down-74): 213 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 381 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 488 us MUL_MAT_ID(ffn_moe_down-76): 305 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 389 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 457 us MUL_MAT_ID(ffn_moe_down-78): 222 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 399 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 362 us MUL_MAT_ID(ffn_moe_down-80): 206 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 396 us MUL_MAT_ID(ffn_moe_down-81): 415 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 401 us MUL_MAT_ID(ffn_moe_down-82): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 366 us MUL_MAT_ID(ffn_moe_down-83): 318 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 412 us MUL_MAT_ID(ffn_moe_down-84): 645 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 412 us MUL_MAT_ID(ffn_moe_down-85): 725 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 437 us MUL_MAT_ID(ffn_moe_down-86): 431 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 378 us MUL_MAT_ID(ffn_moe_down-87): 187 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 519 us MUL_MAT_ID(ffn_moe_down-88): 179 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 409 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 336 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 342 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 85 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 380 us MUL_MAT_ID(ffn_moe_down-25): 332 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 436 us MUL_MAT_ID(ffn_moe_down-26): 198 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 376 us MUL_MAT_ID(ffn_moe_down-27): 189 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 477 us MUL_MAT_ID(ffn_moe_down-28): 208 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 440 us MUL_MAT_ID(ffn_moe_down-29): 203 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 454 us MUL_MAT_ID(ffn_moe_down-30): 219 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 384 us MUL_MAT_ID(ffn_moe_down-31): 197 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 488 us MUL_MAT_ID(ffn_moe_down-32): 187 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 397 us MUL_MAT_ID(ffn_moe_down-33): 208 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 508 us MUL_MAT_ID(ffn_moe_down-34): 286 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 408 us MUL_MAT_ID(ffn_moe_down-35): 204 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 345 us MUL_MAT_ID(ffn_moe_down-36): 210 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 398 us MUL_MAT_ID(ffn_moe_down-37): 261 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 460 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 508 us MUL_MAT_ID(ffn_moe_down-39): 454 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 376 us MUL_MAT_ID(ffn_moe_down-40): 234 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 411 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 391 us MUL_MAT_ID(ffn_moe_down-42): 226 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 406 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 476 us MUL_MAT_ID(ffn_moe_down-44): 179 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 437 us MUL_MAT_ID(ffn_moe_down-45): 320 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 524 us MUL_MAT_ID(ffn_moe_down-46): 207 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 409 us MUL_MAT_ID(ffn_moe_down-47): 269 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 398 us MUL_MAT_ID(ffn_moe_down-48): 243 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 410 us MUL_MAT_ID(ffn_moe_down-49): 219 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 430 us MUL_MAT_ID(ffn_moe_down-50): 285 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 457 us MUL_MAT_ID(ffn_moe_down-51): 206 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 497 us MUL_MAT_ID(ffn_moe_down-52): 224 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 391 us MUL_MAT_ID(ffn_moe_down-53): 207 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 398 us MUL_MAT_ID(ffn_moe_down-54): 224 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 478 us MUL_MAT_ID(ffn_moe_down-55): 193 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 378 us MUL_MAT_ID(ffn_moe_down-56): 210 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 435 us MUL_MAT_ID(ffn_moe_down-57): 189 us ggml_barrier(...): 365 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 451 us MUL_MAT_ID(ffn_moe_down-58): 234 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 467 us MUL_MAT_ID(ffn_moe_down-59): 307 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 435 us MUL_MAT_ID(ffn_moe_down-60): 219 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 431 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 452 us MUL_MAT_ID(ffn_moe_down-62): 178 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 444 us MUL_MAT_ID(ffn_moe_down-63): 247 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 386 us MUL_MAT_ID(ffn_moe_down-64): 277 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 403 us MUL_MAT_ID(ffn_moe_down-65): 412 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 477 us MUL_MAT_ID(ffn_moe_down-66): 234 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 514 us MUL_MAT_ID(ffn_moe_down-67): 210 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 409 us MUL_MAT_ID(ffn_moe_down-68): 378 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 415 us MUL_MAT_ID(ffn_moe_down-69): 494 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 438 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 371 us MUL_MAT_ID(ffn_moe_down-71): 223 us ggml_barrier(...): 452 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 409 us MUL_MAT_ID(ffn_moe_down-72): 192 us ggml_barrier(...): 374 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 454 us MUL_MAT_ID(ffn_moe_down-73): 296 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 386 us MUL_MAT_ID(ffn_moe_down-74): 189 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 406 us MUL_MAT_ID(ffn_moe_down-75): 282 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 477 us MUL_MAT_ID(ffn_moe_down-76): 298 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 407 us MUL_MAT_ID(ffn_moe_down-77): 184 us ggml_barrier(...): 340 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 410 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 462 us MUL_MAT_ID(ffn_moe_down-79): 212 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 424 us MUL_MAT_ID(ffn_moe_down-80): 223 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 446 us MUL_MAT_ID(ffn_moe_down-81): 289 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 439 us MUL_MAT_ID(ffn_moe_down-82): 320 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 373 us MUL_MAT_ID(ffn_moe_down-83): 244 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 386 us MUL_MAT_ID(ffn_moe_down-84): 219 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 479 us MUL_MAT_ID(ffn_moe_down-85): 294 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 387 us MUL_MAT_ID(ffn_moe_down-86): 216 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 442 us MUL_MAT_ID(ffn_moe_down-87): 172 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 404 us MUL_MAT_ID(ffn_moe_down-88): 198 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 347 us MUL_MAT_ID(ffn_moe_down-89): 185 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 424 us MUL_MAT_ID(ffn_moe_down-90): 229 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 428 us MUL_MAT_ID(ffn_moe_down-91): 251 us ggml_barrier(...): 26 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 505 us MUL_MAT_ID(ffn_moe_down-25): 214 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 485 us MUL_MAT_ID(ffn_moe_down-26): 295 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 425 us MUL_MAT_ID(ffn_moe_down-27): 213 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 406 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 392 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 555 us MUL_MAT_ID(ffn_moe_down-30): 189 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 431 us MUL_MAT_ID(ffn_moe_down-31): 217 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 387 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 398 us MUL_MAT_ID(ffn_moe_down-33): 202 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 384 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 341 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 344 us MUL_MAT_ID(ffn_moe_down-35): 212 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 475 us MUL_MAT_ID(ffn_moe_down-36): 173 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 470 us MUL_MAT_ID(ffn_moe_down-37): 250 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 503 us MUL_MAT_ID(ffn_moe_down-38): 177 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 396 us MUL_MAT_ID(ffn_moe_down-39): 463 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 444 us MUL_MAT_ID(ffn_moe_down-40): 234 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 482 us MUL_MAT_ID(ffn_moe_down-41): 198 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 476 us MUL_MAT_ID(ffn_moe_down-42): 193 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 433 us MUL_MAT_ID(ffn_moe_down-43): 206 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 393 us MUL_MAT_ID(ffn_moe_down-44): 206 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 364 us MUL_MAT_ID(ffn_moe_down-45): 208 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 446 us MUL_MAT_ID(ffn_moe_down-46): 360 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 415 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 439 us MUL_MAT_ID(ffn_moe_down-48): 196 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 461 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 473 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 364 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 374 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 444 us MUL_MAT_ID(ffn_moe_down-53): 384 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 367 us MUL_MAT_ID(ffn_moe_down-54): 409 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 411 us MUL_MAT_ID(ffn_moe_down-55): 264 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 388 us MUL_MAT_ID(ffn_moe_down-56): 232 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 433 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 390 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 392 us MUL_MAT_ID(ffn_moe_down-59): 274 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 370 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 366 us MUL_MAT_ID(ffn_moe_down-61): 210 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 435 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 414 us MUL_MAT_ID(ffn_moe_down-63): 186 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 447 us MUL_MAT_ID(ffn_moe_down-64): 212 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 445 us MUL_MAT_ID(ffn_moe_down-65): 231 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 409 us MUL_MAT_ID(ffn_moe_down-66): 185 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 399 us MUL_MAT_ID(ffn_moe_down-67): 185 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 460 us MUL_MAT_ID(ffn_moe_down-68): 189 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 523 us MUL_MAT_ID(ffn_moe_down-69): 204 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 377 us MUL_MAT_ID(ffn_moe_down-70): 210 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 457 us MUL_MAT_ID(ffn_moe_down-71): 186 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 385 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 433 us MUL_MAT_ID(ffn_moe_down-73): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 386 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 557 us MUL_MAT_ID(ffn_moe_down-75): 182 us ggml_barrier(...): 387 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 430 us MUL_MAT_ID(ffn_moe_down-76): 201 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 390 us MUL_MAT_ID(ffn_moe_down-77): 392 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 425 us MUL_MAT_ID(ffn_moe_down-78): 204 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 444 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 409 us MUL_MAT_ID(ffn_moe_down-80): 213 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 437 us MUL_MAT_ID(ffn_moe_down-81): 199 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 439 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 424 us MUL_MAT_ID(ffn_moe_down-83): 209 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 455 us MUL_MAT_ID(ffn_moe_down-84): 453 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 410 us MUL_MAT_ID(ffn_moe_down-85): 643 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 355 us MUL_MAT_ID(ffn_moe_down-86): 185 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 420 us MUL_MAT_ID(ffn_moe_down-87): 356 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 392 us MUL_MAT_ID(ffn_moe_down-88): 218 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 421 us MUL_MAT_ID(ffn_moe_down-89): 185 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 411 us MUL_MAT_ID(ffn_moe_down-90): 495 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 378 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 39 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 412 us MUL_MAT_ID(ffn_moe_down-25): 179 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 407 us MUL_MAT_ID(ffn_moe_down-26): 198 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 372 us MUL_MAT_ID(ffn_moe_down-27): 252 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 411 us MUL_MAT_ID(ffn_moe_down-28): 218 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 428 us MUL_MAT_ID(ffn_moe_down-29): 183 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 384 us MUL_MAT_ID(ffn_moe_down-30): 200 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 338 us MUL_MAT_ID(ffn_moe_down-31): 215 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 399 us MUL_MAT_ID(ffn_moe_down-32): 183 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 384 us MUL_MAT_ID(ffn_moe_down-33): 268 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 454 us MUL_MAT_ID(ffn_moe_down-34): 377 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 408 us MUL_MAT_ID(ffn_moe_down-35): 205 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 422 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 369 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 415 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 451 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 416 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 412 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 422 us MUL_MAT_ID(ffn_moe_down-41): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 469 us MUL_MAT_ID(ffn_moe_down-42): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 405 us MUL_MAT_ID(ffn_moe_down-43): 199 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 404 us MUL_MAT_ID(ffn_moe_down-44): 187 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 472 us MUL_MAT_ID(ffn_moe_down-45): 201 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 437 us MUL_MAT_ID(ffn_moe_down-46): 384 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 415 us MUL_MAT_ID(ffn_moe_down-47): 427 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 371 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 416 us MUL_MAT_ID(ffn_moe_down-49): 213 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 516 us MUL_MAT_ID(ffn_moe_down-50): 180 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 398 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 459 us MUL_MAT_ID(ffn_moe_down-52): 410 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 398 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 417 us MUL_MAT_ID(ffn_moe_down-54): 272 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 415 us MUL_MAT_ID(ffn_moe_down-55): 202 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 423 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 437 us MUL_MAT_ID(ffn_moe_down-57): 274 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 414 us MUL_MAT_ID(ffn_moe_down-58): 251 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 460 us MUL_MAT_ID(ffn_moe_down-59): 423 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 420 us MUL_MAT_ID(ffn_moe_down-60): 443 us ggml_barrier(...): 369 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 397 us MUL_MAT_ID(ffn_moe_down-61): 334 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 414 us MUL_MAT_ID(ffn_moe_down-62): 570 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 430 us MUL_MAT_ID(ffn_moe_down-63): 212 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 434 us MUL_MAT_ID(ffn_moe_down-64): 258 us ggml_barrier(...): 340 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 447 us MUL_MAT_ID(ffn_moe_down-65): 210 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 375 us MUL_MAT_ID(ffn_moe_down-66): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 350 us MUL_MAT_ID(ffn_moe_down-67): 211 us ggml_barrier(...): 430 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 423 us MUL_MAT_ID(ffn_moe_down-68): 261 us ggml_barrier(...): 341 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 399 us MUL_MAT_ID(ffn_moe_down-69): 184 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 409 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 396 us MUL_MAT_ID(ffn_moe_down-71): 186 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 403 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 433 us MUL_MAT_ID(ffn_moe_down-73): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 394 us MUL_MAT_ID(ffn_moe_down-74): 206 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 389 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 386 us MUL_MAT_ID(ffn_moe_down-76): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 423 us MUL_MAT_ID(ffn_moe_down-77): 210 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 407 us MUL_MAT_ID(ffn_moe_down-78): 205 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 382 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 370 us MUL_MAT_ID(ffn_moe_down-80): 298 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 392 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 442 us MUL_MAT_ID(ffn_moe_down-82): 217 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 439 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 429 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 445 us MUL_MAT_ID(ffn_moe_down-85): 730 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 409 us MUL_MAT_ID(ffn_moe_down-86): 483 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 363 us MUL_MAT_ID(ffn_moe_down-87): 187 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 424 us MUL_MAT_ID(ffn_moe_down-88): 237 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 405 us MUL_MAT_ID(ffn_moe_down-89): 185 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 446 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 392 us MUL_MAT_ID(ffn_moe_down-91): 538 us ggml_barrier(...): 31 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 391 us MUL_MAT_ID(ffn_moe_down-25): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 418 us MUL_MAT_ID(ffn_moe_down-26): 208 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 424 us MUL_MAT_ID(ffn_moe_down-27): 207 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 464 us MUL_MAT_ID(ffn_moe_down-28): 281 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 390 us MUL_MAT_ID(ffn_moe_down-29): 186 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 441 us MUL_MAT_ID(ffn_moe_down-30): 169 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 401 us MUL_MAT_ID(ffn_moe_down-31): 399 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 428 us MUL_MAT_ID(ffn_moe_down-32): 203 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 428 us MUL_MAT_ID(ffn_moe_down-33): 208 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 378 us MUL_MAT_ID(ffn_moe_down-34): 349 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 391 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 416 us MUL_MAT_ID(ffn_moe_down-36): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 391 us MUL_MAT_ID(ffn_moe_down-37): 277 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 390 us MUL_MAT_ID(ffn_moe_down-38): 200 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 436 us MUL_MAT_ID(ffn_moe_down-39): 248 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 453 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 404 us MUL_MAT_ID(ffn_moe_down-41): 419 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 394 us MUL_MAT_ID(ffn_moe_down-42): 476 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 368 us MUL_MAT_ID(ffn_moe_down-43): 390 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 404 us MUL_MAT_ID(ffn_moe_down-44): 194 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 429 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 407 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 407 us MUL_MAT_ID(ffn_moe_down-47): 221 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 383 us MUL_MAT_ID(ffn_moe_down-48): 199 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 412 us MUL_MAT_ID(ffn_moe_down-49): 217 us ggml_barrier(...): 365 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 426 us MUL_MAT_ID(ffn_moe_down-50): 317 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 394 us MUL_MAT_ID(ffn_moe_down-51): 183 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 398 us MUL_MAT_ID(ffn_moe_down-52): 229 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 379 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 377 us MUL_MAT_ID(ffn_moe_down-54): 355 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 376 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 400 us MUL_MAT_ID(ffn_moe_down-56): 206 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 412 us MUL_MAT_ID(ffn_moe_down-57): 420 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 390 us MUL_MAT_ID(ffn_moe_down-58): 189 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 417 us MUL_MAT_ID(ffn_moe_down-59): 216 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 377 us MUL_MAT_ID(ffn_moe_down-60): 204 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 394 us MUL_MAT_ID(ffn_moe_down-61): 202 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 408 us MUL_MAT_ID(ffn_moe_down-62): 295 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 376 us MUL_MAT_ID(ffn_moe_down-63): 199 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 379 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 397 us MUL_MAT_ID(ffn_moe_down-65): 214 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 381 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 405 us MUL_MAT_ID(ffn_moe_down-67): 214 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 395 us MUL_MAT_ID(ffn_moe_down-68): 202 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 384 us MUL_MAT_ID(ffn_moe_down-69): 217 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 433 us MUL_MAT_ID(ffn_moe_down-70): 218 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 436 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 359 us MUL_MAT_ID(ffn_moe_down-72): 186 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 361 us MUL_MAT_ID(ffn_moe_down-73): 204 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 384 us MUL_MAT_ID(ffn_moe_down-74): 413 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 376 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 390 us MUL_MAT_ID(ffn_moe_down-76): 188 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 416 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 502 us MUL_MAT_ID(ffn_moe_down-78): 326 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 395 us MUL_MAT_ID(ffn_moe_down-79): 242 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 419 us MUL_MAT_ID(ffn_moe_down-80): 206 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 365 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 357 us MUL_MAT_ID(ffn_moe_down-82): 218 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 433 us MUL_MAT_ID(ffn_moe_down-83): 567 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 418 us MUL_MAT_ID(ffn_moe_down-84): 358 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 371 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 350 us MUL_MAT_ID(ffn_moe_down-86): 483 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 401 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 393 us MUL_MAT_ID(ffn_moe_down-88): 204 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 385 us MUL_MAT_ID(ffn_moe_down-89): 218 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 414 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 391 us MUL_MAT_ID(ffn_moe_down-91): 259 us ggml_barrier(...): 48 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 417 us MUL_MAT_ID(ffn_moe_down-25): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 376 us MUL_MAT_ID(ffn_moe_down-26): 218 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 398 us MUL_MAT_ID(ffn_moe_down-27): 198 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 424 us MUL_MAT_ID(ffn_moe_down-28): 586 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 407 us MUL_MAT_ID(ffn_moe_down-29): 442 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 418 us MUL_MAT_ID(ffn_moe_down-30): 176 us ggml_barrier(...): 335 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 431 us MUL_MAT_ID(ffn_moe_down-31): 185 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 379 us MUL_MAT_ID(ffn_moe_down-32): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 433 us MUL_MAT_ID(ffn_moe_down-33): 401 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 413 us MUL_MAT_ID(ffn_moe_down-34): 188 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 392 us MUL_MAT_ID(ffn_moe_down-35): 182 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 392 us MUL_MAT_ID(ffn_moe_down-36): 215 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 438 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 476 us MUL_MAT_ID(ffn_moe_down-38): 249 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 408 us MUL_MAT_ID(ffn_moe_down-39): 707 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 423 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 437 us MUL_MAT_ID(ffn_moe_down-41): 233 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 448 us MUL_MAT_ID(ffn_moe_down-42): 443 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 366 us MUL_MAT_ID(ffn_moe_down-43): 189 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 438 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 443 us MUL_MAT_ID(ffn_moe_down-45): 241 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 368 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 441 us MUL_MAT_ID(ffn_moe_down-47): 187 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 363 us MUL_MAT_ID(ffn_moe_down-48): 190 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 391 us MUL_MAT_ID(ffn_moe_down-49): 213 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 417 us MUL_MAT_ID(ffn_moe_down-50): 206 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 418 us MUL_MAT_ID(ffn_moe_down-51): 191 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 424 us MUL_MAT_ID(ffn_moe_down-52): 211 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 440 us MUL_MAT_ID(ffn_moe_down-53): 363 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 393 us MUL_MAT_ID(ffn_moe_down-54): 186 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 407 us MUL_MAT_ID(ffn_moe_down-55): 176 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 415 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 421 us MUL_MAT_ID(ffn_moe_down-57): 250 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 414 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 438 us MUL_MAT_ID(ffn_moe_down-59): 311 us ggml_barrier(...): 351 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 390 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 381 us MUL_MAT_ID(ffn_moe_down-61): 200 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 429 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 446 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 355 us MUL_MAT_ID(ffn_moe_down-64): 186 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 435 us MUL_MAT_ID(ffn_moe_down-65): 206 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 399 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 368 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 444 us MUL_MAT_ID(ffn_moe_down-67): 185 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 452 us MUL_MAT_ID(ffn_moe_down-68): 180 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 378 us MUL_MAT_ID(ffn_moe_down-69): 197 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 416 us MUL_MAT_ID(ffn_moe_down-70): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 438 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 351 us MUL_MAT_ID(ffn_moe_down-72): 216 us ggml_barrier(...): 335 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 439 us MUL_MAT_ID(ffn_moe_down-73): 205 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 370 us MUL_MAT_ID(ffn_moe_down-74): 203 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 412 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 415 us MUL_MAT_ID(ffn_moe_down-76): 199 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 422 us MUL_MAT_ID(ffn_moe_down-77): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 441 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 430 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 427 us MUL_MAT_ID(ffn_moe_down-80): 336 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 468 us MUL_MAT_ID(ffn_moe_down-81): 182 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 358 us MUL_MAT_ID(ffn_moe_down-82): 200 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 384 us MUL_MAT_ID(ffn_moe_down-83): 207 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 443 us MUL_MAT_ID(ffn_moe_down-84): 248 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 430 us MUL_MAT_ID(ffn_moe_down-85): 237 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 405 us MUL_MAT_ID(ffn_moe_down-86): 197 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 426 us MUL_MAT_ID(ffn_moe_down-87): 188 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 420 us MUL_MAT_ID(ffn_moe_down-88): 220 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 439 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 403 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 406 us MUL_MAT_ID(ffn_moe_down-91): 646 us ggml_barrier(...): 77 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 389 us MUL_MAT_ID(ffn_moe_down-25): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 381 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 502 us MUL_MAT_ID(ffn_moe_down-27): 190 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 413 us MUL_MAT_ID(ffn_moe_down-28): 593 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 395 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 406 us MUL_MAT_ID(ffn_moe_down-30): 542 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 485 us MUL_MAT_ID(ffn_moe_down-31): 368 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 384 us MUL_MAT_ID(ffn_moe_down-32): 212 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 437 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 344 us MUL_MAT_ID(ffn_moe_down-35): 207 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 402 us MUL_MAT_ID(ffn_moe_down-36): 236 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 395 us MUL_MAT_ID(ffn_moe_down-37): 243 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 397 us MUL_MAT_ID(ffn_moe_down-38): 356 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 394 us MUL_MAT_ID(ffn_moe_down-39): 252 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 370 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 361 us MUL_MAT_ID(ffn_moe_down-41): 210 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 442 us MUL_MAT_ID(ffn_moe_down-42): 207 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 360 us MUL_MAT_ID(ffn_moe_down-43): 220 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 434 us MUL_MAT_ID(ffn_moe_down-44): 313 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 552 us MUL_MAT_ID(ffn_moe_down-45): 284 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 415 us MUL_MAT_ID(ffn_moe_down-46): 217 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 417 us MUL_MAT_ID(ffn_moe_down-47): 243 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 407 us MUL_MAT_ID(ffn_moe_down-48): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 395 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 372 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 406 us MUL_MAT_ID(ffn_moe_down-51): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 339 us MUL_MAT_ID(ffn_moe_down-52): 568 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 417 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 369 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 406 us MUL_MAT_ID(ffn_moe_down-55): 215 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 438 us MUL_MAT_ID(ffn_moe_down-56): 185 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 355 us MUL_MAT_ID(ffn_moe_down-57): 182 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 428 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 418 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 418 us MUL_MAT_ID(ffn_moe_down-60): 211 us ggml_barrier(...): 337 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 380 us MUL_MAT_ID(ffn_moe_down-61): 202 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 366 us MUL_MAT_ID(ffn_moe_down-62): 186 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 399 us MUL_MAT_ID(ffn_moe_down-63): 199 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 446 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 450 us MUL_MAT_ID(ffn_moe_down-65): 214 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 389 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 452 us MUL_MAT_ID(ffn_moe_down-67): 192 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 425 us MUL_MAT_ID(ffn_moe_down-68): 188 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 446 us MUL_MAT_ID(ffn_moe_down-69): 320 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 402 us MUL_MAT_ID(ffn_moe_down-70): 191 us ggml_barrier(...): 462 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 421 us MUL_MAT_ID(ffn_moe_down-71): 203 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 482 us MUL_MAT_ID(ffn_moe_down-72): 302 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 431 us MUL_MAT_ID(ffn_moe_down-73): 213 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 421 us MUL_MAT_ID(ffn_moe_down-74): 189 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 445 us MUL_MAT_ID(ffn_moe_down-75): 211 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 411 us MUL_MAT_ID(ffn_moe_down-76): 332 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 388 us MUL_MAT_ID(ffn_moe_down-77): 181 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 439 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 361 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 404 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 469 us MUL_MAT_ID(ffn_moe_down-81): 209 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 462 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 403 us MUL_MAT_ID(ffn_moe_down-83): 226 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 453 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 412 us MUL_MAT_ID(ffn_moe_down-85): 237 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 376 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 443 us MUL_MAT_ID(ffn_moe_down-87): 228 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 440 us MUL_MAT_ID(ffn_moe_down-88): 225 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 437 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 382 us MUL_MAT_ID(ffn_moe_down-90): 244 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 435 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 16 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 378 us MUL_MAT_ID(ffn_moe_down-25): 177 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 435 us MUL_MAT_ID(ffn_moe_down-26): 603 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 390 us MUL_MAT_ID(ffn_moe_down-27): 216 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 400 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 345 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 422 us MUL_MAT_ID(ffn_moe_down-30): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 429 us MUL_MAT_ID(ffn_moe_down-31): 423 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 446 us MUL_MAT_ID(ffn_moe_down-32): 187 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 430 us MUL_MAT_ID(ffn_moe_down-33): 213 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 428 us MUL_MAT_ID(ffn_moe_down-34): 229 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 395 us MUL_MAT_ID(ffn_moe_down-35): 180 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 405 us MUL_MAT_ID(ffn_moe_down-36): 207 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 419 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 485 us MUL_MAT_ID(ffn_moe_down-38): 163 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 407 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 413 us MUL_MAT_ID(ffn_moe_down-40): 483 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 446 us MUL_MAT_ID(ffn_moe_down-41): 199 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 359 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 403 us MUL_MAT_ID(ffn_moe_down-43): 202 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 384 us MUL_MAT_ID(ffn_moe_down-44): 187 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 402 us MUL_MAT_ID(ffn_moe_down-45): 222 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 367 us MUL_MAT_ID(ffn_moe_down-46): 202 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 385 us MUL_MAT_ID(ffn_moe_down-47): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 420 us MUL_MAT_ID(ffn_moe_down-48): 187 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 349 us MUL_MAT_ID(ffn_moe_down-49): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 461 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 428 us MUL_MAT_ID(ffn_moe_down-51): 248 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 393 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 378 us MUL_MAT_ID(ffn_moe_down-53): 191 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 430 us MUL_MAT_ID(ffn_moe_down-54): 188 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 406 us MUL_MAT_ID(ffn_moe_down-55): 249 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 420 us MUL_MAT_ID(ffn_moe_down-56): 259 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 361 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 407 us MUL_MAT_ID(ffn_moe_down-58): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 470 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 431 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 415 us MUL_MAT_ID(ffn_moe_down-61): 184 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 432 us MUL_MAT_ID(ffn_moe_down-62): 262 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 377 us MUL_MAT_ID(ffn_moe_down-63): 308 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 405 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 463 us MUL_MAT_ID(ffn_moe_down-65): 218 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 386 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 404 us MUL_MAT_ID(ffn_moe_down-67): 228 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 446 us MUL_MAT_ID(ffn_moe_down-68): 599 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 375 us MUL_MAT_ID(ffn_moe_down-69): 199 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 399 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 389 us MUL_MAT_ID(ffn_moe_down-71): 203 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 403 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 421 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 374 us MUL_MAT_ID(ffn_moe_down-74): 206 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 426 us MUL_MAT_ID(ffn_moe_down-75): 186 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 399 us MUL_MAT_ID(ffn_moe_down-76): 188 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 396 us MUL_MAT_ID(ffn_moe_down-77): 359 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 380 us MUL_MAT_ID(ffn_moe_down-78): 366 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 396 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 398 us MUL_MAT_ID(ffn_moe_down-80): 337 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 396 us MUL_MAT_ID(ffn_moe_down-81): 239 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 399 us MUL_MAT_ID(ffn_moe_down-82): 179 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 439 us MUL_MAT_ID(ffn_moe_down-83): 203 us ggml_barrier(...): 378 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 397 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 389 us MUL_MAT_ID(ffn_moe_down-85): 239 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 363 us MUL_MAT_ID(ffn_moe_down-86): 184 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 375 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 387 us MUL_MAT_ID(ffn_moe_down-88): 386 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 388 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 411 us MUL_MAT_ID(ffn_moe_down-90): 230 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 414 us MUL_MAT_ID(ffn_moe_down-91): 241 us ggml_barrier(...): 19 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 392 us MUL_MAT_ID(ffn_moe_down-25): 205 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 395 us MUL_MAT_ID(ffn_moe_down-26): 469 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 375 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 357 us MUL_MAT_ID(ffn_moe_down-28): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 391 us MUL_MAT_ID(ffn_moe_down-29): 224 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 394 us MUL_MAT_ID(ffn_moe_down-30): 200 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 422 us MUL_MAT_ID(ffn_moe_down-31): 193 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 398 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 422 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 382 us MUL_MAT_ID(ffn_moe_down-34): 560 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 395 us MUL_MAT_ID(ffn_moe_down-35): 183 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 377 us MUL_MAT_ID(ffn_moe_down-36): 300 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 431 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 426 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 363 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 402 us MUL_MAT_ID(ffn_moe_down-39): 701 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 377 us MUL_MAT_ID(ffn_moe_down-40): 218 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 434 us MUL_MAT_ID(ffn_moe_down-41): 210 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 396 us MUL_MAT_ID(ffn_moe_down-42): 302 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 475 us MUL_MAT_ID(ffn_moe_down-43): 202 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 401 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 415 us MUL_MAT_ID(ffn_moe_down-45): 186 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 440 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 365 us MUL_MAT_ID(ffn_moe_down-47): 547 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 440 us MUL_MAT_ID(ffn_moe_down-48): 185 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 368 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 397 us MUL_MAT_ID(ffn_moe_down-50): 213 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 450 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 374 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 420 us MUL_MAT_ID(ffn_moe_down-53): 183 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 375 us MUL_MAT_ID(ffn_moe_down-54): 182 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 431 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 454 us MUL_MAT_ID(ffn_moe_down-56): 722 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 424 us MUL_MAT_ID(ffn_moe_down-57): 299 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 358 us MUL_MAT_ID(ffn_moe_down-58): 186 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 416 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 383 us MUL_MAT_ID(ffn_moe_down-60): 186 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 403 us MUL_MAT_ID(ffn_moe_down-61): 213 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 428 us MUL_MAT_ID(ffn_moe_down-62): 211 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 387 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 430 us MUL_MAT_ID(ffn_moe_down-64): 382 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 374 us MUL_MAT_ID(ffn_moe_down-65): 202 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 395 us MUL_MAT_ID(ffn_moe_down-66): 182 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 414 us MUL_MAT_ID(ffn_moe_down-67): 209 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 424 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 388 us MUL_MAT_ID(ffn_moe_down-69): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 357 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 405 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 376 us MUL_MAT_ID(ffn_moe_down-72): 219 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 339 us MUL_MAT_ID(ffn_moe_down-73): 253 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 375 us MUL_MAT_ID(ffn_moe_down-74): 215 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 416 us MUL_MAT_ID(ffn_moe_down-75): 338 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 415 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 376 us MUL_MAT_ID(ffn_moe_down-77): 216 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 451 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 431 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 380 us MUL_MAT_ID(ffn_moe_down-80): 194 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 421 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 368 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 415 us MUL_MAT_ID(ffn_moe_down-83): 192 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 339 us MUL_MAT_ID(ffn_moe_down-84): 226 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 394 us MUL_MAT_ID(ffn_moe_down-85): 228 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 368 us MUL_MAT_ID(ffn_moe_down-86): 343 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 401 us MUL_MAT_ID(ffn_moe_down-87): 188 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 374 us MUL_MAT_ID(ffn_moe_down-88): 204 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 381 us MUL_MAT_ID(ffn_moe_down-89): 183 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 425 us MUL_MAT_ID(ffn_moe_down-90): 643 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 392 us MUL_MAT_ID(ffn_moe_down-91): 260 us ggml_barrier(...): 38 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 384 us MUL_MAT_ID(ffn_moe_down-25): 375 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 413 us MUL_MAT_ID(ffn_moe_down-26): 204 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 367 us MUL_MAT_ID(ffn_moe_down-27): 359 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 426 us MUL_MAT_ID(ffn_moe_down-28): 208 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 446 us MUL_MAT_ID(ffn_moe_down-29): 205 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 399 us MUL_MAT_ID(ffn_moe_down-30): 203 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 428 us MUL_MAT_ID(ffn_moe_down-31): 188 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 376 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 458 us MUL_MAT_ID(ffn_moe_down-33): 367 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 417 us MUL_MAT_ID(ffn_moe_down-34): 189 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 389 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 393 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 368 us MUL_MAT_ID(ffn_moe_down-37): 400 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 419 us MUL_MAT_ID(ffn_moe_down-38): 220 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 385 us MUL_MAT_ID(ffn_moe_down-39): 245 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 419 us MUL_MAT_ID(ffn_moe_down-40): 303 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 390 us MUL_MAT_ID(ffn_moe_down-41): 173 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 408 us MUL_MAT_ID(ffn_moe_down-42): 201 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 442 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 398 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 395 us MUL_MAT_ID(ffn_moe_down-45): 196 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 442 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 370 us MUL_MAT_ID(ffn_moe_down-47): 279 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 417 us MUL_MAT_ID(ffn_moe_down-48): 194 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 386 us MUL_MAT_ID(ffn_moe_down-49): 209 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 395 us MUL_MAT_ID(ffn_moe_down-50): 261 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 405 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 390 us MUL_MAT_ID(ffn_moe_down-52): 208 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 379 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 426 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 376 us MUL_MAT_ID(ffn_moe_down-55): 422 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 369 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 435 us MUL_MAT_ID(ffn_moe_down-57): 216 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 392 us MUL_MAT_ID(ffn_moe_down-58): 340 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 412 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 405 us MUL_MAT_ID(ffn_moe_down-60): 219 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 393 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 447 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 393 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 395 us MUL_MAT_ID(ffn_moe_down-64): 187 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 436 us MUL_MAT_ID(ffn_moe_down-65): 218 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 448 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 407 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 417 us MUL_MAT_ID(ffn_moe_down-68): 191 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 414 us MUL_MAT_ID(ffn_moe_down-69): 208 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 391 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 400 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 375 us MUL_MAT_ID(ffn_moe_down-72): 224 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 448 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 344 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 415 us MUL_MAT_ID(ffn_moe_down-75): 200 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 381 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 392 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 352 us MUL_MAT_ID(ffn_moe_down-78): 208 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 413 us MUL_MAT_ID(ffn_moe_down-79): 235 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 428 us MUL_MAT_ID(ffn_moe_down-80): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 344 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 384 us MUL_MAT_ID(ffn_moe_down-82): 191 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 344 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 391 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 415 us MUL_MAT_ID(ffn_moe_down-85): 223 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 381 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 370 us MUL_MAT_ID(ffn_moe_down-88): 189 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 378 us MUL_MAT_ID(ffn_moe_down-89): 185 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 376 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 386 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 37 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 391 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 424 us MUL_MAT_ID(ffn_moe_down-26): 210 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 414 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 382 us MUL_MAT_ID(ffn_moe_down-28): 186 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 438 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 383 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 398 us MUL_MAT_ID(ffn_moe_down-31): 199 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 457 us MUL_MAT_ID(ffn_moe_down-32): 324 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 401 us MUL_MAT_ID(ffn_moe_down-33): 204 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 503 us MUL_MAT_ID(ffn_moe_down-34): 217 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 391 us MUL_MAT_ID(ffn_moe_down-35): 458 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 491 us MUL_MAT_ID(ffn_moe_down-36): 301 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 421 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 437 us MUL_MAT_ID(ffn_moe_down-38): 209 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 409 us MUL_MAT_ID(ffn_moe_down-39): 400 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 372 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 385 us MUL_MAT_ID(ffn_moe_down-41): 189 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 403 us MUL_MAT_ID(ffn_moe_down-42): 425 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 364 us MUL_MAT_ID(ffn_moe_down-43): 212 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 490 us MUL_MAT_ID(ffn_moe_down-44): 288 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 407 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 388 us MUL_MAT_ID(ffn_moe_down-46): 198 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 403 us MUL_MAT_ID(ffn_moe_down-47): 225 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 432 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 452 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 400 us MUL_MAT_ID(ffn_moe_down-50): 229 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 363 us MUL_MAT_ID(ffn_moe_down-51): 285 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 464 us MUL_MAT_ID(ffn_moe_down-52): 396 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 429 us MUL_MAT_ID(ffn_moe_down-53): 199 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 543 us MUL_MAT_ID(ffn_moe_down-54): 201 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 461 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 376 us MUL_MAT_ID(ffn_moe_down-56): 267 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 418 us MUL_MAT_ID(ffn_moe_down-57): 245 us ggml_barrier(...): 342 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 439 us MUL_MAT_ID(ffn_moe_down-58): 228 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 420 us MUL_MAT_ID(ffn_moe_down-59): 423 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 411 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 426 us MUL_MAT_ID(ffn_moe_down-61): 285 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 400 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 418 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 350 us MUL_MAT_ID(ffn_moe_down-64): 234 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 431 us MUL_MAT_ID(ffn_moe_down-65): 251 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 439 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 342 us MUL_MAT_ID(ffn_moe_down-67): 200 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 449 us MUL_MAT_ID(ffn_moe_down-68): 277 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 381 us MUL_MAT_ID(ffn_moe_down-69): 250 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 429 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 391 us MUL_MAT_ID(ffn_moe_down-71): 215 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 395 us MUL_MAT_ID(ffn_moe_down-72): 206 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 403 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 456 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 477 us MUL_MAT_ID(ffn_moe_down-75): 203 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 378 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 407 us MUL_MAT_ID(ffn_moe_down-77): 252 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 439 us MUL_MAT_ID(ffn_moe_down-78): 172 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 410 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 349 us MUL_MAT_ID(ffn_moe_down-80): 204 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 386 us MUL_MAT_ID(ffn_moe_down-81): 229 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 438 us MUL_MAT_ID(ffn_moe_down-82): 460 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 384 us MUL_MAT_ID(ffn_moe_down-83): 222 us ggml_barrier(...): 334 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 407 us MUL_MAT_ID(ffn_moe_down-84): 195 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 399 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 379 us MUL_MAT_ID(ffn_moe_down-86): 204 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 377 us MUL_MAT_ID(ffn_moe_down-87): 242 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 409 us MUL_MAT_ID(ffn_moe_down-88): 203 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 378 us MUL_MAT_ID(ffn_moe_down-89): 438 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 373 us MUL_MAT_ID(ffn_moe_down-90): 230 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 414 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 55 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 389 us MUL_MAT_ID(ffn_moe_down-25): 205 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 419 us MUL_MAT_ID(ffn_moe_down-26): 302 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 415 us MUL_MAT_ID(ffn_moe_down-27): 204 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 383 us MUL_MAT_ID(ffn_moe_down-28): 341 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 425 us MUL_MAT_ID(ffn_moe_down-29): 335 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 394 us MUL_MAT_ID(ffn_moe_down-30): 212 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 440 us MUL_MAT_ID(ffn_moe_down-31): 175 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 427 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 448 us MUL_MAT_ID(ffn_moe_down-33): 230 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 418 us MUL_MAT_ID(ffn_moe_down-34): 216 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 396 us MUL_MAT_ID(ffn_moe_down-35): 178 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 419 us MUL_MAT_ID(ffn_moe_down-36): 213 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 445 us MUL_MAT_ID(ffn_moe_down-37): 244 us ggml_barrier(...): 509 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 438 us MUL_MAT_ID(ffn_moe_down-38): 334 us ggml_barrier(...): 338 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 468 us MUL_MAT_ID(ffn_moe_down-39): 215 us ggml_barrier(...): 391 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 406 us MUL_MAT_ID(ffn_moe_down-40): 392 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 420 us MUL_MAT_ID(ffn_moe_down-41): 193 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 383 us MUL_MAT_ID(ffn_moe_down-42): 397 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 392 us MUL_MAT_ID(ffn_moe_down-43): 201 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 417 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 419 us MUL_MAT_ID(ffn_moe_down-45): 384 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 386 us MUL_MAT_ID(ffn_moe_down-46): 198 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 441 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 466 us MUL_MAT_ID(ffn_moe_down-48): 216 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 414 us MUL_MAT_ID(ffn_moe_down-49): 199 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 549 us MUL_MAT_ID(ffn_moe_down-50): 184 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 410 us MUL_MAT_ID(ffn_moe_down-51): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 393 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 500 us MUL_MAT_ID(ffn_moe_down-53): 174 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 425 us MUL_MAT_ID(ffn_moe_down-54): 218 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 541 us MUL_MAT_ID(ffn_moe_down-55): 233 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 399 us MUL_MAT_ID(ffn_moe_down-56): 405 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 409 us MUL_MAT_ID(ffn_moe_down-57): 246 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 485 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 423 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 392 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 419 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 433 us MUL_MAT_ID(ffn_moe_down-61): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 414 us MUL_MAT_ID(ffn_moe_down-62): 396 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 395 us MUL_MAT_ID(ffn_moe_down-63): 186 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 401 us MUL_MAT_ID(ffn_moe_down-64): 244 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 365 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 403 us MUL_MAT_ID(ffn_moe_down-66): 236 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 360 us MUL_MAT_ID(ffn_moe_down-67): 207 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 373 us MUL_MAT_ID(ffn_moe_down-68): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 384 us MUL_MAT_ID(ffn_moe_down-69): 243 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 422 us MUL_MAT_ID(ffn_moe_down-70): 187 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 472 us MUL_MAT_ID(ffn_moe_down-71): 201 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 416 us MUL_MAT_ID(ffn_moe_down-72): 571 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 396 us MUL_MAT_ID(ffn_moe_down-73): 225 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 409 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 424 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 397 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 407 us MUL_MAT_ID(ffn_moe_down-77): 353 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 389 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 377 us MUL_MAT_ID(ffn_moe_down-79): 202 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 380 us MUL_MAT_ID(ffn_moe_down-80): 207 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 408 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 437 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 403 us MUL_MAT_ID(ffn_moe_down-83): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 393 us MUL_MAT_ID(ffn_moe_down-84): 222 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 380 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 385 us MUL_MAT_ID(ffn_moe_down-86): 359 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 383 us MUL_MAT_ID(ffn_moe_down-87): 169 us ggml_barrier(...): 365 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 440 us MUL_MAT_ID(ffn_moe_down-88): 196 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 362 us MUL_MAT_ID(ffn_moe_down-89): 187 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 426 us MUL_MAT_ID(ffn_moe_down-90): 238 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 419 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 31 us | glm4moe 355B.A32B Q3_K - Medium | 155.60 GiB | 369.08 B | CUDA | 94 | 8192 | 8192 | q8_0 | q8_0 | graph | 100.00 | tg128 | 10.34 ± 0.00 | Device 0: 48.875 MiB Device 1: 0 MiB Device 2: 0 MiB Device 3: 0 MiB GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 399 us MUL_MAT_ID(ffn_moe_down-25): 198 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 413 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 386 us MUL_MAT_ID(ffn_moe_down-27): 222 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 419 us MUL_MAT_ID(ffn_moe_down-28): 188 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 408 us MUL_MAT_ID(ffn_moe_down-29): 203 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 414 us MUL_MAT_ID(ffn_moe_down-30): 232 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 374 us MUL_MAT_ID(ffn_moe_down-31): 200 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 397 us MUL_MAT_ID(ffn_moe_down-32): 316 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 404 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 381 us MUL_MAT_ID(ffn_moe_down-34): 183 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 402 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 376 us MUL_MAT_ID(ffn_moe_down-36): 202 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 409 us MUL_MAT_ID(ffn_moe_down-37): 249 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 425 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 340 us MUL_MAT_ID(ffn_moe_down-39): 269 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 387 us MUL_MAT_ID(ffn_moe_down-40): 610 us ggml_barrier(...): 366 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 397 us MUL_MAT_ID(ffn_moe_down-41): 204 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 357 us MUL_MAT_ID(ffn_moe_down-42): 211 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 341 us MUL_MAT_ID(ffn_moe_down-43): 209 us ggml_barrier(...): 381 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 404 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 352 us MUL_MAT_ID(ffn_moe_down-45): 188 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 340 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 393 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 386 us MUL_MAT_ID(ffn_moe_down-48): 228 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 390 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 405 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 401 us MUL_MAT_ID(ffn_moe_down-51): 206 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 384 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 380 us MUL_MAT_ID(ffn_moe_down-53): 198 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 406 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 446 us MUL_MAT_ID(ffn_moe_down-55): 189 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 418 us MUL_MAT_ID(ffn_moe_down-56): 205 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 490 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 396 us MUL_MAT_ID(ffn_moe_down-58): 210 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 414 us MUL_MAT_ID(ffn_moe_down-59): 189 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 346 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 384 us MUL_MAT_ID(ffn_moe_down-61): 202 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 447 us MUL_MAT_ID(ffn_moe_down-62): 207 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 435 us MUL_MAT_ID(ffn_moe_down-63): 206 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 395 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 345 us MUL_MAT_ID(ffn_moe_down-65): 205 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 383 us MUL_MAT_ID(ffn_moe_down-66): 199 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 339 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 438 us MUL_MAT_ID(ffn_moe_down-68): 208 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 409 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 363 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 390 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 381 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 407 us MUL_MAT_ID(ffn_moe_down-73): 198 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 395 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 416 us MUL_MAT_ID(ffn_moe_down-75): 205 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 354 us MUL_MAT_ID(ffn_moe_down-76): 204 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 428 us MUL_MAT_ID(ffn_moe_down-77): 205 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 341 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 365 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 343 us MUL_MAT_ID(ffn_moe_down-80): 211 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 467 us MUL_MAT_ID(ffn_moe_down-82): 180 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 397 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 379 us MUL_MAT_ID(ffn_moe_down-84): 243 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 389 us MUL_MAT_ID(ffn_moe_down-85): 239 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 376 us MUL_MAT_ID(ffn_moe_down-86): 215 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 400 us MUL_MAT_ID(ffn_moe_down-87): 240 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 385 us MUL_MAT_ID(ffn_moe_down-88): 501 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 384 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 472 us MUL_MAT_ID(ffn_moe_down-90): 223 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 430 us MUL_MAT_ID(ffn_moe_down-91): 753 us ggml_barrier(...): 205 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 419 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 409 us MUL_MAT_ID(ffn_moe_down-26): 542 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 401 us MUL_MAT_ID(ffn_moe_down-27): 540 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 424 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 372 us MUL_MAT_ID(ffn_moe_down-30): 362 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 434 us MUL_MAT_ID(ffn_moe_down-31): 231 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 397 us MUL_MAT_ID(ffn_moe_down-32): 243 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 461 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 470 us MUL_MAT_ID(ffn_moe_down-34): 221 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 395 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 425 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 351 us MUL_MAT_ID(ffn_moe_down-37): 639 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 397 us MUL_MAT_ID(ffn_moe_down-38): 289 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 380 us MUL_MAT_ID(ffn_moe_down-39): 251 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 374 us MUL_MAT_ID(ffn_moe_down-40): 234 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 454 us MUL_MAT_ID(ffn_moe_down-41): 172 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 473 us MUL_MAT_ID(ffn_moe_down-42): 262 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 449 us MUL_MAT_ID(ffn_moe_down-43): 334 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 465 us MUL_MAT_ID(ffn_moe_down-44): 279 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 395 us MUL_MAT_ID(ffn_moe_down-45): 210 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 444 us MUL_MAT_ID(ffn_moe_down-46): 203 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 537 us MUL_MAT_ID(ffn_moe_down-47): 223 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 474 us MUL_MAT_ID(ffn_moe_down-48): 168 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 433 us MUL_MAT_ID(ffn_moe_down-49): 206 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 358 us MUL_MAT_ID(ffn_moe_down-50): 188 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 391 us MUL_MAT_ID(ffn_moe_down-51): 210 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 361 us MUL_MAT_ID(ffn_moe_down-52): 322 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 387 us MUL_MAT_ID(ffn_moe_down-53): 327 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 473 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 472 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 391 us MUL_MAT_ID(ffn_moe_down-56): 197 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 431 us MUL_MAT_ID(ffn_moe_down-57): 265 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 440 us MUL_MAT_ID(ffn_moe_down-58): 294 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 396 us MUL_MAT_ID(ffn_moe_down-59): 422 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 394 us MUL_MAT_ID(ffn_moe_down-60): 187 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 466 us MUL_MAT_ID(ffn_moe_down-61): 184 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 399 us MUL_MAT_ID(ffn_moe_down-62): 260 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 457 us MUL_MAT_ID(ffn_moe_down-63): 202 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 485 us MUL_MAT_ID(ffn_moe_down-64): 307 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 407 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 418 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 383 us MUL_MAT_ID(ffn_moe_down-67): 291 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 416 us MUL_MAT_ID(ffn_moe_down-68): 228 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 448 us MUL_MAT_ID(ffn_moe_down-69): 179 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 397 us MUL_MAT_ID(ffn_moe_down-70): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 374 us MUL_MAT_ID(ffn_moe_down-71): 167 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 403 us MUL_MAT_ID(ffn_moe_down-72): 300 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 449 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 413 us MUL_MAT_ID(ffn_moe_down-74): 222 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 391 us MUL_MAT_ID(ffn_moe_down-75): 243 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 427 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 381 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 376 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 339 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 359 us MUL_MAT_ID(ffn_moe_down-79): 169 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 450 us MUL_MAT_ID(ffn_moe_down-80): 201 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 441 us MUL_MAT_ID(ffn_moe_down-81): 283 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 431 us MUL_MAT_ID(ffn_moe_down-82): 202 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 481 us MUL_MAT_ID(ffn_moe_down-83): 289 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 491 us MUL_MAT_ID(ffn_moe_down-84): 254 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 438 us MUL_MAT_ID(ffn_moe_down-85): 227 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 505 us MUL_MAT_ID(ffn_moe_down-86): 226 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 466 us MUL_MAT_ID(ffn_moe_down-87): 181 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 489 us MUL_MAT_ID(ffn_moe_down-88): 385 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 405 us MUL_MAT_ID(ffn_moe_down-89): 481 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 460 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 416 us MUL_MAT_ID(ffn_moe_down-91): 222 us ggml_barrier(...): 346 us GET_ROWS(inp_embd): 11 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 394 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 430 us MUL_MAT_ID(ffn_moe_down-26): 257 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 415 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 480 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 377 us MUL_MAT_ID(ffn_moe_down-29): 250 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 364 us MUL_MAT_ID(ffn_moe_down-30): 183 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 423 us MUL_MAT_ID(ffn_moe_down-31): 219 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 396 us MUL_MAT_ID(ffn_moe_down-32): 310 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 395 us MUL_MAT_ID(ffn_moe_down-33): 217 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 475 us MUL_MAT_ID(ffn_moe_down-34): 240 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 407 us MUL_MAT_ID(ffn_moe_down-35): 231 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 394 us MUL_MAT_ID(ffn_moe_down-36): 181 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 386 us MUL_MAT_ID(ffn_moe_down-37): 242 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 406 us MUL_MAT_ID(ffn_moe_down-38): 189 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 486 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 434 us MUL_MAT_ID(ffn_moe_down-40): 234 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 458 us MUL_MAT_ID(ffn_moe_down-41): 190 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 393 us MUL_MAT_ID(ffn_moe_down-42): 216 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 424 us MUL_MAT_ID(ffn_moe_down-43): 182 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 421 us MUL_MAT_ID(ffn_moe_down-44): 241 us ggml_barrier(...): 352 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 384 us MUL_MAT_ID(ffn_moe_down-45): 476 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 397 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 423 us MUL_MAT_ID(ffn_moe_down-47): 237 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 450 us MUL_MAT_ID(ffn_moe_down-48): 571 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 402 us MUL_MAT_ID(ffn_moe_down-49): 209 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 513 us MUL_MAT_ID(ffn_moe_down-50): 182 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 461 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 408 us MUL_MAT_ID(ffn_moe_down-52): 231 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 401 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 360 us MUL_MAT_ID(ffn_moe_down-54): 311 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 416 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 184 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 460 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 393 us MUL_MAT_ID(ffn_moe_down-58): 213 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 390 us MUL_MAT_ID(ffn_moe_down-59): 306 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 393 us MUL_MAT_ID(ffn_moe_down-60): 178 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 400 us MUL_MAT_ID(ffn_moe_down-61): 398 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 379 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 449 us MUL_MAT_ID(ffn_moe_down-63): 210 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 411 us MUL_MAT_ID(ffn_moe_down-64): 405 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 383 us MUL_MAT_ID(ffn_moe_down-65): 199 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 424 us MUL_MAT_ID(ffn_moe_down-66): 206 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 549 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 482 us MUL_MAT_ID(ffn_moe_down-68): 186 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 398 us MUL_MAT_ID(ffn_moe_down-69): 182 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 376 us MUL_MAT_ID(ffn_moe_down-70): 241 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 418 us MUL_MAT_ID(ffn_moe_down-71): 680 us ggml_barrier(...): 351 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 356 us MUL_MAT_ID(ffn_moe_down-72): 240 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 490 us MUL_MAT_ID(ffn_moe_down-73): 177 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 461 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 408 us MUL_MAT_ID(ffn_moe_down-75): 335 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 438 us MUL_MAT_ID(ffn_moe_down-76): 330 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 388 us MUL_MAT_ID(ffn_moe_down-77): 409 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 375 us MUL_MAT_ID(ffn_moe_down-78): 209 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 418 us MUL_MAT_ID(ffn_moe_down-79): 246 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 396 us MUL_MAT_ID(ffn_moe_down-80): 215 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 430 us MUL_MAT_ID(ffn_moe_down-81): 196 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 399 us MUL_MAT_ID(ffn_moe_down-82): 220 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 489 us MUL_MAT_ID(ffn_moe_down-83): 188 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 445 us MUL_MAT_ID(ffn_moe_down-84): 303 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 396 us MUL_MAT_ID(ffn_moe_down-85): 265 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 402 us MUL_MAT_ID(ffn_moe_down-86): 174 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 485 us MUL_MAT_ID(ffn_moe_down-87): 352 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 397 us MUL_MAT_ID(ffn_moe_down-88): 207 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 420 us MUL_MAT_ID(ffn_moe_down-89): 243 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 413 us MUL_MAT_ID(ffn_moe_down-90): 600 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 444 us MUL_MAT_ID(ffn_moe_down-91): 222 us ggml_barrier(...): 206 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 415 us MUL_MAT_ID(ffn_moe_down-25): 569 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 499 us MUL_MAT_ID(ffn_moe_down-26): 283 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 442 us MUL_MAT_ID(ffn_moe_down-27): 227 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 390 us MUL_MAT_ID(ffn_moe_down-28): 605 us ggml_barrier(...): 370 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 459 us MUL_MAT_ID(ffn_moe_down-29): 215 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 394 us MUL_MAT_ID(ffn_moe_down-30): 398 us ggml_barrier(...): 388 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 411 us MUL_MAT_ID(ffn_moe_down-31): 250 us ggml_barrier(...): 380 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 400 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 427 us MUL_MAT_ID(ffn_moe_down-33): 321 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 433 us MUL_MAT_ID(ffn_moe_down-34): 234 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 409 us MUL_MAT_ID(ffn_moe_down-35): 331 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 393 us MUL_MAT_ID(ffn_moe_down-36): 187 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 475 us MUL_MAT_ID(ffn_moe_down-37): 296 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 398 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 403 us MUL_MAT_ID(ffn_moe_down-39): 248 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 404 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 410 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 454 us MUL_MAT_ID(ffn_moe_down-42): 280 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 391 us MUL_MAT_ID(ffn_moe_down-43): 201 us ggml_barrier(...): 462 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 405 us MUL_MAT_ID(ffn_moe_down-44): 208 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 443 us MUL_MAT_ID(ffn_moe_down-45): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 455 us MUL_MAT_ID(ffn_moe_down-46): 679 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 416 us MUL_MAT_ID(ffn_moe_down-47): 412 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 370 us MUL_MAT_ID(ffn_moe_down-48): 202 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 396 us MUL_MAT_ID(ffn_moe_down-49): 197 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 408 us MUL_MAT_ID(ffn_moe_down-50): 309 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 558 us MUL_MAT_ID(ffn_moe_down-51): 221 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 388 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 429 us MUL_MAT_ID(ffn_moe_down-53): 312 us ggml_barrier(...): 439 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 383 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 374 us MUL_MAT_ID(ffn_moe_down-55): 203 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 418 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 399 us MUL_MAT_ID(ffn_moe_down-57): 189 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 400 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 376 us MUL_MAT_ID(ffn_moe_down-59): 190 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 452 us MUL_MAT_ID(ffn_moe_down-60): 197 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 401 us MUL_MAT_ID(ffn_moe_down-61): 186 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 510 us MUL_MAT_ID(ffn_moe_down-62): 292 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 381 us MUL_MAT_ID(ffn_moe_down-63): 501 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 350 us MUL_MAT_ID(ffn_moe_down-64): 243 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 370 us MUL_MAT_ID(ffn_moe_down-65): 334 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 406 us MUL_MAT_ID(ffn_moe_down-66): 440 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 434 us MUL_MAT_ID(ffn_moe_down-67): 188 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 393 us MUL_MAT_ID(ffn_moe_down-68): 202 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 420 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 340 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 378 us MUL_MAT_ID(ffn_moe_down-70): 169 us ggml_barrier(...): 382 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 408 us MUL_MAT_ID(ffn_moe_down-71): 208 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 365 us MUL_MAT_ID(ffn_moe_down-72): 189 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 383 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 482 us MUL_MAT_ID(ffn_moe_down-74): 170 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 388 us MUL_MAT_ID(ffn_moe_down-75): 207 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 369 us MUL_MAT_ID(ffn_moe_down-76): 172 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 405 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 470 us MUL_MAT_ID(ffn_moe_down-78): 176 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 430 us MUL_MAT_ID(ffn_moe_down-79): 267 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 421 us MUL_MAT_ID(ffn_moe_down-80): 373 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 400 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 355 us MUL_MAT_ID(ffn_moe_down-82): 183 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 387 us MUL_MAT_ID(ffn_moe_down-83): 282 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 451 us MUL_MAT_ID(ffn_moe_down-84): 247 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 368 us MUL_MAT_ID(ffn_moe_down-85): 452 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 401 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 408 us MUL_MAT_ID(ffn_moe_down-87): 220 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 428 us MUL_MAT_ID(ffn_moe_down-88): 291 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 394 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 340 us MUL_MAT_ID(ffn_moe_down-90): 237 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 396 us MUL_MAT_ID(ffn_moe_down-91): 233 us ggml_barrier(...): 44 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 442 us MUL_MAT_ID(ffn_moe_down-25): 172 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 397 us MUL_MAT_ID(ffn_moe_down-26): 236 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 386 us MUL_MAT_ID(ffn_moe_down-27): 219 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 403 us MUL_MAT_ID(ffn_moe_down-28): 224 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 392 us MUL_MAT_ID(ffn_moe_down-29): 205 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 462 us MUL_MAT_ID(ffn_moe_down-30): 223 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 386 us MUL_MAT_ID(ffn_moe_down-31): 586 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 409 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 377 us MUL_MAT_ID(ffn_moe_down-33): 169 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 375 us MUL_MAT_ID(ffn_moe_down-34): 543 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 400 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 417 us MUL_MAT_ID(ffn_moe_down-36): 475 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 406 us MUL_MAT_ID(ffn_moe_down-37): 262 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 354 us MUL_MAT_ID(ffn_moe_down-38): 255 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 421 us MUL_MAT_ID(ffn_moe_down-39): 239 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 413 us MUL_MAT_ID(ffn_moe_down-40): 247 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 397 us MUL_MAT_ID(ffn_moe_down-41): 399 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 438 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 449 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 462 us MUL_MAT_ID(ffn_moe_down-44): 181 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 412 us MUL_MAT_ID(ffn_moe_down-45): 192 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 349 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 456 us MUL_MAT_ID(ffn_moe_down-47): 302 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 391 us MUL_MAT_ID(ffn_moe_down-48): 623 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 408 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 386 us MUL_MAT_ID(ffn_moe_down-50): 576 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 419 us MUL_MAT_ID(ffn_moe_down-51): 265 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 383 us MUL_MAT_ID(ffn_moe_down-52): 294 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 455 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 421 us MUL_MAT_ID(ffn_moe_down-54): 221 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 427 us MUL_MAT_ID(ffn_moe_down-55): 237 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 420 us MUL_MAT_ID(ffn_moe_down-56): 252 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 479 us MUL_MAT_ID(ffn_moe_down-57): 226 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 404 us MUL_MAT_ID(ffn_moe_down-58): 386 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 403 us MUL_MAT_ID(ffn_moe_down-59): 466 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 371 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 351 us MUL_MAT_ID(ffn_moe_down-61): 246 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 421 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 392 us MUL_MAT_ID(ffn_moe_down-63): 714 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 383 us MUL_MAT_ID(ffn_moe_down-64): 275 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 411 us MUL_MAT_ID(ffn_moe_down-65): 302 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 349 us MUL_MAT_ID(ffn_moe_down-66): 218 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 479 us MUL_MAT_ID(ffn_moe_down-67): 204 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 379 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 393 us MUL_MAT_ID(ffn_moe_down-69): 200 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 404 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 380 us MUL_MAT_ID(ffn_moe_down-71): 254 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 393 us MUL_MAT_ID(ffn_moe_down-72): 250 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 396 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 394 us MUL_MAT_ID(ffn_moe_down-74): 192 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 365 us MUL_MAT_ID(ffn_moe_down-75): 267 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 394 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 399 us MUL_MAT_ID(ffn_moe_down-77): 204 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 349 us MUL_MAT_ID(ffn_moe_down-78): 425 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 461 us MUL_MAT_ID(ffn_moe_down-79): 324 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 396 us MUL_MAT_ID(ffn_moe_down-80): 202 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 392 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 426 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 514 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 398 us MUL_MAT_ID(ffn_moe_down-85): 216 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 540 us MUL_MAT_ID(ffn_moe_down-86): 227 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 453 us MUL_MAT_ID(ffn_moe_down-87): 161 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 387 us MUL_MAT_ID(ffn_moe_down-88): 388 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 398 us MUL_MAT_ID(ffn_moe_down-89): 490 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 364 us MUL_MAT_ID(ffn_moe_down-90): 240 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 542 us MUL_MAT_ID(ffn_moe_down-91): 271 us ggml_barrier(...): 101 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 186 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 391 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 445 us MUL_MAT_ID(ffn_moe_down-27): 202 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 466 us MUL_MAT_ID(ffn_moe_down-28): 253 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 394 us MUL_MAT_ID(ffn_moe_down-29): 216 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 433 us MUL_MAT_ID(ffn_moe_down-30): 203 us ggml_barrier(...): 363 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 380 us MUL_MAT_ID(ffn_moe_down-31): 232 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 417 us MUL_MAT_ID(ffn_moe_down-32): 180 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 434 us MUL_MAT_ID(ffn_moe_down-33): 299 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 448 us MUL_MAT_ID(ffn_moe_down-34): 232 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 387 us MUL_MAT_ID(ffn_moe_down-35): 218 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 393 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 384 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 421 us MUL_MAT_ID(ffn_moe_down-38): 185 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 384 us MUL_MAT_ID(ffn_moe_down-39): 222 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 411 us MUL_MAT_ID(ffn_moe_down-40): 314 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 387 us MUL_MAT_ID(ffn_moe_down-41): 186 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 372 us MUL_MAT_ID(ffn_moe_down-42): 201 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 390 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 414 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 374 us MUL_MAT_ID(ffn_moe_down-45): 197 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 403 us MUL_MAT_ID(ffn_moe_down-46): 258 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 455 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 489 us MUL_MAT_ID(ffn_moe_down-48): 191 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 404 us MUL_MAT_ID(ffn_moe_down-49): 197 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 381 us MUL_MAT_ID(ffn_moe_down-50): 199 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 374 us MUL_MAT_ID(ffn_moe_down-51): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 377 us MUL_MAT_ID(ffn_moe_down-52): 199 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 398 us MUL_MAT_ID(ffn_moe_down-53): 185 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 458 us MUL_MAT_ID(ffn_moe_down-54): 221 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 378 us MUL_MAT_ID(ffn_moe_down-55): 194 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 473 us MUL_MAT_ID(ffn_moe_down-56): 185 us ggml_barrier(...): 343 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 434 us MUL_MAT_ID(ffn_moe_down-57): 270 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 397 us MUL_MAT_ID(ffn_moe_down-58): 532 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 404 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 389 us MUL_MAT_ID(ffn_moe_down-60): 207 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 466 us MUL_MAT_ID(ffn_moe_down-61): 538 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 378 us MUL_MAT_ID(ffn_moe_down-62): 187 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 424 us MUL_MAT_ID(ffn_moe_down-63): 204 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 427 us MUL_MAT_ID(ffn_moe_down-64): 204 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 448 us MUL_MAT_ID(ffn_moe_down-65): 196 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 351 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 373 us MUL_MAT_ID(ffn_moe_down-67): 189 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 569 us MUL_MAT_ID(ffn_moe_down-68): 198 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 454 us MUL_MAT_ID(ffn_moe_down-69): 184 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 398 us MUL_MAT_ID(ffn_moe_down-70): 431 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 518 us MUL_MAT_ID(ffn_moe_down-71): 280 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 429 us MUL_MAT_ID(ffn_moe_down-72): 308 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 355 us MUL_MAT_ID(ffn_moe_down-73): 221 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 398 us MUL_MAT_ID(ffn_moe_down-74): 223 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 449 us MUL_MAT_ID(ffn_moe_down-75): 373 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 394 us MUL_MAT_ID(ffn_moe_down-76): 407 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 411 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 446 us MUL_MAT_ID(ffn_moe_down-78): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 442 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 469 us MUL_MAT_ID(ffn_moe_down-80): 566 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 450 us MUL_MAT_ID(ffn_moe_down-81): 290 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 384 us MUL_MAT_ID(ffn_moe_down-82): 254 us ggml_barrier(...): 498 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 407 us MUL_MAT_ID(ffn_moe_down-83): 247 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 399 us MUL_MAT_ID(ffn_moe_down-84): 659 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 438 us MUL_MAT_ID(ffn_moe_down-85): 251 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 397 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 466 us MUL_MAT_ID(ffn_moe_down-87): 292 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 423 us MUL_MAT_ID(ffn_moe_down-88): 268 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 423 us MUL_MAT_ID(ffn_moe_down-89): 215 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 423 us MUL_MAT_ID(ffn_moe_down-90): 205 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 423 us MUL_MAT_ID(ffn_moe_down-91): 213 us ggml_barrier(...): 224 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 379 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 421 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 368 us MUL_MAT_ID(ffn_moe_down-27): 222 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 389 us MUL_MAT_ID(ffn_moe_down-28): 303 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 385 us MUL_MAT_ID(ffn_moe_down-29): 405 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 219 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 399 us MUL_MAT_ID(ffn_moe_down-31): 375 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 391 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 336 us MUL_MAT_ID(ffn_moe_down-33): 544 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 384 us MUL_MAT_ID(ffn_moe_down-34): 178 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 434 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 400 us MUL_MAT_ID(ffn_moe_down-36): 195 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 411 us MUL_MAT_ID(ffn_moe_down-37): 243 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 397 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 390 us MUL_MAT_ID(ffn_moe_down-39): 246 us ggml_barrier(...): 464 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 396 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 390 us MUL_MAT_ID(ffn_moe_down-41): 213 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 460 us MUL_MAT_ID(ffn_moe_down-42): 203 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 425 us MUL_MAT_ID(ffn_moe_down-43): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 427 us MUL_MAT_ID(ffn_moe_down-44): 285 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 492 us MUL_MAT_ID(ffn_moe_down-45): 177 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 383 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 451 us MUL_MAT_ID(ffn_moe_down-47): 389 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 368 us MUL_MAT_ID(ffn_moe_down-48): 205 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 384 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 394 us MUL_MAT_ID(ffn_moe_down-50): 188 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 391 us MUL_MAT_ID(ffn_moe_down-51): 345 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 353 us MUL_MAT_ID(ffn_moe_down-52): 203 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 389 us MUL_MAT_ID(ffn_moe_down-53): 205 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 440 us MUL_MAT_ID(ffn_moe_down-54): 210 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 434 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 410 us MUL_MAT_ID(ffn_moe_down-56): 205 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 439 us MUL_MAT_ID(ffn_moe_down-57): 530 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 393 us MUL_MAT_ID(ffn_moe_down-58): 217 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 365 us MUL_MAT_ID(ffn_moe_down-59): 249 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 417 us MUL_MAT_ID(ffn_moe_down-60): 184 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 490 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 403 us MUL_MAT_ID(ffn_moe_down-62): 209 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 402 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 345 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 383 us MUL_MAT_ID(ffn_moe_down-65): 200 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 343 us MUL_MAT_ID(ffn_moe_down-66): 435 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 395 us MUL_MAT_ID(ffn_moe_down-67): 200 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 398 us MUL_MAT_ID(ffn_moe_down-68): 207 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 438 us MUL_MAT_ID(ffn_moe_down-69): 353 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 419 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 355 us MUL_MAT_ID(ffn_moe_down-71): 191 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 383 us MUL_MAT_ID(ffn_moe_down-72): 358 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 376 us MUL_MAT_ID(ffn_moe_down-73): 183 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 455 us MUL_MAT_ID(ffn_moe_down-74): 204 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 420 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 342 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 388 us MUL_MAT_ID(ffn_moe_down-76): 421 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 435 us MUL_MAT_ID(ffn_moe_down-77): 206 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 421 us MUL_MAT_ID(ffn_moe_down-78): 204 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 433 us MUL_MAT_ID(ffn_moe_down-79): 573 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 385 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 434 us MUL_MAT_ID(ffn_moe_down-81): 200 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 417 us MUL_MAT_ID(ffn_moe_down-82): 363 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 448 us MUL_MAT_ID(ffn_moe_down-83): 184 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 428 us MUL_MAT_ID(ffn_moe_down-84): 250 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 390 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 406 us MUL_MAT_ID(ffn_moe_down-86): 182 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 362 us MUL_MAT_ID(ffn_moe_down-87): 219 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 436 us MUL_MAT_ID(ffn_moe_down-88): 217 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 432 us MUL_MAT_ID(ffn_moe_down-89): 203 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 365 us MUL_MAT_ID(ffn_moe_down-90): 247 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 401 us MUL_MAT_ID(ffn_moe_down-91): 472 us ggml_barrier(...): 62 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 402 us MUL_MAT_ID(ffn_moe_down-25): 366 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 389 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 379 us MUL_MAT_ID(ffn_moe_down-27): 337 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 394 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 398 us MUL_MAT_ID(ffn_moe_down-29): 208 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 351 us MUL_MAT_ID(ffn_moe_down-30): 201 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 392 us MUL_MAT_ID(ffn_moe_down-31): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 422 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 401 us MUL_MAT_ID(ffn_moe_down-33): 292 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 360 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 399 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 415 us MUL_MAT_ID(ffn_moe_down-36): 313 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 379 us MUL_MAT_ID(ffn_moe_down-37): 389 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 390 us MUL_MAT_ID(ffn_moe_down-38): 228 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 372 us MUL_MAT_ID(ffn_moe_down-39): 272 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 405 us MUL_MAT_ID(ffn_moe_down-40): 749 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 372 us MUL_MAT_ID(ffn_moe_down-41): 210 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 346 us MUL_MAT_ID(ffn_moe_down-42): 203 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 403 us MUL_MAT_ID(ffn_moe_down-43): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 376 us MUL_MAT_ID(ffn_moe_down-44): 213 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 414 us MUL_MAT_ID(ffn_moe_down-45): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 390 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 380 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 385 us MUL_MAT_ID(ffn_moe_down-48): 209 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 414 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 374 us MUL_MAT_ID(ffn_moe_down-50): 204 us ggml_barrier(...): 348 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 394 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 393 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 413 us MUL_MAT_ID(ffn_moe_down-53): 195 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 402 us MUL_MAT_ID(ffn_moe_down-54): 190 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 353 us MUL_MAT_ID(ffn_moe_down-55): 234 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 383 us MUL_MAT_ID(ffn_moe_down-56): 201 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 437 us MUL_MAT_ID(ffn_moe_down-57): 183 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 382 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 410 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 342 us MUL_MAT_ID(ffn_moe_down-60): 201 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 414 us MUL_MAT_ID(ffn_moe_down-61): 243 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 431 us MUL_MAT_ID(ffn_moe_down-62): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 384 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 376 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 390 us MUL_MAT_ID(ffn_moe_down-65): 546 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 410 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 359 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 392 us MUL_MAT_ID(ffn_moe_down-67): 204 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 202 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 382 us MUL_MAT_ID(ffn_moe_down-69): 201 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 416 us MUL_MAT_ID(ffn_moe_down-70): 215 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 401 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 350 us MUL_MAT_ID(ffn_moe_down-72): 524 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 354 us MUL_MAT_ID(ffn_moe_down-73): 197 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 419 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 391 us MUL_MAT_ID(ffn_moe_down-75): 262 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 380 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 406 us MUL_MAT_ID(ffn_moe_down-77): 240 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 432 us MUL_MAT_ID(ffn_moe_down-78): 205 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 394 us MUL_MAT_ID(ffn_moe_down-79): 223 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 434 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 357 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 423 us MUL_MAT_ID(ffn_moe_down-82): 189 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 392 us MUL_MAT_ID(ffn_moe_down-83): 191 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 370 us MUL_MAT_ID(ffn_moe_down-84): 284 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 426 us MUL_MAT_ID(ffn_moe_down-85): 732 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 409 us MUL_MAT_ID(ffn_moe_down-86): 214 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 392 us MUL_MAT_ID(ffn_moe_down-87): 183 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 440 us MUL_MAT_ID(ffn_moe_down-88): 476 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 392 us MUL_MAT_ID(ffn_moe_down-89): 183 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 467 us MUL_MAT_ID(ffn_moe_down-90): 244 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 378 us MUL_MAT_ID(ffn_moe_down-91): 696 us ggml_barrier(...): 42 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 376 us MUL_MAT_ID(ffn_moe_down-25): 225 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 446 us MUL_MAT_ID(ffn_moe_down-26): 231 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 461 us MUL_MAT_ID(ffn_moe_down-27): 203 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 412 us MUL_MAT_ID(ffn_moe_down-28): 213 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 386 us MUL_MAT_ID(ffn_moe_down-29): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 382 us MUL_MAT_ID(ffn_moe_down-30): 201 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 344 us MUL_MAT_ID(ffn_moe_down-31): 219 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 395 us MUL_MAT_ID(ffn_moe_down-32): 180 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 441 us MUL_MAT_ID(ffn_moe_down-33): 270 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 394 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 398 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 210 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 379 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 424 us MUL_MAT_ID(ffn_moe_down-38): 414 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 378 us MUL_MAT_ID(ffn_moe_down-39): 227 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 364 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 421 us MUL_MAT_ID(ffn_moe_down-41): 237 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 383 us MUL_MAT_ID(ffn_moe_down-42): 214 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 417 us MUL_MAT_ID(ffn_moe_down-43): 608 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 402 us MUL_MAT_ID(ffn_moe_down-44): 202 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 425 us MUL_MAT_ID(ffn_moe_down-45): 209 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 416 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 382 us MUL_MAT_ID(ffn_moe_down-47): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 343 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 381 us MUL_MAT_ID(ffn_moe_down-49): 230 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 378 us MUL_MAT_ID(ffn_moe_down-50): 184 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 378 us MUL_MAT_ID(ffn_moe_down-51): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 477 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 401 us MUL_MAT_ID(ffn_moe_down-53): 191 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 420 us MUL_MAT_ID(ffn_moe_down-54): 185 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 409 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 481 us MUL_MAT_ID(ffn_moe_down-56): 207 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 398 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 422 us MUL_MAT_ID(ffn_moe_down-58): 357 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 412 us MUL_MAT_ID(ffn_moe_down-59): 205 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 412 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 372 us MUL_MAT_ID(ffn_moe_down-61): 380 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 364 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 395 us MUL_MAT_ID(ffn_moe_down-63): 187 us ggml_barrier(...): 439 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 392 us MUL_MAT_ID(ffn_moe_down-64): 200 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 371 us MUL_MAT_ID(ffn_moe_down-65): 325 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 435 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 420 us MUL_MAT_ID(ffn_moe_down-67): 244 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 408 us MUL_MAT_ID(ffn_moe_down-68): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 421 us MUL_MAT_ID(ffn_moe_down-69): 194 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 398 us MUL_MAT_ID(ffn_moe_down-70): 251 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 412 us MUL_MAT_ID(ffn_moe_down-71): 205 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 437 us MUL_MAT_ID(ffn_moe_down-72): 206 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 362 us MUL_MAT_ID(ffn_moe_down-73): 180 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 421 us MUL_MAT_ID(ffn_moe_down-74): 173 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 369 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 419 us MUL_MAT_ID(ffn_moe_down-76): 178 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 391 us MUL_MAT_ID(ffn_moe_down-77): 214 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 456 us MUL_MAT_ID(ffn_moe_down-78): 188 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 431 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 373 us MUL_MAT_ID(ffn_moe_down-80): 253 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 457 us MUL_MAT_ID(ffn_moe_down-81): 209 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 447 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 380 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 351 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 442 us MUL_MAT_ID(ffn_moe_down-85): 227 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 424 us MUL_MAT_ID(ffn_moe_down-86): 589 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 387 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 385 us MUL_MAT_ID(ffn_moe_down-88): 257 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 402 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 373 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 399 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 55 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 425 us MUL_MAT_ID(ffn_moe_down-25): 387 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 373 us MUL_MAT_ID(ffn_moe_down-26): 180 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 373 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 363 us MUL_MAT_ID(ffn_moe_down-28): 215 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 439 us MUL_MAT_ID(ffn_moe_down-29): 191 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 391 us MUL_MAT_ID(ffn_moe_down-30): 203 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 355 us MUL_MAT_ID(ffn_moe_down-31): 193 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 364 us MUL_MAT_ID(ffn_moe_down-32): 215 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 396 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 385 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 409 us MUL_MAT_ID(ffn_moe_down-35): 202 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 377 us MUL_MAT_ID(ffn_moe_down-36): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 398 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 416 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 361 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 378 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 495 us MUL_MAT_ID(ffn_moe_down-41): 215 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 399 us MUL_MAT_ID(ffn_moe_down-42): 204 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 448 us MUL_MAT_ID(ffn_moe_down-43): 213 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 382 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 413 us MUL_MAT_ID(ffn_moe_down-45): 214 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 362 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 464 us MUL_MAT_ID(ffn_moe_down-47): 284 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 409 us MUL_MAT_ID(ffn_moe_down-48): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 389 us MUL_MAT_ID(ffn_moe_down-49): 553 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 396 us MUL_MAT_ID(ffn_moe_down-50): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 374 us MUL_MAT_ID(ffn_moe_down-51): 546 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 396 us MUL_MAT_ID(ffn_moe_down-52): 210 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 378 us MUL_MAT_ID(ffn_moe_down-53): 187 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 363 us MUL_MAT_ID(ffn_moe_down-54): 220 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 369 us MUL_MAT_ID(ffn_moe_down-55): 216 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 404 us MUL_MAT_ID(ffn_moe_down-56): 216 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 410 us MUL_MAT_ID(ffn_moe_down-57): 198 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 396 us MUL_MAT_ID(ffn_moe_down-58): 266 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 416 us MUL_MAT_ID(ffn_moe_down-59): 190 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 422 us MUL_MAT_ID(ffn_moe_down-60): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 360 us MUL_MAT_ID(ffn_moe_down-61): 182 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 383 us MUL_MAT_ID(ffn_moe_down-62): 213 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 390 us MUL_MAT_ID(ffn_moe_down-63): 201 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 392 us MUL_MAT_ID(ffn_moe_down-64): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 365 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 409 us MUL_MAT_ID(ffn_moe_down-66): 219 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 404 us MUL_MAT_ID(ffn_moe_down-67): 196 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 416 us MUL_MAT_ID(ffn_moe_down-68): 384 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 390 us MUL_MAT_ID(ffn_moe_down-69): 298 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 405 us MUL_MAT_ID(ffn_moe_down-70): 191 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 395 us MUL_MAT_ID(ffn_moe_down-71): 437 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 348 us MUL_MAT_ID(ffn_moe_down-72): 213 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 403 us MUL_MAT_ID(ffn_moe_down-73): 423 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 409 us MUL_MAT_ID(ffn_moe_down-74): 208 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 430 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 419 us MUL_MAT_ID(ffn_moe_down-76): 199 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 442 us MUL_MAT_ID(ffn_moe_down-77): 208 us ggml_barrier(...): 539 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 388 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 388 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 401 us MUL_MAT_ID(ffn_moe_down-79): 203 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 408 us MUL_MAT_ID(ffn_moe_down-80): 206 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 445 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 374 us MUL_MAT_ID(ffn_moe_down-82): 230 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 383 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 395 us MUL_MAT_ID(ffn_moe_down-84): 632 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 358 us MUL_MAT_ID(ffn_moe_down-85): 289 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 400 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 440 us MUL_MAT_ID(ffn_moe_down-87): 371 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 392 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 339 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 372 us MUL_MAT_ID(ffn_moe_down-89): 191 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 414 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 382 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 47 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 352 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 375 us MUL_MAT_ID(ffn_moe_down-26): 370 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 382 us MUL_MAT_ID(ffn_moe_down-27): 583 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 391 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 432 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 413 us MUL_MAT_ID(ffn_moe_down-30): 215 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 424 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 378 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 394 us MUL_MAT_ID(ffn_moe_down-33): 319 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 390 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 402 us MUL_MAT_ID(ffn_moe_down-35): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 407 us MUL_MAT_ID(ffn_moe_down-36): 307 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 383 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 404 us MUL_MAT_ID(ffn_moe_down-38): 234 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 382 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 382 us MUL_MAT_ID(ffn_moe_down-40): 520 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 388 us MUL_MAT_ID(ffn_moe_down-41): 445 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 387 us MUL_MAT_ID(ffn_moe_down-42): 209 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 401 us MUL_MAT_ID(ffn_moe_down-43): 184 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 397 us MUL_MAT_ID(ffn_moe_down-44): 177 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 431 us MUL_MAT_ID(ffn_moe_down-45): 209 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 420 us MUL_MAT_ID(ffn_moe_down-46): 306 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 417 us MUL_MAT_ID(ffn_moe_down-47): 205 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 354 us MUL_MAT_ID(ffn_moe_down-48): 202 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 415 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 369 us MUL_MAT_ID(ffn_moe_down-50): 215 us ggml_barrier(...): 401 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 377 us MUL_MAT_ID(ffn_moe_down-51): 203 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 420 us MUL_MAT_ID(ffn_moe_down-52): 210 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 379 us MUL_MAT_ID(ffn_moe_down-53): 230 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 408 us MUL_MAT_ID(ffn_moe_down-54): 571 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 412 us MUL_MAT_ID(ffn_moe_down-55): 193 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 412 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 371 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 394 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 377 us MUL_MAT_ID(ffn_moe_down-59): 204 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 451 us MUL_MAT_ID(ffn_moe_down-60): 202 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 361 us MUL_MAT_ID(ffn_moe_down-61): 574 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 397 us MUL_MAT_ID(ffn_moe_down-62): 222 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 408 us MUL_MAT_ID(ffn_moe_down-63): 185 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 409 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 354 us MUL_MAT_ID(ffn_moe_down-65): 440 us ggml_barrier(...): 340 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 399 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 345 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 427 us MUL_MAT_ID(ffn_moe_down-68): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 368 us MUL_MAT_ID(ffn_moe_down-69): 187 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 440 us MUL_MAT_ID(ffn_moe_down-70): 188 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 451 us MUL_MAT_ID(ffn_moe_down-71): 267 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 413 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 405 us MUL_MAT_ID(ffn_moe_down-73): 301 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 420 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 421 us MUL_MAT_ID(ffn_moe_down-75): 209 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 459 us MUL_MAT_ID(ffn_moe_down-76): 215 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 492 us MUL_MAT_ID(ffn_moe_down-77): 174 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 395 us MUL_MAT_ID(ffn_moe_down-78): 205 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 402 us MUL_MAT_ID(ffn_moe_down-79): 301 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 405 us MUL_MAT_ID(ffn_moe_down-80): 257 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 490 us MUL_MAT_ID(ffn_moe_down-81): 220 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 473 us MUL_MAT_ID(ffn_moe_down-82): 309 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 352 us MUL_MAT_ID(ffn_moe_down-83): 227 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 462 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 481 us MUL_MAT_ID(ffn_moe_down-85): 241 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 375 us MUL_MAT_ID(ffn_moe_down-86): 183 us ggml_barrier(...): 533 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 410 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 451 us MUL_MAT_ID(ffn_moe_down-88): 222 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 456 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 425 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 529 us MUL_MAT_ID(ffn_moe_down-91): 291 us ggml_barrier(...): 68 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 399 us MUL_MAT_ID(ffn_moe_down-25): 441 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 395 us MUL_MAT_ID(ffn_moe_down-26): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 349 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 353 us MUL_MAT_ID(ffn_moe_down-28): 285 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 206 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 431 us MUL_MAT_ID(ffn_moe_down-30): 226 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 452 us MUL_MAT_ID(ffn_moe_down-31): 184 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 355 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 405 us MUL_MAT_ID(ffn_moe_down-33): 186 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 414 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 360 us MUL_MAT_ID(ffn_moe_down-35): 381 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 387 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 428 us MUL_MAT_ID(ffn_moe_down-38): 199 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 396 us MUL_MAT_ID(ffn_moe_down-39): 241 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 412 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 419 us MUL_MAT_ID(ffn_moe_down-41): 191 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 343 us MUL_MAT_ID(ffn_moe_down-42): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 383 us MUL_MAT_ID(ffn_moe_down-43): 215 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 419 us MUL_MAT_ID(ffn_moe_down-44): 183 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 340 us MUL_MAT_ID(ffn_moe_down-45): 198 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 382 us MUL_MAT_ID(ffn_moe_down-46): 290 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 414 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 400 us MUL_MAT_ID(ffn_moe_down-48): 207 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 510 us MUL_MAT_ID(ffn_moe_down-49): 200 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 399 us MUL_MAT_ID(ffn_moe_down-50): 214 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 394 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 440 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 333 us MUL_MAT_ID(ffn_moe_down-53): 206 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 385 us MUL_MAT_ID(ffn_moe_down-54): 217 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 390 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 430 us MUL_MAT_ID(ffn_moe_down-56): 207 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 415 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 416 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 396 us MUL_MAT_ID(ffn_moe_down-59): 265 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 350 us MUL_MAT_ID(ffn_moe_down-60): 210 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 401 us MUL_MAT_ID(ffn_moe_down-61): 207 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 351 us MUL_MAT_ID(ffn_moe_down-62): 218 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 438 us MUL_MAT_ID(ffn_moe_down-63): 320 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 397 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 386 us MUL_MAT_ID(ffn_moe_down-65): 262 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 385 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 380 us MUL_MAT_ID(ffn_moe_down-67): 179 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 408 us MUL_MAT_ID(ffn_moe_down-68): 187 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 402 us MUL_MAT_ID(ffn_moe_down-69): 201 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 433 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 340 us MUL_MAT_ID(ffn_moe_down-71): 350 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 387 us MUL_MAT_ID(ffn_moe_down-72): 266 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 431 us MUL_MAT_ID(ffn_moe_down-73): 183 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 388 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 405 us MUL_MAT_ID(ffn_moe_down-75): 202 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 429 us MUL_MAT_ID(ffn_moe_down-76): 566 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 372 us MUL_MAT_ID(ffn_moe_down-77): 187 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 413 us MUL_MAT_ID(ffn_moe_down-78): 190 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 432 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 362 us MUL_MAT_ID(ffn_moe_down-80): 254 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 436 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 397 us MUL_MAT_ID(ffn_moe_down-82): 229 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 467 us MUL_MAT_ID(ffn_moe_down-83): 284 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 405 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 421 us MUL_MAT_ID(ffn_moe_down-85): 246 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 359 us MUL_MAT_ID(ffn_moe_down-86): 213 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 417 us MUL_MAT_ID(ffn_moe_down-87): 178 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 377 us MUL_MAT_ID(ffn_moe_down-88): 191 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 399 us MUL_MAT_ID(ffn_moe_down-89): 241 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 396 us MUL_MAT_ID(ffn_moe_down-90): 274 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 433 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 49 us GET_ROWS(inp_embd): 11 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 382 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 411 us MUL_MAT_ID(ffn_moe_down-26): 211 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 391 us MUL_MAT_ID(ffn_moe_down-27): 356 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 398 us MUL_MAT_ID(ffn_moe_down-28): 207 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 432 us MUL_MAT_ID(ffn_moe_down-29): 243 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 384 us MUL_MAT_ID(ffn_moe_down-30): 300 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 381 us MUL_MAT_ID(ffn_moe_down-31): 578 us ggml_barrier(...): 434 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 411 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 346 us MUL_MAT_ID(ffn_moe_down-33): 205 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 408 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 412 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 416 us MUL_MAT_ID(ffn_moe_down-36): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 391 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 358 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 411 us MUL_MAT_ID(ffn_moe_down-39): 479 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 443 us MUL_MAT_ID(ffn_moe_down-40): 207 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 418 us MUL_MAT_ID(ffn_moe_down-41): 439 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 420 us MUL_MAT_ID(ffn_moe_down-42): 220 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 440 us MUL_MAT_ID(ffn_moe_down-43): 204 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 399 us MUL_MAT_ID(ffn_moe_down-44): 179 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 349 us MUL_MAT_ID(ffn_moe_down-45): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 380 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 402 us MUL_MAT_ID(ffn_moe_down-47): 197 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 410 us MUL_MAT_ID(ffn_moe_down-48): 185 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 387 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 437 us MUL_MAT_ID(ffn_moe_down-50): 201 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 414 us MUL_MAT_ID(ffn_moe_down-51): 191 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 384 us MUL_MAT_ID(ffn_moe_down-52): 207 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 451 us MUL_MAT_ID(ffn_moe_down-53): 184 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 408 us MUL_MAT_ID(ffn_moe_down-54): 186 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 427 us MUL_MAT_ID(ffn_moe_down-55): 180 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 416 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 448 us MUL_MAT_ID(ffn_moe_down-57): 183 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 431 us MUL_MAT_ID(ffn_moe_down-58): 189 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 394 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 426 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 456 us MUL_MAT_ID(ffn_moe_down-61): 220 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 408 us MUL_MAT_ID(ffn_moe_down-62): 268 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 418 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 448 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 411 us MUL_MAT_ID(ffn_moe_down-65): 289 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 402 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 453 us MUL_MAT_ID(ffn_moe_down-67): 181 us ggml_barrier(...): 411 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 426 us MUL_MAT_ID(ffn_moe_down-68): 416 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 387 us MUL_MAT_ID(ffn_moe_down-69): 195 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 383 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 423 us MUL_MAT_ID(ffn_moe_down-71): 395 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 357 us MUL_MAT_ID(ffn_moe_down-72): 547 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 412 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 438 us MUL_MAT_ID(ffn_moe_down-74): 187 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 432 us MUL_MAT_ID(ffn_moe_down-75): 467 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 390 us MUL_MAT_ID(ffn_moe_down-76): 187 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 434 us MUL_MAT_ID(ffn_moe_down-77): 191 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 426 us MUL_MAT_ID(ffn_moe_down-78): 300 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 402 us MUL_MAT_ID(ffn_moe_down-79): 357 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 422 us MUL_MAT_ID(ffn_moe_down-80): 317 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 402 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 480 us MUL_MAT_ID(ffn_moe_down-82): 442 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 422 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 432 us MUL_MAT_ID(ffn_moe_down-84): 219 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 359 us MUL_MAT_ID(ffn_moe_down-85): 246 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 429 us MUL_MAT_ID(ffn_moe_down-86): 237 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 517 us MUL_MAT_ID(ffn_moe_down-87): 227 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 413 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 447 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 413 us MUL_MAT_ID(ffn_moe_down-90): 236 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 392 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 42 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 392 us MUL_MAT_ID(ffn_moe_down-25): 206 us ggml_barrier(...): 391 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 413 us MUL_MAT_ID(ffn_moe_down-26): 406 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 412 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 445 us MUL_MAT_ID(ffn_moe_down-28): 205 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 413 us MUL_MAT_ID(ffn_moe_down-29): 299 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 419 us MUL_MAT_ID(ffn_moe_down-30): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 349 us MUL_MAT_ID(ffn_moe_down-31): 186 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 388 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 389 us MUL_MAT_ID(ffn_moe_down-33): 441 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 395 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 452 us MUL_MAT_ID(ffn_moe_down-35): 306 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 407 us MUL_MAT_ID(ffn_moe_down-36): 381 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 368 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 366 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 399 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 435 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 378 us MUL_MAT_ID(ffn_moe_down-41): 189 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 417 us MUL_MAT_ID(ffn_moe_down-42): 182 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 403 us MUL_MAT_ID(ffn_moe_down-43): 442 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 372 us MUL_MAT_ID(ffn_moe_down-44): 175 us ggml_barrier(...): 365 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 366 us MUL_MAT_ID(ffn_moe_down-45): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 398 us MUL_MAT_ID(ffn_moe_down-46): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 442 us MUL_MAT_ID(ffn_moe_down-47): 201 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 399 us MUL_MAT_ID(ffn_moe_down-48): 209 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 406 us MUL_MAT_ID(ffn_moe_down-49): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 448 us MUL_MAT_ID(ffn_moe_down-50): 187 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 424 us MUL_MAT_ID(ffn_moe_down-51): 204 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 431 us MUL_MAT_ID(ffn_moe_down-52): 456 us ggml_barrier(...): 360 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 462 us MUL_MAT_ID(ffn_moe_down-53): 191 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 442 us MUL_MAT_ID(ffn_moe_down-54): 243 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 408 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 429 us MUL_MAT_ID(ffn_moe_down-56): 206 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 490 us MUL_MAT_ID(ffn_moe_down-57): 174 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 413 us MUL_MAT_ID(ffn_moe_down-58): 208 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 398 us MUL_MAT_ID(ffn_moe_down-59): 210 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 393 us MUL_MAT_ID(ffn_moe_down-60): 238 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 422 us MUL_MAT_ID(ffn_moe_down-61): 196 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 428 us MUL_MAT_ID(ffn_moe_down-62): 213 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 432 us MUL_MAT_ID(ffn_moe_down-63): 198 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 441 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 424 us MUL_MAT_ID(ffn_moe_down-65): 212 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 430 us MUL_MAT_ID(ffn_moe_down-66): 398 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 363 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 394 us MUL_MAT_ID(ffn_moe_down-68): 204 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 448 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 401 us MUL_MAT_ID(ffn_moe_down-70): 207 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 443 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 350 us MUL_MAT_ID(ffn_moe_down-72): 464 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 443 us MUL_MAT_ID(ffn_moe_down-73): 204 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 385 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 431 us MUL_MAT_ID(ffn_moe_down-75): 214 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 451 us MUL_MAT_ID(ffn_moe_down-76): 218 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 397 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 438 us MUL_MAT_ID(ffn_moe_down-78): 308 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 452 us MUL_MAT_ID(ffn_moe_down-79): 203 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 456 us MUL_MAT_ID(ffn_moe_down-80): 210 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 420 us MUL_MAT_ID(ffn_moe_down-81): 199 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 393 us MUL_MAT_ID(ffn_moe_down-82): 205 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 429 us MUL_MAT_ID(ffn_moe_down-83): 209 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 452 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 427 us MUL_MAT_ID(ffn_moe_down-85): 747 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 405 us MUL_MAT_ID(ffn_moe_down-86): 362 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 429 us MUL_MAT_ID(ffn_moe_down-87): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 447 us MUL_MAT_ID(ffn_moe_down-88): 229 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 489 us MUL_MAT_ID(ffn_moe_down-89): 172 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 361 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 398 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 32 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 394 us MUL_MAT_ID(ffn_moe_down-25): 202 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 407 us MUL_MAT_ID(ffn_moe_down-26): 215 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 389 us MUL_MAT_ID(ffn_moe_down-27): 233 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 422 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 415 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 411 us MUL_MAT_ID(ffn_moe_down-30): 280 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 376 us MUL_MAT_ID(ffn_moe_down-31): 199 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 394 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 447 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 391 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 381 us MUL_MAT_ID(ffn_moe_down-35): 182 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 333 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 389 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 392 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 390 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 455 us MUL_MAT_ID(ffn_moe_down-40): 608 us ggml_barrier(...): 551 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 377 us MUL_MAT_ID(ffn_moe_down-41): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 442 us MUL_MAT_ID(ffn_moe_down-42): 271 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 377 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 428 us MUL_MAT_ID(ffn_moe_down-44): 411 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 473 us MUL_MAT_ID(ffn_moe_down-45): 246 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 422 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 456 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 344 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 412 us MUL_MAT_ID(ffn_moe_down-49): 587 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 404 us MUL_MAT_ID(ffn_moe_down-50): 315 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 416 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 366 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 369 us MUL_MAT_ID(ffn_moe_down-52): 409 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 404 us MUL_MAT_ID(ffn_moe_down-53): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 408 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 367 us MUL_MAT_ID(ffn_moe_down-55): 193 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 418 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 444 us MUL_MAT_ID(ffn_moe_down-57): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 391 us MUL_MAT_ID(ffn_moe_down-58): 189 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 405 us MUL_MAT_ID(ffn_moe_down-59): 190 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 440 us MUL_MAT_ID(ffn_moe_down-60): 569 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 385 us MUL_MAT_ID(ffn_moe_down-61): 209 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 343 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 397 us MUL_MAT_ID(ffn_moe_down-63): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 440 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 379 us MUL_MAT_ID(ffn_moe_down-65): 211 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 437 us MUL_MAT_ID(ffn_moe_down-66): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 417 us MUL_MAT_ID(ffn_moe_down-67): 216 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 449 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 397 us MUL_MAT_ID(ffn_moe_down-69): 530 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 364 us MUL_MAT_ID(ffn_moe_down-70): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 414 us MUL_MAT_ID(ffn_moe_down-71): 207 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 396 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 361 us MUL_MAT_ID(ffn_moe_down-73): 187 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 384 us MUL_MAT_ID(ffn_moe_down-74): 203 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 405 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 366 us MUL_MAT_ID(ffn_moe_down-76): 184 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 384 us MUL_MAT_ID(ffn_moe_down-77): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 428 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 409 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 354 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 200 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 373 us MUL_MAT_ID(ffn_moe_down-82): 191 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 432 us MUL_MAT_ID(ffn_moe_down-83): 209 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 397 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 360 us MUL_MAT_ID(ffn_moe_down-85): 234 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 390 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 405 us MUL_MAT_ID(ffn_moe_down-87): 186 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 355 us MUL_MAT_ID(ffn_moe_down-88): 205 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 407 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 396 us MUL_MAT_ID(ffn_moe_down-90): 230 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 421 us MUL_MAT_ID(ffn_moe_down-91): 290 us ggml_barrier(...): 192 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 399 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 381 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 395 us MUL_MAT_ID(ffn_moe_down-27): 256 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 392 us MUL_MAT_ID(ffn_moe_down-28): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 415 us MUL_MAT_ID(ffn_moe_down-29): 200 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 404 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 379 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 442 us MUL_MAT_ID(ffn_moe_down-32): 458 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 388 us MUL_MAT_ID(ffn_moe_down-33): 196 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 427 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 416 us MUL_MAT_ID(ffn_moe_down-35): 211 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 385 us MUL_MAT_ID(ffn_moe_down-36): 205 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 380 us MUL_MAT_ID(ffn_moe_down-37): 291 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 367 us MUL_MAT_ID(ffn_moe_down-38): 211 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 405 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 424 us MUL_MAT_ID(ffn_moe_down-40): 235 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 393 us MUL_MAT_ID(ffn_moe_down-41): 265 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 407 us MUL_MAT_ID(ffn_moe_down-42): 189 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 397 us MUL_MAT_ID(ffn_moe_down-43): 189 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 424 us MUL_MAT_ID(ffn_moe_down-44): 214 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 592 us MUL_MAT_ID(ffn_moe_down-45): 200 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 403 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 379 us MUL_MAT_ID(ffn_moe_down-47): 282 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 388 us MUL_MAT_ID(ffn_moe_down-48): 591 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 378 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 394 us MUL_MAT_ID(ffn_moe_down-50): 257 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 437 us MUL_MAT_ID(ffn_moe_down-51): 283 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 424 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 495 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 405 us MUL_MAT_ID(ffn_moe_down-53): 214 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 388 us MUL_MAT_ID(ffn_moe_down-54): 183 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 416 us MUL_MAT_ID(ffn_moe_down-55): 215 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 420 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 396 us MUL_MAT_ID(ffn_moe_down-57): 183 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 389 us MUL_MAT_ID(ffn_moe_down-58): 251 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 435 us MUL_MAT_ID(ffn_moe_down-59): 264 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 407 us MUL_MAT_ID(ffn_moe_down-60): 185 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 384 us MUL_MAT_ID(ffn_moe_down-61): 201 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 379 us MUL_MAT_ID(ffn_moe_down-62): 588 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 412 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 375 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 401 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 339 us MUL_MAT_ID(ffn_moe_down-65): 401 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 410 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 371 us MUL_MAT_ID(ffn_moe_down-67): 185 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 417 us MUL_MAT_ID(ffn_moe_down-68): 553 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 425 us MUL_MAT_ID(ffn_moe_down-69): 313 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 385 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 452 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 417 us MUL_MAT_ID(ffn_moe_down-72): 502 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 326 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 405 us MUL_MAT_ID(ffn_moe_down-74): 200 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 369 us MUL_MAT_ID(ffn_moe_down-75): 422 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 385 us MUL_MAT_ID(ffn_moe_down-76): 387 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 375 us MUL_MAT_ID(ffn_moe_down-77): 191 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 421 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 500 us MUL_MAT_ID(ffn_moe_down-79): 276 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 377 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 400 us MUL_MAT_ID(ffn_moe_down-81): 200 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 346 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 378 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 362 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 425 us MUL_MAT_ID(ffn_moe_down-85): 245 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 374 us MUL_MAT_ID(ffn_moe_down-86): 185 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 488 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 491 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 449 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 423 us MUL_MAT_ID(ffn_moe_down-89): 409 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 397 us MUL_MAT_ID(ffn_moe_down-90): 283 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 390 us MUL_MAT_ID(ffn_moe_down-91): 733 us ggml_barrier(...): 39 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 384 us MUL_MAT_ID(ffn_moe_down-25): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 402 us MUL_MAT_ID(ffn_moe_down-26): 208 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 454 us MUL_MAT_ID(ffn_moe_down-27): 206 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 432 us MUL_MAT_ID(ffn_moe_down-28): 210 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 408 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 423 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 386 us MUL_MAT_ID(ffn_moe_down-31): 323 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 393 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 400 us MUL_MAT_ID(ffn_moe_down-33): 413 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 404 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 406 us MUL_MAT_ID(ffn_moe_down-35): 202 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 371 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 409 us MUL_MAT_ID(ffn_moe_down-37): 258 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 483 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 395 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 414 us MUL_MAT_ID(ffn_moe_down-40): 241 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 411 us MUL_MAT_ID(ffn_moe_down-41): 406 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 388 us MUL_MAT_ID(ffn_moe_down-42): 251 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 411 us MUL_MAT_ID(ffn_moe_down-43): 202 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 420 us MUL_MAT_ID(ffn_moe_down-44): 352 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 394 us MUL_MAT_ID(ffn_moe_down-45): 202 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 396 us MUL_MAT_ID(ffn_moe_down-46): 399 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 373 us MUL_MAT_ID(ffn_moe_down-47): 201 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 392 us MUL_MAT_ID(ffn_moe_down-48): 211 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 406 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 401 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 373 us MUL_MAT_ID(ffn_moe_down-51): 171 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 383 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 409 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 450 us MUL_MAT_ID(ffn_moe_down-54): 206 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 381 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 416 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 552 us MUL_MAT_ID(ffn_moe_down-57): 227 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 393 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 439 us MUL_MAT_ID(ffn_moe_down-59): 281 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 390 us MUL_MAT_ID(ffn_moe_down-60): 572 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 384 us MUL_MAT_ID(ffn_moe_down-61): 217 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 383 us MUL_MAT_ID(ffn_moe_down-62): 249 us ggml_barrier(...): 506 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 542 us MUL_MAT_ID(ffn_moe_down-63): 436 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 416 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 381 us MUL_MAT_ID(ffn_moe_down-65): 287 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 408 us MUL_MAT_ID(ffn_moe_down-66): 210 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 394 us MUL_MAT_ID(ffn_moe_down-67): 189 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 435 us MUL_MAT_ID(ffn_moe_down-68): 298 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 441 us MUL_MAT_ID(ffn_moe_down-69): 171 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 366 us MUL_MAT_ID(ffn_moe_down-70): 191 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 415 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 337 us MUL_MAT_ID(ffn_moe_down-72): 192 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 431 us MUL_MAT_ID(ffn_moe_down-73): 211 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 411 us MUL_MAT_ID(ffn_moe_down-74): 399 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 400 us MUL_MAT_ID(ffn_moe_down-75): 398 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 702 us MUL_MAT_ID(ffn_moe_down-76): 348 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 364 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 398 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 417 us MUL_MAT_ID(ffn_moe_down-79): 427 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 425 us MUL_MAT_ID(ffn_moe_down-80): 193 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 408 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 412 us MUL_MAT_ID(ffn_moe_down-82): 563 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 406 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 393 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 402 us MUL_MAT_ID(ffn_moe_down-85): 371 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 390 us MUL_MAT_ID(ffn_moe_down-86): 226 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 464 us MUL_MAT_ID(ffn_moe_down-87): 186 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 397 us MUL_MAT_ID(ffn_moe_down-88): 206 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 405 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 452 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 415 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 25 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 525 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 383 us MUL_MAT_ID(ffn_moe_down-26): 277 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 495 us MUL_MAT_ID(ffn_moe_down-27): 182 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 440 us MUL_MAT_ID(ffn_moe_down-28): 293 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 418 us MUL_MAT_ID(ffn_moe_down-29): 290 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 417 us MUL_MAT_ID(ffn_moe_down-30): 538 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 491 us MUL_MAT_ID(ffn_moe_down-31): 341 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 383 us MUL_MAT_ID(ffn_moe_down-32): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 424 us MUL_MAT_ID(ffn_moe_down-33): 177 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 402 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 399 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 381 us MUL_MAT_ID(ffn_moe_down-36): 273 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 406 us MUL_MAT_ID(ffn_moe_down-37): 297 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 440 us MUL_MAT_ID(ffn_moe_down-38): 298 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 464 us MUL_MAT_ID(ffn_moe_down-39): 284 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 411 us MUL_MAT_ID(ffn_moe_down-40): 418 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 399 us MUL_MAT_ID(ffn_moe_down-41): 204 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 399 us MUL_MAT_ID(ffn_moe_down-42): 206 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 359 us MUL_MAT_ID(ffn_moe_down-43): 186 us ggml_barrier(...): 386 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 393 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 396 us MUL_MAT_ID(ffn_moe_down-45): 230 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 404 us MUL_MAT_ID(ffn_moe_down-46): 210 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 420 us MUL_MAT_ID(ffn_moe_down-47): 385 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 366 us MUL_MAT_ID(ffn_moe_down-48): 200 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 407 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 363 us MUL_MAT_ID(ffn_moe_down-50): 216 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 374 us MUL_MAT_ID(ffn_moe_down-51): 483 us ggml_barrier(...): 391 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 393 us MUL_MAT_ID(ffn_moe_down-52): 207 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 407 us MUL_MAT_ID(ffn_moe_down-53): 288 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 421 us MUL_MAT_ID(ffn_moe_down-54): 195 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 429 us MUL_MAT_ID(ffn_moe_down-55): 211 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 383 us MUL_MAT_ID(ffn_moe_down-56): 234 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 440 us MUL_MAT_ID(ffn_moe_down-57): 193 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 412 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 413 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 400 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 380 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 367 us MUL_MAT_ID(ffn_moe_down-61): 185 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 513 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 401 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 381 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 368 us MUL_MAT_ID(ffn_moe_down-65): 519 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 374 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 429 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 396 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 512 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 407 us MUL_MAT_ID(ffn_moe_down-69): 201 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 442 us MUL_MAT_ID(ffn_moe_down-70): 441 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 366 us MUL_MAT_ID(ffn_moe_down-71): 191 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 408 us MUL_MAT_ID(ffn_moe_down-72): 225 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 374 us MUL_MAT_ID(ffn_moe_down-73): 262 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 427 us MUL_MAT_ID(ffn_moe_down-74): 203 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 408 us MUL_MAT_ID(ffn_moe_down-75): 206 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 469 us MUL_MAT_ID(ffn_moe_down-76): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 444 us MUL_MAT_ID(ffn_moe_down-77): 196 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 408 us MUL_MAT_ID(ffn_moe_down-78): 219 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 411 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 366 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 459 us MUL_MAT_ID(ffn_moe_down-80): 456 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 411 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 374 us MUL_MAT_ID(ffn_moe_down-82): 197 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 459 us MUL_MAT_ID(ffn_moe_down-83): 288 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 413 us MUL_MAT_ID(ffn_moe_down-84): 621 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 390 us MUL_MAT_ID(ffn_moe_down-85): 734 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 409 us MUL_MAT_ID(ffn_moe_down-86): 386 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 420 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 406 us MUL_MAT_ID(ffn_moe_down-88): 222 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 352 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 394 us MUL_MAT_ID(ffn_moe_down-90): 619 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 414 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 17 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 369 us MUL_MAT_ID(ffn_moe_down-25): 208 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 425 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 363 us MUL_MAT_ID(ffn_moe_down-27): 204 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 368 us MUL_MAT_ID(ffn_moe_down-28): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 359 us MUL_MAT_ID(ffn_moe_down-29): 200 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 392 us MUL_MAT_ID(ffn_moe_down-30): 214 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 380 us MUL_MAT_ID(ffn_moe_down-31): 217 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 413 us MUL_MAT_ID(ffn_moe_down-32): 184 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 338 us MUL_MAT_ID(ffn_moe_down-33): 190 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 391 us MUL_MAT_ID(ffn_moe_down-34): 384 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 425 us MUL_MAT_ID(ffn_moe_down-35): 299 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 355 us MUL_MAT_ID(ffn_moe_down-36): 178 us ggml_barrier(...): 346 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 420 us MUL_MAT_ID(ffn_moe_down-37): 504 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 373 us MUL_MAT_ID(ffn_moe_down-38): 182 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 418 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 395 us MUL_MAT_ID(ffn_moe_down-40): 244 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 395 us MUL_MAT_ID(ffn_moe_down-41): 188 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 484 us MUL_MAT_ID(ffn_moe_down-42): 253 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 364 us MUL_MAT_ID(ffn_moe_down-43): 315 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 392 us MUL_MAT_ID(ffn_moe_down-44): 210 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 457 us MUL_MAT_ID(ffn_moe_down-45): 236 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 399 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 410 us MUL_MAT_ID(ffn_moe_down-47): 205 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 370 us MUL_MAT_ID(ffn_moe_down-48): 183 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 435 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 362 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 384 us MUL_MAT_ID(ffn_moe_down-51): 209 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 427 us MUL_MAT_ID(ffn_moe_down-52): 176 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 371 us MUL_MAT_ID(ffn_moe_down-53): 206 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 456 us MUL_MAT_ID(ffn_moe_down-54): 215 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 422 us MUL_MAT_ID(ffn_moe_down-55): 281 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 206 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 432 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 455 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 418 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 458 us MUL_MAT_ID(ffn_moe_down-60): 346 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 413 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 442 us MUL_MAT_ID(ffn_moe_down-62): 206 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 421 us MUL_MAT_ID(ffn_moe_down-63): 343 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 390 us MUL_MAT_ID(ffn_moe_down-64): 434 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 423 us MUL_MAT_ID(ffn_moe_down-65): 240 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 527 us MUL_MAT_ID(ffn_moe_down-66): 205 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 427 us MUL_MAT_ID(ffn_moe_down-67): 229 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 389 us MUL_MAT_ID(ffn_moe_down-68): 189 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 489 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 398 us MUL_MAT_ID(ffn_moe_down-70): 339 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 390 us MUL_MAT_ID(ffn_moe_down-71): 197 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 444 us MUL_MAT_ID(ffn_moe_down-72): 181 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 408 us MUL_MAT_ID(ffn_moe_down-73): 332 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 413 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 439 us MUL_MAT_ID(ffn_moe_down-75): 323 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 402 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 424 us MUL_MAT_ID(ffn_moe_down-77): 232 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 393 us MUL_MAT_ID(ffn_moe_down-78): 213 us ggml_barrier(...): 402 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 385 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 390 us MUL_MAT_ID(ffn_moe_down-80): 365 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 377 us MUL_MAT_ID(ffn_moe_down-81): 185 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 392 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 432 us MUL_MAT_ID(ffn_moe_down-83): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 505 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 428 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 424 us MUL_MAT_ID(ffn_moe_down-86): 208 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 399 us MUL_MAT_ID(ffn_moe_down-87): 226 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 408 us MUL_MAT_ID(ffn_moe_down-88): 231 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 364 us MUL_MAT_ID(ffn_moe_down-89): 299 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 416 us MUL_MAT_ID(ffn_moe_down-90): 243 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 371 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 35 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 384 us MUL_MAT_ID(ffn_moe_down-25): 183 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 442 us MUL_MAT_ID(ffn_moe_down-26): 203 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 476 us MUL_MAT_ID(ffn_moe_down-27): 222 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 372 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 383 us MUL_MAT_ID(ffn_moe_down-29): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 468 us MUL_MAT_ID(ffn_moe_down-30): 207 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 424 us MUL_MAT_ID(ffn_moe_down-31): 199 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 402 us MUL_MAT_ID(ffn_moe_down-32): 444 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 220 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 335 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 388 us MUL_MAT_ID(ffn_moe_down-36): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 406 us MUL_MAT_ID(ffn_moe_down-37): 238 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 388 us MUL_MAT_ID(ffn_moe_down-38): 176 us ggml_barrier(...): 356 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 460 us MUL_MAT_ID(ffn_moe_down-39): 676 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 363 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 365 us MUL_MAT_ID(ffn_moe_down-41): 385 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 407 us MUL_MAT_ID(ffn_moe_down-42): 579 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 408 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 411 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 378 us MUL_MAT_ID(ffn_moe_down-45): 182 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 404 us MUL_MAT_ID(ffn_moe_down-46): 210 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 481 us MUL_MAT_ID(ffn_moe_down-47): 204 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 406 us MUL_MAT_ID(ffn_moe_down-48): 204 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 433 us MUL_MAT_ID(ffn_moe_down-49): 204 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 524 us MUL_MAT_ID(ffn_moe_down-50): 288 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 434 us MUL_MAT_ID(ffn_moe_down-51): 329 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 440 us MUL_MAT_ID(ffn_moe_down-52): 244 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 426 us MUL_MAT_ID(ffn_moe_down-53): 179 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 409 us MUL_MAT_ID(ffn_moe_down-54): 185 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 386 us MUL_MAT_ID(ffn_moe_down-55): 189 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 413 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 386 us MUL_MAT_ID(ffn_moe_down-57): 411 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 433 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 428 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 423 us MUL_MAT_ID(ffn_moe_down-60): 419 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 427 us MUL_MAT_ID(ffn_moe_down-61): 206 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 434 us MUL_MAT_ID(ffn_moe_down-62): 203 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 481 us MUL_MAT_ID(ffn_moe_down-63): 208 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 417 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 474 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 407 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 444 us MUL_MAT_ID(ffn_moe_down-67): 224 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 486 us MUL_MAT_ID(ffn_moe_down-68): 237 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 391 us MUL_MAT_ID(ffn_moe_down-69): 182 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 379 us MUL_MAT_ID(ffn_moe_down-70): 193 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 421 us MUL_MAT_ID(ffn_moe_down-71): 203 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 443 us MUL_MAT_ID(ffn_moe_down-72): 216 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 333 us MUL_MAT_ID(ffn_moe_down-73): 382 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 391 us MUL_MAT_ID(ffn_moe_down-74): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 428 us MUL_MAT_ID(ffn_moe_down-75): 182 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 401 us MUL_MAT_ID(ffn_moe_down-76): 287 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 446 us MUL_MAT_ID(ffn_moe_down-77): 209 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 380 us MUL_MAT_ID(ffn_moe_down-78): 189 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 580 us MUL_MAT_ID(ffn_moe_down-79): 372 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 507 us MUL_MAT_ID(ffn_moe_down-80): 244 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 426 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 415 us MUL_MAT_ID(ffn_moe_down-82): 208 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 400 us MUL_MAT_ID(ffn_moe_down-83): 260 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 394 us MUL_MAT_ID(ffn_moe_down-84): 259 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 433 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 403 us MUL_MAT_ID(ffn_moe_down-86): 226 us ggml_barrier(...): 367 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 411 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 434 us MUL_MAT_ID(ffn_moe_down-88): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 470 us MUL_MAT_ID(ffn_moe_down-89): 177 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 544 us MUL_MAT_ID(ffn_moe_down-90): 291 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 477 us MUL_MAT_ID(ffn_moe_down-91): 270 us ggml_barrier(...): 129 us GET_ROWS(inp_embd): 11 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 507 us MUL_MAT_ID(ffn_moe_down-25): 205 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 407 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 422 us MUL_MAT_ID(ffn_moe_down-27): 222 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 368 us MUL_MAT_ID(ffn_moe_down-28): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 205 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 357 us MUL_MAT_ID(ffn_moe_down-30): 202 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 438 us MUL_MAT_ID(ffn_moe_down-31): 205 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 373 us MUL_MAT_ID(ffn_moe_down-32): 209 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 384 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 379 us MUL_MAT_ID(ffn_moe_down-34): 186 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 411 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 399 us MUL_MAT_ID(ffn_moe_down-36): 209 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 341 us MUL_MAT_ID(ffn_moe_down-37): 239 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 415 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 350 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 351 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 402 us MUL_MAT_ID(ffn_moe_down-41): 576 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 381 us MUL_MAT_ID(ffn_moe_down-42): 199 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 342 us MUL_MAT_ID(ffn_moe_down-43): 204 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 375 us MUL_MAT_ID(ffn_moe_down-44): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 368 us MUL_MAT_ID(ffn_moe_down-45): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 385 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 411 us MUL_MAT_ID(ffn_moe_down-47): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 401 us MUL_MAT_ID(ffn_moe_down-48): 209 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 448 us MUL_MAT_ID(ffn_moe_down-49): 185 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 378 us MUL_MAT_ID(ffn_moe_down-50): 187 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 347 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 374 us MUL_MAT_ID(ffn_moe_down-52): 182 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 412 us MUL_MAT_ID(ffn_moe_down-53): 192 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 382 us MUL_MAT_ID(ffn_moe_down-54): 372 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 488 us MUL_MAT_ID(ffn_moe_down-55): 229 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 419 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 423 us MUL_MAT_ID(ffn_moe_down-57): 187 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 390 us MUL_MAT_ID(ffn_moe_down-58): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 421 us MUL_MAT_ID(ffn_moe_down-59): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 502 us MUL_MAT_ID(ffn_moe_down-60): 254 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 379 us MUL_MAT_ID(ffn_moe_down-61): 203 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 378 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 355 us MUL_MAT_ID(ffn_moe_down-63): 309 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 396 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 418 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 413 us MUL_MAT_ID(ffn_moe_down-66): 393 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 393 us MUL_MAT_ID(ffn_moe_down-67): 220 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 423 us MUL_MAT_ID(ffn_moe_down-68): 186 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 376 us MUL_MAT_ID(ffn_moe_down-69): 219 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 343 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 387 us MUL_MAT_ID(ffn_moe_down-71): 221 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 449 us MUL_MAT_ID(ffn_moe_down-72): 424 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 401 us MUL_MAT_ID(ffn_moe_down-73): 195 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 354 us MUL_MAT_ID(ffn_moe_down-74): 207 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 445 us MUL_MAT_ID(ffn_moe_down-75): 213 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 393 us MUL_MAT_ID(ffn_moe_down-76): 220 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 392 us MUL_MAT_ID(ffn_moe_down-77): 193 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 379 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 400 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 402 us MUL_MAT_ID(ffn_moe_down-80): 200 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 363 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 381 us MUL_MAT_ID(ffn_moe_down-82): 356 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 409 us MUL_MAT_ID(ffn_moe_down-83): 207 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 427 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 413 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 417 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 409 us MUL_MAT_ID(ffn_moe_down-87): 204 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 406 us MUL_MAT_ID(ffn_moe_down-88): 377 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 377 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 416 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 365 us MUL_MAT_ID(ffn_moe_down-91): 244 us ggml_barrier(...): 73 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 369 us MUL_MAT_ID(ffn_moe_down-25): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 396 us MUL_MAT_ID(ffn_moe_down-26): 204 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 426 us MUL_MAT_ID(ffn_moe_down-27): 242 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 481 us MUL_MAT_ID(ffn_moe_down-28): 329 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 416 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 381 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 395 us MUL_MAT_ID(ffn_moe_down-30): 221 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 407 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 401 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 432 us MUL_MAT_ID(ffn_moe_down-33): 353 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 439 us MUL_MAT_ID(ffn_moe_down-34): 226 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 408 us MUL_MAT_ID(ffn_moe_down-35): 237 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 409 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 391 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 421 us MUL_MAT_ID(ffn_moe_down-38): 215 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 436 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 449 us MUL_MAT_ID(ffn_moe_down-40): 241 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 416 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 455 us MUL_MAT_ID(ffn_moe_down-42): 399 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 424 us MUL_MAT_ID(ffn_moe_down-43): 294 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 425 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 423 us MUL_MAT_ID(ffn_moe_down-45): 238 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 380 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 405 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 430 us MUL_MAT_ID(ffn_moe_down-48): 187 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 425 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 339 us MUL_MAT_ID(ffn_moe_down-50): 539 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 399 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 393 us MUL_MAT_ID(ffn_moe_down-52): 216 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 358 us MUL_MAT_ID(ffn_moe_down-53): 314 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 345 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 387 us MUL_MAT_ID(ffn_moe_down-55): 221 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 417 us MUL_MAT_ID(ffn_moe_down-56): 189 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 415 us MUL_MAT_ID(ffn_moe_down-57): 203 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 422 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 397 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 394 us MUL_MAT_ID(ffn_moe_down-60): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 360 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 369 us MUL_MAT_ID(ffn_moe_down-62): 204 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 376 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 339 us MUL_MAT_ID(ffn_moe_down-64): 322 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 359 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 357 us MUL_MAT_ID(ffn_moe_down-66): 185 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 427 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 427 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 365 us MUL_MAT_ID(ffn_moe_down-68): 185 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 402 us MUL_MAT_ID(ffn_moe_down-69): 370 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 373 us MUL_MAT_ID(ffn_moe_down-70): 214 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 389 us MUL_MAT_ID(ffn_moe_down-71): 280 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 474 us MUL_MAT_ID(ffn_moe_down-72): 203 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 403 us MUL_MAT_ID(ffn_moe_down-73): 206 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 394 us MUL_MAT_ID(ffn_moe_down-74): 213 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 361 us MUL_MAT_ID(ffn_moe_down-75): 200 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 370 us MUL_MAT_ID(ffn_moe_down-76): 171 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 378 us MUL_MAT_ID(ffn_moe_down-77): 189 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 447 us MUL_MAT_ID(ffn_moe_down-78): 185 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 383 us MUL_MAT_ID(ffn_moe_down-79): 216 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 407 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 341 us MUL_MAT_ID(ffn_moe_down-81): 233 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 423 us MUL_MAT_ID(ffn_moe_down-82): 425 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 372 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 431 us MUL_MAT_ID(ffn_moe_down-84): 321 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 395 us MUL_MAT_ID(ffn_moe_down-85): 226 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 425 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 430 us MUL_MAT_ID(ffn_moe_down-87): 177 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 397 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 347 us MUL_MAT_ID(ffn_moe_down-89): 310 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 398 us MUL_MAT_ID(ffn_moe_down-90): 230 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 384 us MUL_MAT_ID(ffn_moe_down-91): 228 us ggml_barrier(...): 63 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 400 us MUL_MAT_ID(ffn_moe_down-25): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 487 us MUL_MAT_ID(ffn_moe_down-26): 197 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 398 us MUL_MAT_ID(ffn_moe_down-27): 320 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 188 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 426 us MUL_MAT_ID(ffn_moe_down-29): 197 us ggml_barrier(...): 360 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 396 us MUL_MAT_ID(ffn_moe_down-30): 211 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 390 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 409 us MUL_MAT_ID(ffn_moe_down-32): 211 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 437 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 428 us MUL_MAT_ID(ffn_moe_down-34): 337 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 394 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 384 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 450 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 422 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 362 us MUL_MAT_ID(ffn_moe_down-39): 630 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 360 us MUL_MAT_ID(ffn_moe_down-40): 238 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 395 us MUL_MAT_ID(ffn_moe_down-41): 408 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 398 us MUL_MAT_ID(ffn_moe_down-42): 395 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 412 us MUL_MAT_ID(ffn_moe_down-43): 169 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 391 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 418 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 384 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 440 us MUL_MAT_ID(ffn_moe_down-47): 235 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 403 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 390 us MUL_MAT_ID(ffn_moe_down-49): 192 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 390 us MUL_MAT_ID(ffn_moe_down-50): 223 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 466 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 399 us MUL_MAT_ID(ffn_moe_down-52): 204 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 409 us MUL_MAT_ID(ffn_moe_down-53): 355 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 393 us MUL_MAT_ID(ffn_moe_down-54): 312 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 435 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 390 us MUL_MAT_ID(ffn_moe_down-56): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 399 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 420 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 457 us MUL_MAT_ID(ffn_moe_down-59): 395 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 376 us MUL_MAT_ID(ffn_moe_down-60): 254 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 451 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 437 us MUL_MAT_ID(ffn_moe_down-62): 226 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 202 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 402 us MUL_MAT_ID(ffn_moe_down-64): 215 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 420 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 373 us MUL_MAT_ID(ffn_moe_down-66): 187 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 401 us MUL_MAT_ID(ffn_moe_down-67): 188 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 392 us MUL_MAT_ID(ffn_moe_down-68): 296 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 437 us MUL_MAT_ID(ffn_moe_down-69): 285 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 426 us MUL_MAT_ID(ffn_moe_down-70): 346 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 413 us MUL_MAT_ID(ffn_moe_down-71): 206 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 361 us MUL_MAT_ID(ffn_moe_down-72): 188 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 489 us MUL_MAT_ID(ffn_moe_down-73): 184 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 428 us MUL_MAT_ID(ffn_moe_down-74): 191 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 408 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 398 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 424 us MUL_MAT_ID(ffn_moe_down-77): 247 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 425 us MUL_MAT_ID(ffn_moe_down-78): 410 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 643 us MUL_MAT_ID(ffn_moe_down-79): 248 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 482 us MUL_MAT_ID(ffn_moe_down-80): 240 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 394 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 396 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 366 us MUL_MAT_ID(ffn_moe_down-83): 591 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 382 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 429 us MUL_MAT_ID(ffn_moe_down-85): 225 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 448 us MUL_MAT_ID(ffn_moe_down-86): 399 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 421 us MUL_MAT_ID(ffn_moe_down-87): 199 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 509 us MUL_MAT_ID(ffn_moe_down-88): 197 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 476 us MUL_MAT_ID(ffn_moe_down-89): 167 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 400 us MUL_MAT_ID(ffn_moe_down-90): 611 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 376 us MUL_MAT_ID(ffn_moe_down-91): 264 us ggml_barrier(...): 42 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 398 us MUL_MAT_ID(ffn_moe_down-25): 394 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 439 us MUL_MAT_ID(ffn_moe_down-26): 206 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 454 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 361 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 403 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 435 us MUL_MAT_ID(ffn_moe_down-30): 188 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 371 us MUL_MAT_ID(ffn_moe_down-31): 175 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 415 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 580 us MUL_MAT_ID(ffn_moe_down-33): 237 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 462 us MUL_MAT_ID(ffn_moe_down-34): 224 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 422 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 463 us MUL_MAT_ID(ffn_moe_down-36): 387 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 415 us MUL_MAT_ID(ffn_moe_down-37): 249 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 433 us MUL_MAT_ID(ffn_moe_down-38): 184 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 407 us MUL_MAT_ID(ffn_moe_down-39): 578 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 426 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 441 us MUL_MAT_ID(ffn_moe_down-41): 194 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 448 us MUL_MAT_ID(ffn_moe_down-42): 211 us ggml_barrier(...): 420 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 409 us MUL_MAT_ID(ffn_moe_down-43): 184 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 386 us MUL_MAT_ID(ffn_moe_down-44): 209 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 394 us MUL_MAT_ID(ffn_moe_down-45): 218 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 458 us MUL_MAT_ID(ffn_moe_down-46): 280 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 417 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 401 us MUL_MAT_ID(ffn_moe_down-48): 199 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 408 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 348 us MUL_MAT_ID(ffn_moe_down-50): 176 us ggml_barrier(...): 351 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 402 us MUL_MAT_ID(ffn_moe_down-51): 208 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 433 us MUL_MAT_ID(ffn_moe_down-52): 202 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 443 us MUL_MAT_ID(ffn_moe_down-53): 292 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 385 us MUL_MAT_ID(ffn_moe_down-54): 253 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 216 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 431 us MUL_MAT_ID(ffn_moe_down-56): 342 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 402 us MUL_MAT_ID(ffn_moe_down-57): 228 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 433 us MUL_MAT_ID(ffn_moe_down-58): 237 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 445 us MUL_MAT_ID(ffn_moe_down-59): 292 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 452 us MUL_MAT_ID(ffn_moe_down-60): 214 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 432 us MUL_MAT_ID(ffn_moe_down-61): 417 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 364 us MUL_MAT_ID(ffn_moe_down-62): 214 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 375 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 422 us MUL_MAT_ID(ffn_moe_down-64): 184 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 423 us MUL_MAT_ID(ffn_moe_down-65): 196 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 401 us MUL_MAT_ID(ffn_moe_down-66): 256 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 485 us MUL_MAT_ID(ffn_moe_down-67): 269 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 473 us MUL_MAT_ID(ffn_moe_down-68): 201 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 421 us MUL_MAT_ID(ffn_moe_down-69): 173 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 407 us MUL_MAT_ID(ffn_moe_down-70): 302 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 367 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 467 us MUL_MAT_ID(ffn_moe_down-72): 278 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 402 us MUL_MAT_ID(ffn_moe_down-73): 219 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 404 us MUL_MAT_ID(ffn_moe_down-74): 282 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 414 us MUL_MAT_ID(ffn_moe_down-75): 204 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 423 us MUL_MAT_ID(ffn_moe_down-76): 356 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 412 us MUL_MAT_ID(ffn_moe_down-77): 197 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 394 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 422 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 397 us MUL_MAT_ID(ffn_moe_down-80): 292 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 383 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 401 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 404 us MUL_MAT_ID(ffn_moe_down-83): 316 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 412 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 404 us MUL_MAT_ID(ffn_moe_down-85): 200 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 387 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 436 us MUL_MAT_ID(ffn_moe_down-87): 228 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 452 us MUL_MAT_ID(ffn_moe_down-88): 197 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 433 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 365 us MUL_MAT_ID(ffn_moe_down-90): 574 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 413 us MUL_MAT_ID(ffn_moe_down-91): 462 us ggml_barrier(...): 294 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 414 us MUL_MAT_ID(ffn_moe_down-25): 198 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 420 us MUL_MAT_ID(ffn_moe_down-26): 348 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 398 us MUL_MAT_ID(ffn_moe_down-27): 211 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 353 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 381 us MUL_MAT_ID(ffn_moe_down-29): 200 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 376 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 409 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 375 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 353 us MUL_MAT_ID(ffn_moe_down-33): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 451 us MUL_MAT_ID(ffn_moe_down-35): 232 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 393 us MUL_MAT_ID(ffn_moe_down-36): 228 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 379 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 384 us MUL_MAT_ID(ffn_moe_down-38): 188 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 392 us MUL_MAT_ID(ffn_moe_down-39): 228 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 400 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 540 us MUL_MAT_ID(ffn_moe_down-41): 186 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 416 us MUL_MAT_ID(ffn_moe_down-42): 220 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 446 us MUL_MAT_ID(ffn_moe_down-43): 199 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 369 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 420 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 367 us MUL_MAT_ID(ffn_moe_down-46): 211 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 564 us MUL_MAT_ID(ffn_moe_down-47): 253 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 437 us MUL_MAT_ID(ffn_moe_down-48): 413 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 388 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 390 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 345 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 435 us MUL_MAT_ID(ffn_moe_down-52): 186 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 352 us MUL_MAT_ID(ffn_moe_down-53): 261 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 389 us MUL_MAT_ID(ffn_moe_down-54): 491 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 470 us MUL_MAT_ID(ffn_moe_down-55): 281 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 445 us MUL_MAT_ID(ffn_moe_down-56): 292 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 382 us MUL_MAT_ID(ffn_moe_down-57): 195 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 448 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 386 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 426 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 413 us MUL_MAT_ID(ffn_moe_down-61): 302 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 429 us MUL_MAT_ID(ffn_moe_down-62): 419 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 367 us MUL_MAT_ID(ffn_moe_down-63): 203 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 415 us MUL_MAT_ID(ffn_moe_down-64): 414 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 405 us MUL_MAT_ID(ffn_moe_down-65): 203 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 341 us MUL_MAT_ID(ffn_moe_down-66): 371 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 407 us MUL_MAT_ID(ffn_moe_down-67): 388 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 390 us MUL_MAT_ID(ffn_moe_down-68): 429 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 358 us MUL_MAT_ID(ffn_moe_down-69): 201 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 373 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 382 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 348 us MUL_MAT_ID(ffn_moe_down-72): 212 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 402 us MUL_MAT_ID(ffn_moe_down-73): 413 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 383 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 358 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 376 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 348 us MUL_MAT_ID(ffn_moe_down-77): 483 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 386 us MUL_MAT_ID(ffn_moe_down-78): 204 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 375 us MUL_MAT_ID(ffn_moe_down-79): 211 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 353 us MUL_MAT_ID(ffn_moe_down-80): 234 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 411 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 387 us MUL_MAT_ID(ffn_moe_down-83): 192 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 394 us MUL_MAT_ID(ffn_moe_down-84): 693 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 387 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 378 us MUL_MAT_ID(ffn_moe_down-86): 186 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 454 us MUL_MAT_ID(ffn_moe_down-87): 199 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 374 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 381 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 406 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 396 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 37 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 377 us MUL_MAT_ID(ffn_moe_down-25): 291 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 419 us MUL_MAT_ID(ffn_moe_down-26): 204 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 416 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 422 us MUL_MAT_ID(ffn_moe_down-28): 221 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 474 us MUL_MAT_ID(ffn_moe_down-29): 408 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 412 us MUL_MAT_ID(ffn_moe_down-30): 474 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 394 us MUL_MAT_ID(ffn_moe_down-31): 192 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 357 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 405 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 456 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 442 us MUL_MAT_ID(ffn_moe_down-35): 201 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 402 us MUL_MAT_ID(ffn_moe_down-36): 192 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 406 us MUL_MAT_ID(ffn_moe_down-37): 238 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 414 us MUL_MAT_ID(ffn_moe_down-38): 209 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 423 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 414 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 410 us MUL_MAT_ID(ffn_moe_down-41): 200 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 434 us MUL_MAT_ID(ffn_moe_down-42): 586 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 360 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 396 us MUL_MAT_ID(ffn_moe_down-44): 311 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 448 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 423 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 353 us MUL_MAT_ID(ffn_moe_down-47): 200 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 380 us MUL_MAT_ID(ffn_moe_down-48): 204 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 445 us MUL_MAT_ID(ffn_moe_down-49): 548 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 451 us MUL_MAT_ID(ffn_moe_down-50): 213 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 451 us MUL_MAT_ID(ffn_moe_down-51): 205 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 440 us MUL_MAT_ID(ffn_moe_down-52): 197 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 416 us MUL_MAT_ID(ffn_moe_down-53): 278 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 401 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 447 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 435 us MUL_MAT_ID(ffn_moe_down-57): 200 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 433 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 433 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 441 us MUL_MAT_ID(ffn_moe_down-60): 202 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 423 us MUL_MAT_ID(ffn_moe_down-61): 186 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 401 us MUL_MAT_ID(ffn_moe_down-62): 534 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 352 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 440 us MUL_MAT_ID(ffn_moe_down-64): 207 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 415 us MUL_MAT_ID(ffn_moe_down-65): 200 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 426 us MUL_MAT_ID(ffn_moe_down-66): 183 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 368 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 446 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 380 us MUL_MAT_ID(ffn_moe_down-68): 180 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 357 us MUL_MAT_ID(ffn_moe_down-69): 205 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 426 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 457 us MUL_MAT_ID(ffn_moe_down-71): 222 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 420 us MUL_MAT_ID(ffn_moe_down-72): 192 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 456 us MUL_MAT_ID(ffn_moe_down-73): 204 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 431 us MUL_MAT_ID(ffn_moe_down-74): 562 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 413 us MUL_MAT_ID(ffn_moe_down-75): 188 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 447 us MUL_MAT_ID(ffn_moe_down-76): 534 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 439 us MUL_MAT_ID(ffn_moe_down-77): 205 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 458 us MUL_MAT_ID(ffn_moe_down-78): 216 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 434 us MUL_MAT_ID(ffn_moe_down-79): 189 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 353 us MUL_MAT_ID(ffn_moe_down-80): 185 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 363 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 393 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 399 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 381 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 392 us MUL_MAT_ID(ffn_moe_down-86): 224 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 389 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 392 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 373 us MUL_MAT_ID(ffn_moe_down-89): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 407 us MUL_MAT_ID(ffn_moe_down-90): 241 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 405 us MUL_MAT_ID(ffn_moe_down-91): 225 us ggml_barrier(...): 14 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 421 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 412 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 445 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 476 us MUL_MAT_ID(ffn_moe_down-28): 180 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 445 us MUL_MAT_ID(ffn_moe_down-29): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 415 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 346 us MUL_MAT_ID(ffn_moe_down-31): 289 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 451 us MUL_MAT_ID(ffn_moe_down-32): 416 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 397 us MUL_MAT_ID(ffn_moe_down-33): 206 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 502 us MUL_MAT_ID(ffn_moe_down-34): 399 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 398 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 399 us MUL_MAT_ID(ffn_moe_down-36): 207 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 401 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 438 us MUL_MAT_ID(ffn_moe_down-38): 295 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 410 us MUL_MAT_ID(ffn_moe_down-39): 231 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 454 us MUL_MAT_ID(ffn_moe_down-40): 281 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 388 us MUL_MAT_ID(ffn_moe_down-41): 406 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 398 us MUL_MAT_ID(ffn_moe_down-42): 438 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 402 us MUL_MAT_ID(ffn_moe_down-43): 458 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 397 us MUL_MAT_ID(ffn_moe_down-44): 524 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 353 us MUL_MAT_ID(ffn_moe_down-45): 209 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 385 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 353 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 429 us MUL_MAT_ID(ffn_moe_down-48): 454 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 380 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 505 us MUL_MAT_ID(ffn_moe_down-50): 199 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 414 us MUL_MAT_ID(ffn_moe_down-51): 214 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 454 us MUL_MAT_ID(ffn_moe_down-52): 216 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 445 us MUL_MAT_ID(ffn_moe_down-53): 765 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 351 us MUL_MAT_ID(ffn_moe_down-54): 188 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 374 us MUL_MAT_ID(ffn_moe_down-55): 204 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 359 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 453 us MUL_MAT_ID(ffn_moe_down-57): 558 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 429 us MUL_MAT_ID(ffn_moe_down-58): 202 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 406 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 429 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 371 us MUL_MAT_ID(ffn_moe_down-61): 210 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 373 us MUL_MAT_ID(ffn_moe_down-62): 285 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 403 us MUL_MAT_ID(ffn_moe_down-63): 209 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 420 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 445 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 389 us MUL_MAT_ID(ffn_moe_down-66): 320 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 465 us MUL_MAT_ID(ffn_moe_down-67): 211 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 367 us MUL_MAT_ID(ffn_moe_down-68): 201 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 439 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 380 us MUL_MAT_ID(ffn_moe_down-70): 191 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 482 us MUL_MAT_ID(ffn_moe_down-71): 423 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 424 us MUL_MAT_ID(ffn_moe_down-72): 466 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 522 us MUL_MAT_ID(ffn_moe_down-73): 198 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 393 us MUL_MAT_ID(ffn_moe_down-74): 222 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 435 us MUL_MAT_ID(ffn_moe_down-75): 229 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 432 us MUL_MAT_ID(ffn_moe_down-76): 199 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 427 us MUL_MAT_ID(ffn_moe_down-77): 228 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 512 us MUL_MAT_ID(ffn_moe_down-78): 345 us ggml_barrier(...): 424 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 405 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 397 us MUL_MAT_ID(ffn_moe_down-80): 285 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 479 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 419 us MUL_MAT_ID(ffn_moe_down-82): 272 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 425 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 362 us MUL_MAT_ID(ffn_moe_down-84): 653 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 388 us MUL_MAT_ID(ffn_moe_down-85): 242 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 410 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 473 us MUL_MAT_ID(ffn_moe_down-87): 249 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 493 us MUL_MAT_ID(ffn_moe_down-88): 247 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 519 us MUL_MAT_ID(ffn_moe_down-89): 232 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 400 us MUL_MAT_ID(ffn_moe_down-90): 226 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 420 us MUL_MAT_ID(ffn_moe_down-91): 553 us ggml_barrier(...): 12 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 424 us MUL_MAT_ID(ffn_moe_down-25): 191 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 396 us MUL_MAT_ID(ffn_moe_down-26): 206 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 348 us MUL_MAT_ID(ffn_moe_down-27): 211 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 416 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 470 us MUL_MAT_ID(ffn_moe_down-29): 455 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 386 us MUL_MAT_ID(ffn_moe_down-30): 390 us ggml_barrier(...): 380 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 380 us MUL_MAT_ID(ffn_moe_down-31): 209 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 452 us MUL_MAT_ID(ffn_moe_down-32): 267 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 450 us MUL_MAT_ID(ffn_moe_down-33): 184 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 412 us MUL_MAT_ID(ffn_moe_down-34): 188 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 411 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 391 us MUL_MAT_ID(ffn_moe_down-36): 230 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 426 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 432 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 345 us MUL_MAT_ID(ffn_moe_down-39): 243 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 415 us MUL_MAT_ID(ffn_moe_down-40): 239 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 393 us MUL_MAT_ID(ffn_moe_down-41): 375 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 383 us MUL_MAT_ID(ffn_moe_down-42): 432 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 496 us MUL_MAT_ID(ffn_moe_down-43): 173 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 383 us MUL_MAT_ID(ffn_moe_down-44): 187 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 350 us MUL_MAT_ID(ffn_moe_down-45): 167 us ggml_barrier(...): 364 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 403 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 424 us MUL_MAT_ID(ffn_moe_down-47): 188 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 403 us MUL_MAT_ID(ffn_moe_down-48): 275 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 413 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 397 us MUL_MAT_ID(ffn_moe_down-50): 216 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 404 us MUL_MAT_ID(ffn_moe_down-51): 203 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 497 us MUL_MAT_ID(ffn_moe_down-52): 252 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 424 us MUL_MAT_ID(ffn_moe_down-53): 268 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 404 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 417 us MUL_MAT_ID(ffn_moe_down-55): 194 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 423 us MUL_MAT_ID(ffn_moe_down-56): 232 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 408 us MUL_MAT_ID(ffn_moe_down-57): 337 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 406 us MUL_MAT_ID(ffn_moe_down-58): 174 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 421 us MUL_MAT_ID(ffn_moe_down-59): 216 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 378 us MUL_MAT_ID(ffn_moe_down-60): 203 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 403 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 432 us MUL_MAT_ID(ffn_moe_down-62): 206 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 432 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 420 us MUL_MAT_ID(ffn_moe_down-64): 212 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 418 us MUL_MAT_ID(ffn_moe_down-65): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 419 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 420 us MUL_MAT_ID(ffn_moe_down-67): 398 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 441 us MUL_MAT_ID(ffn_moe_down-68): 202 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 431 us MUL_MAT_ID(ffn_moe_down-69): 203 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 532 us MUL_MAT_ID(ffn_moe_down-70): 166 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 439 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 395 us MUL_MAT_ID(ffn_moe_down-72): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 440 us MUL_MAT_ID(ffn_moe_down-73): 213 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 429 us MUL_MAT_ID(ffn_moe_down-74): 301 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 445 us MUL_MAT_ID(ffn_moe_down-75): 265 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 425 us MUL_MAT_ID(ffn_moe_down-76): 433 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 401 us MUL_MAT_ID(ffn_moe_down-77): 208 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 459 us MUL_MAT_ID(ffn_moe_down-78): 195 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 397 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 408 us MUL_MAT_ID(ffn_moe_down-80): 184 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 399 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 437 us MUL_MAT_ID(ffn_moe_down-82): 184 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 429 us MUL_MAT_ID(ffn_moe_down-83): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 434 us MUL_MAT_ID(ffn_moe_down-84): 636 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 384 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 373 us MUL_MAT_ID(ffn_moe_down-86): 177 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 433 us MUL_MAT_ID(ffn_moe_down-87): 510 us ggml_barrier(...): 535 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 349 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 493 us MUL_MAT_ID(ffn_moe_down-89): 206 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 382 us MUL_MAT_ID(ffn_moe_down-90): 236 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 384 us MUL_MAT_ID(ffn_moe_down-91): 405 us ggml_barrier(...): 59 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 376 us MUL_MAT_ID(ffn_moe_down-25): 414 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 372 us MUL_MAT_ID(ffn_moe_down-26): 220 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 451 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 452 us MUL_MAT_ID(ffn_moe_down-28): 184 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 424 us MUL_MAT_ID(ffn_moe_down-29): 206 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 460 us MUL_MAT_ID(ffn_moe_down-30): 185 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 430 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 414 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 481 us MUL_MAT_ID(ffn_moe_down-33): 230 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 460 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 406 us MUL_MAT_ID(ffn_moe_down-35): 206 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 474 us MUL_MAT_ID(ffn_moe_down-36): 191 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 390 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 417 us MUL_MAT_ID(ffn_moe_down-38): 188 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 416 us MUL_MAT_ID(ffn_moe_down-39): 231 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 452 us MUL_MAT_ID(ffn_moe_down-40): 619 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 358 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 426 us MUL_MAT_ID(ffn_moe_down-42): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 446 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 460 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 420 us MUL_MAT_ID(ffn_moe_down-45): 208 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 409 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 394 us MUL_MAT_ID(ffn_moe_down-47): 283 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 509 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 376 us MUL_MAT_ID(ffn_moe_down-49): 200 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 396 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 424 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 442 us MUL_MAT_ID(ffn_moe_down-52): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 346 us MUL_MAT_ID(ffn_moe_down-53): 413 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 548 us MUL_MAT_ID(ffn_moe_down-54): 340 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 496 us MUL_MAT_ID(ffn_moe_down-55): 205 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 426 us MUL_MAT_ID(ffn_moe_down-56): 224 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 410 us MUL_MAT_ID(ffn_moe_down-57): 208 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 436 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 473 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 375 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 422 us MUL_MAT_ID(ffn_moe_down-61): 314 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 437 us MUL_MAT_ID(ffn_moe_down-62): 376 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 399 us MUL_MAT_ID(ffn_moe_down-63): 195 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 395 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 439 us MUL_MAT_ID(ffn_moe_down-65): 188 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 427 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 416 us MUL_MAT_ID(ffn_moe_down-67): 192 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 405 us MUL_MAT_ID(ffn_moe_down-68): 201 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 393 us MUL_MAT_ID(ffn_moe_down-69): 172 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 385 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 412 us MUL_MAT_ID(ffn_moe_down-71): 202 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 426 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 436 us MUL_MAT_ID(ffn_moe_down-73): 299 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 438 us MUL_MAT_ID(ffn_moe_down-74): 224 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 419 us MUL_MAT_ID(ffn_moe_down-75): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 427 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 426 us MUL_MAT_ID(ffn_moe_down-77): 193 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 376 us MUL_MAT_ID(ffn_moe_down-78): 575 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 453 us MUL_MAT_ID(ffn_moe_down-79): 205 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 407 us MUL_MAT_ID(ffn_moe_down-80): 207 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 430 us MUL_MAT_ID(ffn_moe_down-81): 216 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 354 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 381 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 405 us MUL_MAT_ID(ffn_moe_down-84): 240 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 432 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 411 us MUL_MAT_ID(ffn_moe_down-86): 189 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 415 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 379 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 409 us MUL_MAT_ID(ffn_moe_down-89): 235 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 432 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 388 us MUL_MAT_ID(ffn_moe_down-91): 321 us ggml_barrier(...): 30 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 377 us MUL_MAT_ID(ffn_moe_down-25): 268 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 403 us MUL_MAT_ID(ffn_moe_down-26): 204 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 421 us MUL_MAT_ID(ffn_moe_down-27): 191 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 405 us MUL_MAT_ID(ffn_moe_down-28): 186 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 425 us MUL_MAT_ID(ffn_moe_down-29): 197 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 360 us MUL_MAT_ID(ffn_moe_down-30): 186 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 408 us MUL_MAT_ID(ffn_moe_down-31): 188 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 353 us MUL_MAT_ID(ffn_moe_down-32): 231 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 385 us MUL_MAT_ID(ffn_moe_down-33): 236 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 392 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 448 us MUL_MAT_ID(ffn_moe_down-35): 202 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 379 us MUL_MAT_ID(ffn_moe_down-36): 185 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 322 us MUL_MAT_ID(ffn_moe_down-37): 294 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 436 us MUL_MAT_ID(ffn_moe_down-38): 197 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 376 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 374 us MUL_MAT_ID(ffn_moe_down-40): 321 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 423 us MUL_MAT_ID(ffn_moe_down-41): 207 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 443 us MUL_MAT_ID(ffn_moe_down-42): 208 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 375 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 448 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 403 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 353 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 460 us MUL_MAT_ID(ffn_moe_down-47): 198 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 402 us MUL_MAT_ID(ffn_moe_down-48): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 439 us MUL_MAT_ID(ffn_moe_down-49): 385 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 404 us MUL_MAT_ID(ffn_moe_down-50): 193 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 444 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 425 us MUL_MAT_ID(ffn_moe_down-52): 235 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 363 us MUL_MAT_ID(ffn_moe_down-53): 439 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 438 us MUL_MAT_ID(ffn_moe_down-54): 413 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 451 us MUL_MAT_ID(ffn_moe_down-55): 204 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 444 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 460 us MUL_MAT_ID(ffn_moe_down-57): 292 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 436 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 377 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 420 us MUL_MAT_ID(ffn_moe_down-60): 169 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 423 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 399 us MUL_MAT_ID(ffn_moe_down-62): 260 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 387 us MUL_MAT_ID(ffn_moe_down-63): 285 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 451 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 364 us MUL_MAT_ID(ffn_moe_down-65): 186 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 412 us MUL_MAT_ID(ffn_moe_down-66): 197 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 357 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 521 us MUL_MAT_ID(ffn_moe_down-68): 217 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 417 us MUL_MAT_ID(ffn_moe_down-69): 341 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 394 us MUL_MAT_ID(ffn_moe_down-70): 198 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 434 us MUL_MAT_ID(ffn_moe_down-71): 220 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 393 us MUL_MAT_ID(ffn_moe_down-72): 201 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 368 us MUL_MAT_ID(ffn_moe_down-73): 195 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 401 us MUL_MAT_ID(ffn_moe_down-74): 209 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 360 us MUL_MAT_ID(ffn_moe_down-75): 212 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 443 us MUL_MAT_ID(ffn_moe_down-76): 203 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 433 us MUL_MAT_ID(ffn_moe_down-77): 199 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 367 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 372 us MUL_MAT_ID(ffn_moe_down-79): 185 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 416 us MUL_MAT_ID(ffn_moe_down-80): 240 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 364 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 401 us MUL_MAT_ID(ffn_moe_down-82): 594 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 378 us MUL_MAT_ID(ffn_moe_down-83): 215 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 381 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 405 us MUL_MAT_ID(ffn_moe_down-85): 234 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 401 us MUL_MAT_ID(ffn_moe_down-86): 439 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 402 us MUL_MAT_ID(ffn_moe_down-87): 300 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 388 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 385 us MUL_MAT_ID(ffn_moe_down-89): 581 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 396 us MUL_MAT_ID(ffn_moe_down-90): 652 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 387 us MUL_MAT_ID(ffn_moe_down-91): 226 us ggml_barrier(...): 9 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 371 us MUL_MAT_ID(ffn_moe_down-25): 178 us ggml_barrier(...): 340 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 424 us MUL_MAT_ID(ffn_moe_down-26): 188 us ggml_barrier(...): 544 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 556 us MUL_MAT_ID(ffn_moe_down-27): 237 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 458 us MUL_MAT_ID(ffn_moe_down-28): 243 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 491 us MUL_MAT_ID(ffn_moe_down-29): 228 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 409 us MUL_MAT_ID(ffn_moe_down-30): 266 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 442 us MUL_MAT_ID(ffn_moe_down-31): 395 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 411 us MUL_MAT_ID(ffn_moe_down-32): 237 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 504 us MUL_MAT_ID(ffn_moe_down-33): 575 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 434 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 358 us MUL_MAT_ID(ffn_moe_down-35): 206 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 424 us MUL_MAT_ID(ffn_moe_down-36): 507 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 408 us MUL_MAT_ID(ffn_moe_down-37): 404 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 379 us MUL_MAT_ID(ffn_moe_down-38): 626 us ggml_barrier(...): 383 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 463 us MUL_MAT_ID(ffn_moe_down-39): 783 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 427 us MUL_MAT_ID(ffn_moe_down-40): 265 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 381 us MUL_MAT_ID(ffn_moe_down-41): 275 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 416 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 438 us MUL_MAT_ID(ffn_moe_down-43): 197 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 377 us MUL_MAT_ID(ffn_moe_down-44): 205 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 396 us MUL_MAT_ID(ffn_moe_down-45): 217 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 403 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 353 us MUL_MAT_ID(ffn_moe_down-47): 185 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 390 us MUL_MAT_ID(ffn_moe_down-48): 212 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 391 us MUL_MAT_ID(ffn_moe_down-49): 223 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 448 us MUL_MAT_ID(ffn_moe_down-50): 220 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 455 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 454 us MUL_MAT_ID(ffn_moe_down-52): 203 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 423 us MUL_MAT_ID(ffn_moe_down-53): 454 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 350 us MUL_MAT_ID(ffn_moe_down-54): 210 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 394 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 405 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 424 us MUL_MAT_ID(ffn_moe_down-57): 207 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 435 us MUL_MAT_ID(ffn_moe_down-58): 210 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 427 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 451 us MUL_MAT_ID(ffn_moe_down-60): 453 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 423 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 354 us MUL_MAT_ID(ffn_moe_down-62): 387 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 388 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 355 us MUL_MAT_ID(ffn_moe_down-64): 196 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 384 us MUL_MAT_ID(ffn_moe_down-65): 206 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 392 us MUL_MAT_ID(ffn_moe_down-66): 501 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 408 us MUL_MAT_ID(ffn_moe_down-67): 514 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 406 us MUL_MAT_ID(ffn_moe_down-68): 212 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 406 us MUL_MAT_ID(ffn_moe_down-69): 188 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 403 us MUL_MAT_ID(ffn_moe_down-70): 193 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 390 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 327 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 398 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 382 us MUL_MAT_ID(ffn_moe_down-73): 202 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 390 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 386 us MUL_MAT_ID(ffn_moe_down-75): 216 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 409 us MUL_MAT_ID(ffn_moe_down-76): 213 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 445 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 471 us MUL_MAT_ID(ffn_moe_down-78): 369 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 519 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 407 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 413 us MUL_MAT_ID(ffn_moe_down-81): 257 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 482 us MUL_MAT_ID(ffn_moe_down-82): 257 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 475 us MUL_MAT_ID(ffn_moe_down-83): 277 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 436 us MUL_MAT_ID(ffn_moe_down-84): 250 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 393 us MUL_MAT_ID(ffn_moe_down-85): 243 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 418 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 199 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 373 us MUL_MAT_ID(ffn_moe_down-88): 203 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 376 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 370 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 442 us MUL_MAT_ID(ffn_moe_down-91): 372 us ggml_barrier(...): 19 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 405 us MUL_MAT_ID(ffn_moe_down-25): 183 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 440 us MUL_MAT_ID(ffn_moe_down-26): 185 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 357 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 393 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 439 us MUL_MAT_ID(ffn_moe_down-29): 532 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 432 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 375 us MUL_MAT_ID(ffn_moe_down-31): 202 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 375 us MUL_MAT_ID(ffn_moe_down-32): 211 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 391 us MUL_MAT_ID(ffn_moe_down-33): 189 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 385 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 439 us MUL_MAT_ID(ffn_moe_down-35): 327 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 364 us MUL_MAT_ID(ffn_moe_down-36): 334 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 408 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 401 us MUL_MAT_ID(ffn_moe_down-38): 276 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 426 us MUL_MAT_ID(ffn_moe_down-39): 265 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 691 us MUL_MAT_ID(ffn_moe_down-40): 306 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 429 us MUL_MAT_ID(ffn_moe_down-41): 215 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 416 us MUL_MAT_ID(ffn_moe_down-42): 211 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 418 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 402 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 387 us MUL_MAT_ID(ffn_moe_down-45): 532 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 385 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 396 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 350 us MUL_MAT_ID(ffn_moe_down-48): 200 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 411 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 240 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 553 us MUL_MAT_ID(ffn_moe_down-51): 283 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 394 us MUL_MAT_ID(ffn_moe_down-52): 187 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 377 us MUL_MAT_ID(ffn_moe_down-53): 315 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 412 us MUL_MAT_ID(ffn_moe_down-54): 204 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 422 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 382 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 426 us MUL_MAT_ID(ffn_moe_down-57): 223 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 386 us MUL_MAT_ID(ffn_moe_down-58): 189 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 392 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 481 us MUL_MAT_ID(ffn_moe_down-61): 204 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 426 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 371 us MUL_MAT_ID(ffn_moe_down-63): 199 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 397 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 355 us MUL_MAT_ID(ffn_moe_down-65): 195 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 416 us MUL_MAT_ID(ffn_moe_down-66): 201 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 375 us MUL_MAT_ID(ffn_moe_down-67): 227 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 393 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 430 us MUL_MAT_ID(ffn_moe_down-69): 184 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 399 us MUL_MAT_ID(ffn_moe_down-70): 590 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 398 us MUL_MAT_ID(ffn_moe_down-71): 455 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 409 us MUL_MAT_ID(ffn_moe_down-72): 224 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 424 us MUL_MAT_ID(ffn_moe_down-73): 371 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 380 us MUL_MAT_ID(ffn_moe_down-74): 215 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 384 us MUL_MAT_ID(ffn_moe_down-75): 199 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 419 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 398 us MUL_MAT_ID(ffn_moe_down-77): 200 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 366 us MUL_MAT_ID(ffn_moe_down-78): 210 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 375 us MUL_MAT_ID(ffn_moe_down-79): 356 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 405 us MUL_MAT_ID(ffn_moe_down-80): 183 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 351 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 376 us MUL_MAT_ID(ffn_moe_down-82): 570 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 360 us MUL_MAT_ID(ffn_moe_down-83): 187 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 417 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 386 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 369 us MUL_MAT_ID(ffn_moe_down-86): 414 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 362 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 387 us MUL_MAT_ID(ffn_moe_down-88): 227 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 393 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 420 us MUL_MAT_ID(ffn_moe_down-90): 417 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 398 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 25 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 380 us MUL_MAT_ID(ffn_moe_down-25): 239 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 383 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 393 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 364 us MUL_MAT_ID(ffn_moe_down-28): 598 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 419 us MUL_MAT_ID(ffn_moe_down-29): 185 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 440 us MUL_MAT_ID(ffn_moe_down-30): 234 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 513 us MUL_MAT_ID(ffn_moe_down-31): 233 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 404 us MUL_MAT_ID(ffn_moe_down-32): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 433 us MUL_MAT_ID(ffn_moe_down-33): 367 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 379 us MUL_MAT_ID(ffn_moe_down-34): 210 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 388 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 405 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 405 us MUL_MAT_ID(ffn_moe_down-37): 238 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 397 us MUL_MAT_ID(ffn_moe_down-38): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 395 us MUL_MAT_ID(ffn_moe_down-39): 644 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 385 us MUL_MAT_ID(ffn_moe_down-40): 247 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 458 us MUL_MAT_ID(ffn_moe_down-41): 216 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 460 us MUL_MAT_ID(ffn_moe_down-42): 426 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 409 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 423 us MUL_MAT_ID(ffn_moe_down-44): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 430 us MUL_MAT_ID(ffn_moe_down-45): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 394 us MUL_MAT_ID(ffn_moe_down-46): 230 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 553 us MUL_MAT_ID(ffn_moe_down-47): 340 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 650 us MUL_MAT_ID(ffn_moe_down-48): 268 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 485 us MUL_MAT_ID(ffn_moe_down-49): 246 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 473 us MUL_MAT_ID(ffn_moe_down-50): 431 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 616 us MUL_MAT_ID(ffn_moe_down-51): 208 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 426 us MUL_MAT_ID(ffn_moe_down-52): 444 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 424 us MUL_MAT_ID(ffn_moe_down-53): 569 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 399 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 447 us MUL_MAT_ID(ffn_moe_down-55): 378 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 395 us MUL_MAT_ID(ffn_moe_down-57): 204 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 503 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 404 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 368 us MUL_MAT_ID(ffn_moe_down-60): 203 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 497 us MUL_MAT_ID(ffn_moe_down-61): 734 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 362 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 449 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 380 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 429 us MUL_MAT_ID(ffn_moe_down-65): 196 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 434 us MUL_MAT_ID(ffn_moe_down-66): 285 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 415 us MUL_MAT_ID(ffn_moe_down-67): 216 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 428 us MUL_MAT_ID(ffn_moe_down-68): 442 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 414 us MUL_MAT_ID(ffn_moe_down-69): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 422 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 399 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 346 us MUL_MAT_ID(ffn_moe_down-72): 456 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 401 us MUL_MAT_ID(ffn_moe_down-73): 203 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 425 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 381 us MUL_MAT_ID(ffn_moe_down-75): 202 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 414 us MUL_MAT_ID(ffn_moe_down-76): 187 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 439 us MUL_MAT_ID(ffn_moe_down-77): 205 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 406 us MUL_MAT_ID(ffn_moe_down-78): 182 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 357 us MUL_MAT_ID(ffn_moe_down-79): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 429 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 383 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 387 us MUL_MAT_ID(ffn_moe_down-82): 273 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 403 us MUL_MAT_ID(ffn_moe_down-83): 441 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 541 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 415 us MUL_MAT_ID(ffn_moe_down-85): 226 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 359 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 386 us MUL_MAT_ID(ffn_moe_down-87): 566 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 360 us MUL_MAT_ID(ffn_moe_down-88): 203 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 384 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 425 us MUL_MAT_ID(ffn_moe_down-90): 238 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 432 us MUL_MAT_ID(ffn_moe_down-91): 258 us ggml_barrier(...): 10 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 430 us MUL_MAT_ID(ffn_moe_down-26): 252 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 463 us MUL_MAT_ID(ffn_moe_down-27): 185 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 391 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 423 us MUL_MAT_ID(ffn_moe_down-29): 443 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 405 us MUL_MAT_ID(ffn_moe_down-30): 208 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 417 us MUL_MAT_ID(ffn_moe_down-31): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 401 us MUL_MAT_ID(ffn_moe_down-32): 245 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 447 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 419 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 395 us MUL_MAT_ID(ffn_moe_down-35): 244 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 366 us MUL_MAT_ID(ffn_moe_down-36): 257 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 361 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 356 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 428 us MUL_MAT_ID(ffn_moe_down-39): 310 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 389 us MUL_MAT_ID(ffn_moe_down-40): 323 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 388 us MUL_MAT_ID(ffn_moe_down-41): 227 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 400 us MUL_MAT_ID(ffn_moe_down-42): 208 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 420 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 394 us MUL_MAT_ID(ffn_moe_down-44): 225 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 373 us MUL_MAT_ID(ffn_moe_down-45): 474 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 401 us MUL_MAT_ID(ffn_moe_down-46): 212 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 386 us MUL_MAT_ID(ffn_moe_down-47): 208 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 402 us MUL_MAT_ID(ffn_moe_down-48): 202 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 342 us MUL_MAT_ID(ffn_moe_down-49): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 347 us MUL_MAT_ID(ffn_moe_down-50): 216 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 422 us MUL_MAT_ID(ffn_moe_down-51): 218 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 412 us MUL_MAT_ID(ffn_moe_down-52): 380 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 389 us MUL_MAT_ID(ffn_moe_down-53): 204 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 442 us MUL_MAT_ID(ffn_moe_down-54): 286 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 410 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 346 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 427 us MUL_MAT_ID(ffn_moe_down-57): 196 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 413 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 413 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 418 us MUL_MAT_ID(ffn_moe_down-60): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 364 us MUL_MAT_ID(ffn_moe_down-61): 298 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 419 us MUL_MAT_ID(ffn_moe_down-62): 221 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 694 us MUL_MAT_ID(ffn_moe_down-63): 382 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 347 us MUL_MAT_ID(ffn_moe_down-64): 200 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 458 us MUL_MAT_ID(ffn_moe_down-65): 203 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 392 us MUL_MAT_ID(ffn_moe_down-66): 211 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 393 us MUL_MAT_ID(ffn_moe_down-67): 217 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 383 us MUL_MAT_ID(ffn_moe_down-68): 589 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 379 us MUL_MAT_ID(ffn_moe_down-69): 219 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 359 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 414 us MUL_MAT_ID(ffn_moe_down-71): 403 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 405 us MUL_MAT_ID(ffn_moe_down-72): 215 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 417 us MUL_MAT_ID(ffn_moe_down-73): 373 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 644 us MUL_MAT_ID(ffn_moe_down-74): 292 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 405 us MUL_MAT_ID(ffn_moe_down-75): 379 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 441 us MUL_MAT_ID(ffn_moe_down-76): 200 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 454 us MUL_MAT_ID(ffn_moe_down-77): 544 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 424 us MUL_MAT_ID(ffn_moe_down-78): 188 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 454 us MUL_MAT_ID(ffn_moe_down-79): 379 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 432 us MUL_MAT_ID(ffn_moe_down-80): 207 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 427 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 417 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 374 us MUL_MAT_ID(ffn_moe_down-83): 390 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 386 us MUL_MAT_ID(ffn_moe_down-84): 249 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 448 us MUL_MAT_ID(ffn_moe_down-85): 600 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 370 us MUL_MAT_ID(ffn_moe_down-86): 182 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 421 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 360 us MUL_MAT_ID(ffn_moe_down-88): 204 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 419 us MUL_MAT_ID(ffn_moe_down-89): 204 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 401 us MUL_MAT_ID(ffn_moe_down-90): 238 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 417 us MUL_MAT_ID(ffn_moe_down-91): 400 us ggml_barrier(...): 22 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 376 us MUL_MAT_ID(ffn_moe_down-25): 208 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 458 us MUL_MAT_ID(ffn_moe_down-26): 199 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 443 us MUL_MAT_ID(ffn_moe_down-27): 301 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 429 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 439 us MUL_MAT_ID(ffn_moe_down-29): 192 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 430 us MUL_MAT_ID(ffn_moe_down-30): 191 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 500 us MUL_MAT_ID(ffn_moe_down-31): 184 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 432 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 345 us MUL_MAT_ID(ffn_moe_down-33): 380 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 405 us MUL_MAT_ID(ffn_moe_down-34): 188 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 387 us MUL_MAT_ID(ffn_moe_down-35): 189 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 391 us MUL_MAT_ID(ffn_moe_down-36): 230 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 398 us MUL_MAT_ID(ffn_moe_down-37): 238 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 416 us MUL_MAT_ID(ffn_moe_down-38): 185 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 375 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 396 us MUL_MAT_ID(ffn_moe_down-40): 228 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 404 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 431 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 372 us MUL_MAT_ID(ffn_moe_down-43): 396 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 391 us MUL_MAT_ID(ffn_moe_down-44): 418 us ggml_barrier(...): 353 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 409 us MUL_MAT_ID(ffn_moe_down-45): 271 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 407 us MUL_MAT_ID(ffn_moe_down-46): 205 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 393 us MUL_MAT_ID(ffn_moe_down-47): 219 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 426 us MUL_MAT_ID(ffn_moe_down-48): 180 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 413 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 341 us MUL_MAT_ID(ffn_moe_down-50): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 414 us MUL_MAT_ID(ffn_moe_down-51): 201 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 359 us MUL_MAT_ID(ffn_moe_down-52): 271 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 406 us MUL_MAT_ID(ffn_moe_down-53): 391 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 392 us MUL_MAT_ID(ffn_moe_down-54): 219 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 400 us MUL_MAT_ID(ffn_moe_down-55): 237 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 451 us MUL_MAT_ID(ffn_moe_down-56): 235 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 417 us MUL_MAT_ID(ffn_moe_down-57): 179 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 415 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 365 us MUL_MAT_ID(ffn_moe_down-59): 190 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 418 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 455 us MUL_MAT_ID(ffn_moe_down-61): 202 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 211 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 375 us MUL_MAT_ID(ffn_moe_down-63): 526 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 333 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 553 us MUL_MAT_ID(ffn_moe_down-65): 361 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 440 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 484 us MUL_MAT_ID(ffn_moe_down-67): 227 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 440 us MUL_MAT_ID(ffn_moe_down-68): 206 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 440 us MUL_MAT_ID(ffn_moe_down-69): 214 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 409 us MUL_MAT_ID(ffn_moe_down-70): 309 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 381 us MUL_MAT_ID(ffn_moe_down-71): 248 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 421 us MUL_MAT_ID(ffn_moe_down-72): 205 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 662 us MUL_MAT_ID(ffn_moe_down-73): 312 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 350 us MUL_MAT_ID(ffn_moe_down-74): 203 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 401 us MUL_MAT_ID(ffn_moe_down-75): 246 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 399 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 419 us MUL_MAT_ID(ffn_moe_down-77): 193 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 369 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 457 us MUL_MAT_ID(ffn_moe_down-79): 217 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 385 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 387 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 339 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 361 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 452 us MUL_MAT_ID(ffn_moe_down-84): 245 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 409 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 379 us MUL_MAT_ID(ffn_moe_down-86): 198 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 392 us MUL_MAT_ID(ffn_moe_down-87): 576 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 370 us MUL_MAT_ID(ffn_moe_down-88): 306 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 405 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 369 us MUL_MAT_ID(ffn_moe_down-90): 225 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 392 us MUL_MAT_ID(ffn_moe_down-91): 350 us ggml_barrier(...): 37 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 371 us MUL_MAT_ID(ffn_moe_down-25): 204 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 407 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 343 us MUL_MAT_ID(ffn_moe_down-27): 213 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 382 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 395 us MUL_MAT_ID(ffn_moe_down-29): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 385 us MUL_MAT_ID(ffn_moe_down-31): 188 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 390 us MUL_MAT_ID(ffn_moe_down-32): 190 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 341 us MUL_MAT_ID(ffn_moe_down-33): 221 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 404 us MUL_MAT_ID(ffn_moe_down-34): 206 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 424 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 427 us MUL_MAT_ID(ffn_moe_down-36): 215 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 393 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 403 us MUL_MAT_ID(ffn_moe_down-38): 183 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 366 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 374 us MUL_MAT_ID(ffn_moe_down-40): 215 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 440 us MUL_MAT_ID(ffn_moe_down-41): 575 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 370 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 413 us MUL_MAT_ID(ffn_moe_down-43): 246 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 418 us MUL_MAT_ID(ffn_moe_down-44): 206 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 337 us MUL_MAT_ID(ffn_moe_down-45): 268 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 370 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 389 us MUL_MAT_ID(ffn_moe_down-47): 216 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 439 us MUL_MAT_ID(ffn_moe_down-48): 226 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 379 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 375 us MUL_MAT_ID(ffn_moe_down-50): 206 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 380 us MUL_MAT_ID(ffn_moe_down-51): 322 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 430 us MUL_MAT_ID(ffn_moe_down-52): 212 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 380 us MUL_MAT_ID(ffn_moe_down-53): 191 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 440 us MUL_MAT_ID(ffn_moe_down-54): 191 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 457 us MUL_MAT_ID(ffn_moe_down-55): 205 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 427 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 466 us MUL_MAT_ID(ffn_moe_down-57): 203 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 428 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 437 us MUL_MAT_ID(ffn_moe_down-59): 298 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 414 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 405 us MUL_MAT_ID(ffn_moe_down-61): 211 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 375 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 496 us MUL_MAT_ID(ffn_moe_down-63): 390 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 403 us MUL_MAT_ID(ffn_moe_down-64): 183 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 405 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 405 us MUL_MAT_ID(ffn_moe_down-66): 238 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 393 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 462 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 368 us MUL_MAT_ID(ffn_moe_down-69): 185 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 406 us MUL_MAT_ID(ffn_moe_down-70): 211 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 390 us MUL_MAT_ID(ffn_moe_down-71): 218 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 522 us MUL_MAT_ID(ffn_moe_down-72): 213 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 417 us MUL_MAT_ID(ffn_moe_down-73): 181 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 420 us MUL_MAT_ID(ffn_moe_down-74): 211 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 442 us MUL_MAT_ID(ffn_moe_down-75): 189 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 382 us MUL_MAT_ID(ffn_moe_down-76): 219 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 420 us MUL_MAT_ID(ffn_moe_down-77): 223 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 410 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 423 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 384 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 406 us MUL_MAT_ID(ffn_moe_down-81): 551 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 365 us MUL_MAT_ID(ffn_moe_down-82): 283 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 421 us MUL_MAT_ID(ffn_moe_down-83): 222 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 339 us MUL_MAT_ID(ffn_moe_down-84): 256 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 420 us MUL_MAT_ID(ffn_moe_down-85): 274 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 474 us MUL_MAT_ID(ffn_moe_down-86): 254 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 693 us MUL_MAT_ID(ffn_moe_down-87): 336 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 429 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 467 us MUL_MAT_ID(ffn_moe_down-89): 208 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 391 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 421 us MUL_MAT_ID(ffn_moe_down-91): 265 us ggml_barrier(...): 240 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 433 us MUL_MAT_ID(ffn_moe_down-25): 403 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 406 us MUL_MAT_ID(ffn_moe_down-26): 185 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 395 us MUL_MAT_ID(ffn_moe_down-27): 230 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 482 us MUL_MAT_ID(ffn_moe_down-28): 474 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 411 us MUL_MAT_ID(ffn_moe_down-29): 204 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 428 us MUL_MAT_ID(ffn_moe_down-30): 441 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 350 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 416 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 410 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 409 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 390 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 356 us MUL_MAT_ID(ffn_moe_down-36): 225 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 398 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 406 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 385 us MUL_MAT_ID(ffn_moe_down-39): 725 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 389 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 414 us MUL_MAT_ID(ffn_moe_down-41): 194 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 433 us MUL_MAT_ID(ffn_moe_down-42): 224 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 400 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 367 us MUL_MAT_ID(ffn_moe_down-44): 193 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 356 us MUL_MAT_ID(ffn_moe_down-45): 220 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 388 us MUL_MAT_ID(ffn_moe_down-46): 216 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 406 us MUL_MAT_ID(ffn_moe_down-47): 213 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 418 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 347 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 409 us MUL_MAT_ID(ffn_moe_down-50): 396 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 471 us MUL_MAT_ID(ffn_moe_down-51): 495 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 407 us MUL_MAT_ID(ffn_moe_down-52): 203 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 366 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 379 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 353 us MUL_MAT_ID(ffn_moe_down-55): 186 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 364 us MUL_MAT_ID(ffn_moe_down-56): 185 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 414 us MUL_MAT_ID(ffn_moe_down-57): 185 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 384 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 381 us MUL_MAT_ID(ffn_moe_down-59): 224 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 401 us MUL_MAT_ID(ffn_moe_down-60): 205 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 348 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 364 us MUL_MAT_ID(ffn_moe_down-63): 199 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 391 us MUL_MAT_ID(ffn_moe_down-64): 230 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 386 us MUL_MAT_ID(ffn_moe_down-65): 199 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 430 us MUL_MAT_ID(ffn_moe_down-66): 218 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 378 us MUL_MAT_ID(ffn_moe_down-67): 219 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 418 us MUL_MAT_ID(ffn_moe_down-68): 188 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 370 us MUL_MAT_ID(ffn_moe_down-69): 198 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 355 us MUL_MAT_ID(ffn_moe_down-70): 183 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 351 us MUL_MAT_ID(ffn_moe_down-71): 209 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 359 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 370 us MUL_MAT_ID(ffn_moe_down-73): 198 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 369 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 396 us MUL_MAT_ID(ffn_moe_down-75): 177 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 410 us MUL_MAT_ID(ffn_moe_down-76): 552 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 377 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 385 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 398 us MUL_MAT_ID(ffn_moe_down-79): 214 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 373 us MUL_MAT_ID(ffn_moe_down-80): 202 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 338 us MUL_MAT_ID(ffn_moe_down-81): 211 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 373 us MUL_MAT_ID(ffn_moe_down-82): 189 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 394 us MUL_MAT_ID(ffn_moe_down-83): 189 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 343 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 394 us MUL_MAT_ID(ffn_moe_down-85): 239 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 404 us MUL_MAT_ID(ffn_moe_down-86): 217 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 454 us MUL_MAT_ID(ffn_moe_down-87): 595 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 367 us MUL_MAT_ID(ffn_moe_down-88): 202 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 395 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 378 us MUL_MAT_ID(ffn_moe_down-90): 246 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 403 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 40 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 208 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 402 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 398 us MUL_MAT_ID(ffn_moe_down-27): 188 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 436 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 431 us MUL_MAT_ID(ffn_moe_down-29): 206 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 400 us MUL_MAT_ID(ffn_moe_down-30): 224 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 623 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 405 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 431 us MUL_MAT_ID(ffn_moe_down-33): 207 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 400 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 360 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 395 us MUL_MAT_ID(ffn_moe_down-36): 348 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 401 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 343 us MUL_MAT_ID(ffn_moe_down-39): 335 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 333 us MUL_MAT_ID(ffn_moe_down-40): 256 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 418 us MUL_MAT_ID(ffn_moe_down-41): 190 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 402 us MUL_MAT_ID(ffn_moe_down-42): 575 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 356 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 386 us MUL_MAT_ID(ffn_moe_down-44): 242 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 456 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 353 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 397 us MUL_MAT_ID(ffn_moe_down-47): 192 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 398 us MUL_MAT_ID(ffn_moe_down-48): 183 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 375 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 367 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 335 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 442 us MUL_MAT_ID(ffn_moe_down-52): 223 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 390 us MUL_MAT_ID(ffn_moe_down-53): 187 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 380 us MUL_MAT_ID(ffn_moe_down-54): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 341 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 402 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 355 us MUL_MAT_ID(ffn_moe_down-57): 173 us ggml_barrier(...): 327 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 398 us MUL_MAT_ID(ffn_moe_down-58): 285 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 380 us MUL_MAT_ID(ffn_moe_down-59): 204 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 416 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 403 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 350 us MUL_MAT_ID(ffn_moe_down-62): 244 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 388 us MUL_MAT_ID(ffn_moe_down-63): 184 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 409 us MUL_MAT_ID(ffn_moe_down-64): 570 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 352 us MUL_MAT_ID(ffn_moe_down-65): 224 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 430 us MUL_MAT_ID(ffn_moe_down-66): 289 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 395 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 428 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 370 us MUL_MAT_ID(ffn_moe_down-69): 210 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 504 us MUL_MAT_ID(ffn_moe_down-70): 511 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 370 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 372 us MUL_MAT_ID(ffn_moe_down-72): 188 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 370 us MUL_MAT_ID(ffn_moe_down-73): 207 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 374 us MUL_MAT_ID(ffn_moe_down-74): 562 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 368 us MUL_MAT_ID(ffn_moe_down-75): 212 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 400 us MUL_MAT_ID(ffn_moe_down-76): 198 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 391 us MUL_MAT_ID(ffn_moe_down-77): 184 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 342 us MUL_MAT_ID(ffn_moe_down-78): 185 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 426 us MUL_MAT_ID(ffn_moe_down-79): 223 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 382 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 407 us MUL_MAT_ID(ffn_moe_down-81): 516 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 368 us MUL_MAT_ID(ffn_moe_down-82): 405 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 381 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 536 us MUL_MAT_ID(ffn_moe_down-84): 216 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 416 us MUL_MAT_ID(ffn_moe_down-85): 228 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 412 us MUL_MAT_ID(ffn_moe_down-86): 211 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 351 us MUL_MAT_ID(ffn_moe_down-87): 307 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 404 us MUL_MAT_ID(ffn_moe_down-88): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 378 us MUL_MAT_ID(ffn_moe_down-89): 206 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 390 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 364 us MUL_MAT_ID(ffn_moe_down-91): 712 us ggml_barrier(...): 101 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 388 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 424 us MUL_MAT_ID(ffn_moe_down-26): 203 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 375 us MUL_MAT_ID(ffn_moe_down-27): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 361 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 389 us MUL_MAT_ID(ffn_moe_down-29): 351 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 448 us MUL_MAT_ID(ffn_moe_down-30): 227 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 455 us MUL_MAT_ID(ffn_moe_down-31): 218 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 612 us MUL_MAT_ID(ffn_moe_down-32): 311 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 368 us MUL_MAT_ID(ffn_moe_down-33): 205 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 359 us MUL_MAT_ID(ffn_moe_down-34): 190 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 378 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 387 us MUL_MAT_ID(ffn_moe_down-36): 206 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 392 us MUL_MAT_ID(ffn_moe_down-37): 225 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 391 us MUL_MAT_ID(ffn_moe_down-38): 375 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 401 us MUL_MAT_ID(ffn_moe_down-39): 250 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 400 us MUL_MAT_ID(ffn_moe_down-40): 228 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 363 us MUL_MAT_ID(ffn_moe_down-41): 327 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 367 us MUL_MAT_ID(ffn_moe_down-42): 190 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 467 us MUL_MAT_ID(ffn_moe_down-43): 315 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 475 us MUL_MAT_ID(ffn_moe_down-44): 187 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 394 us MUL_MAT_ID(ffn_moe_down-45): 180 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 351 us MUL_MAT_ID(ffn_moe_down-46): 180 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 353 us MUL_MAT_ID(ffn_moe_down-47): 408 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 359 us MUL_MAT_ID(ffn_moe_down-48): 232 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 429 us MUL_MAT_ID(ffn_moe_down-49): 314 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 399 us MUL_MAT_ID(ffn_moe_down-50): 206 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 429 us MUL_MAT_ID(ffn_moe_down-51): 188 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 415 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 386 us MUL_MAT_ID(ffn_moe_down-53): 203 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 377 us MUL_MAT_ID(ffn_moe_down-54): 203 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 410 us MUL_MAT_ID(ffn_moe_down-55): 218 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 417 us MUL_MAT_ID(ffn_moe_down-56): 218 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 386 us MUL_MAT_ID(ffn_moe_down-57): 225 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 410 us MUL_MAT_ID(ffn_moe_down-58): 346 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 444 us MUL_MAT_ID(ffn_moe_down-59): 355 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 400 us MUL_MAT_ID(ffn_moe_down-60): 217 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 423 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 401 us MUL_MAT_ID(ffn_moe_down-62): 210 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 394 us MUL_MAT_ID(ffn_moe_down-63): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 416 us MUL_MAT_ID(ffn_moe_down-64): 323 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 410 us MUL_MAT_ID(ffn_moe_down-65): 300 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 414 us MUL_MAT_ID(ffn_moe_down-66): 202 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 435 us MUL_MAT_ID(ffn_moe_down-67): 196 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 424 us MUL_MAT_ID(ffn_moe_down-68): 353 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 396 us MUL_MAT_ID(ffn_moe_down-69): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 390 us MUL_MAT_ID(ffn_moe_down-70): 211 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 389 us MUL_MAT_ID(ffn_moe_down-71): 212 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 389 us MUL_MAT_ID(ffn_moe_down-72): 210 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 468 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 379 us MUL_MAT_ID(ffn_moe_down-74): 200 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 439 us MUL_MAT_ID(ffn_moe_down-75): 202 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 410 us MUL_MAT_ID(ffn_moe_down-76): 201 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 394 us MUL_MAT_ID(ffn_moe_down-77): 207 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 416 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 444 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 543 us MUL_MAT_ID(ffn_moe_down-80): 181 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 426 us MUL_MAT_ID(ffn_moe_down-81): 405 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 405 us MUL_MAT_ID(ffn_moe_down-82): 199 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 401 us MUL_MAT_ID(ffn_moe_down-83): 314 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 405 us MUL_MAT_ID(ffn_moe_down-84): 662 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 383 us MUL_MAT_ID(ffn_moe_down-85): 448 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 394 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 454 us MUL_MAT_ID(ffn_moe_down-87): 207 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 425 us MUL_MAT_ID(ffn_moe_down-88): 181 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 452 us MUL_MAT_ID(ffn_moe_down-89): 373 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 397 us MUL_MAT_ID(ffn_moe_down-90): 243 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 428 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 26 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 405 us MUL_MAT_ID(ffn_moe_down-25): 380 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 368 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 402 us MUL_MAT_ID(ffn_moe_down-27): 215 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 395 us MUL_MAT_ID(ffn_moe_down-28): 188 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 430 us MUL_MAT_ID(ffn_moe_down-29): 181 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 394 us MUL_MAT_ID(ffn_moe_down-30): 353 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 423 us MUL_MAT_ID(ffn_moe_down-31): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 411 us MUL_MAT_ID(ffn_moe_down-32): 204 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 397 us MUL_MAT_ID(ffn_moe_down-33): 222 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 373 us MUL_MAT_ID(ffn_moe_down-34): 219 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 405 us MUL_MAT_ID(ffn_moe_down-35): 184 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 358 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 389 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 350 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 410 us MUL_MAT_ID(ffn_moe_down-40): 225 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 404 us MUL_MAT_ID(ffn_moe_down-41): 189 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 401 us MUL_MAT_ID(ffn_moe_down-42): 269 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 417 us MUL_MAT_ID(ffn_moe_down-43): 242 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 754 us MUL_MAT_ID(ffn_moe_down-44): 534 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 387 us MUL_MAT_ID(ffn_moe_down-45): 216 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 418 us MUL_MAT_ID(ffn_moe_down-46): 188 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 406 us MUL_MAT_ID(ffn_moe_down-47): 343 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 371 us MUL_MAT_ID(ffn_moe_down-48): 254 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 430 us MUL_MAT_ID(ffn_moe_down-49): 200 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 417 us MUL_MAT_ID(ffn_moe_down-50): 460 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 399 us MUL_MAT_ID(ffn_moe_down-51): 169 us ggml_barrier(...): 334 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 410 us MUL_MAT_ID(ffn_moe_down-52): 266 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 393 us MUL_MAT_ID(ffn_moe_down-53): 190 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 533 us MUL_MAT_ID(ffn_moe_down-54): 219 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 639 us MUL_MAT_ID(ffn_moe_down-55): 253 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 390 us MUL_MAT_ID(ffn_moe_down-56): 208 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 420 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 425 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 413 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 370 us MUL_MAT_ID(ffn_moe_down-60): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 372 us MUL_MAT_ID(ffn_moe_down-61): 403 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 355 us MUL_MAT_ID(ffn_moe_down-62): 368 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 411 us MUL_MAT_ID(ffn_moe_down-63): 364 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 355 us MUL_MAT_ID(ffn_moe_down-64): 270 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 410 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 394 us MUL_MAT_ID(ffn_moe_down-66): 409 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 412 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 409 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 355 us MUL_MAT_ID(ffn_moe_down-69): 195 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 390 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 414 us MUL_MAT_ID(ffn_moe_down-71): 220 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 384 us MUL_MAT_ID(ffn_moe_down-72): 215 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 377 us MUL_MAT_ID(ffn_moe_down-73): 219 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 413 us MUL_MAT_ID(ffn_moe_down-74): 223 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 400 us MUL_MAT_ID(ffn_moe_down-75): 418 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 438 us MUL_MAT_ID(ffn_moe_down-76): 558 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 403 us MUL_MAT_ID(ffn_moe_down-77): 422 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 383 us MUL_MAT_ID(ffn_moe_down-78): 353 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 418 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 358 us MUL_MAT_ID(ffn_moe_down-80): 191 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 373 us MUL_MAT_ID(ffn_moe_down-81): 207 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 434 us MUL_MAT_ID(ffn_moe_down-82): 184 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 363 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 390 us MUL_MAT_ID(ffn_moe_down-84): 242 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 419 us MUL_MAT_ID(ffn_moe_down-85): 645 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 389 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 352 us MUL_MAT_ID(ffn_moe_down-87): 383 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 193 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 350 us MUL_MAT_ID(ffn_moe_down-89): 372 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 362 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 421 us MUL_MAT_ID(ffn_moe_down-91): 247 us ggml_barrier(...): 6 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 290 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 378 us MUL_MAT_ID(ffn_moe_down-26): 220 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 423 us MUL_MAT_ID(ffn_moe_down-27): 438 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 420 us MUL_MAT_ID(ffn_moe_down-28): 209 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 447 us MUL_MAT_ID(ffn_moe_down-29): 216 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 391 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 389 us MUL_MAT_ID(ffn_moe_down-31): 210 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 361 us MUL_MAT_ID(ffn_moe_down-32): 218 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 402 us MUL_MAT_ID(ffn_moe_down-33): 400 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 387 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 420 us MUL_MAT_ID(ffn_moe_down-35): 189 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 413 us MUL_MAT_ID(ffn_moe_down-36): 392 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 393 us MUL_MAT_ID(ffn_moe_down-37): 241 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 393 us MUL_MAT_ID(ffn_moe_down-38): 184 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 404 us MUL_MAT_ID(ffn_moe_down-39): 243 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 358 us MUL_MAT_ID(ffn_moe_down-40): 393 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 382 us MUL_MAT_ID(ffn_moe_down-41): 215 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 381 us MUL_MAT_ID(ffn_moe_down-42): 193 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 398 us MUL_MAT_ID(ffn_moe_down-43): 186 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 387 us MUL_MAT_ID(ffn_moe_down-44): 219 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 414 us MUL_MAT_ID(ffn_moe_down-45): 211 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 427 us MUL_MAT_ID(ffn_moe_down-46): 221 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 525 us MUL_MAT_ID(ffn_moe_down-47): 223 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 480 us MUL_MAT_ID(ffn_moe_down-48): 354 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 404 us MUL_MAT_ID(ffn_moe_down-49): 206 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 404 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 413 us MUL_MAT_ID(ffn_moe_down-51): 209 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 409 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 451 us MUL_MAT_ID(ffn_moe_down-53): 196 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 482 us MUL_MAT_ID(ffn_moe_down-54): 288 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 493 us MUL_MAT_ID(ffn_moe_down-55): 448 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 765 us MUL_MAT_ID(ffn_moe_down-56): 315 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 518 us MUL_MAT_ID(ffn_moe_down-57): 189 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 434 us MUL_MAT_ID(ffn_moe_down-58): 341 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 378 us MUL_MAT_ID(ffn_moe_down-59): 233 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 389 us MUL_MAT_ID(ffn_moe_down-60): 222 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 436 us MUL_MAT_ID(ffn_moe_down-61): 213 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 439 us MUL_MAT_ID(ffn_moe_down-62): 245 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 434 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 426 us MUL_MAT_ID(ffn_moe_down-64): 196 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 400 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 465 us MUL_MAT_ID(ffn_moe_down-66): 326 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 627 us MUL_MAT_ID(ffn_moe_down-67): 428 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 443 us MUL_MAT_ID(ffn_moe_down-68): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 447 us MUL_MAT_ID(ffn_moe_down-69): 203 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 431 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 415 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 466 us MUL_MAT_ID(ffn_moe_down-72): 187 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 424 us MUL_MAT_ID(ffn_moe_down-73): 206 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 484 us MUL_MAT_ID(ffn_moe_down-74): 190 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 408 us MUL_MAT_ID(ffn_moe_down-75): 365 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 403 us MUL_MAT_ID(ffn_moe_down-76): 252 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 415 us MUL_MAT_ID(ffn_moe_down-77): 297 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 398 us MUL_MAT_ID(ffn_moe_down-78): 254 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 414 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 406 us MUL_MAT_ID(ffn_moe_down-80): 212 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 401 us MUL_MAT_ID(ffn_moe_down-81): 530 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 416 us MUL_MAT_ID(ffn_moe_down-82): 214 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 393 us MUL_MAT_ID(ffn_moe_down-83): 315 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 522 us MUL_MAT_ID(ffn_moe_down-84): 287 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 358 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 427 us MUL_MAT_ID(ffn_moe_down-86): 196 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 396 us MUL_MAT_ID(ffn_moe_down-87): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 414 us MUL_MAT_ID(ffn_moe_down-88): 206 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 370 us MUL_MAT_ID(ffn_moe_down-89): 306 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 395 us MUL_MAT_ID(ffn_moe_down-90): 245 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 457 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 11 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 206 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 406 us MUL_MAT_ID(ffn_moe_down-26): 198 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 349 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 365 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 384 us MUL_MAT_ID(ffn_moe_down-29): 217 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 423 us MUL_MAT_ID(ffn_moe_down-30): 372 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 359 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 405 us MUL_MAT_ID(ffn_moe_down-32): 209 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 369 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 425 us MUL_MAT_ID(ffn_moe_down-34): 202 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 447 us MUL_MAT_ID(ffn_moe_down-35): 471 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 364 us MUL_MAT_ID(ffn_moe_down-36): 544 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 366 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 439 us MUL_MAT_ID(ffn_moe_down-38): 200 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 366 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 392 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 521 us MUL_MAT_ID(ffn_moe_down-41): 164 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 420 us MUL_MAT_ID(ffn_moe_down-42): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 388 us MUL_MAT_ID(ffn_moe_down-43): 196 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 363 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 388 us MUL_MAT_ID(ffn_moe_down-45): 373 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 377 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 451 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 383 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 457 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 413 us MUL_MAT_ID(ffn_moe_down-50): 201 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 443 us MUL_MAT_ID(ffn_moe_down-51): 275 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 381 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 347 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 381 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 340 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 434 us MUL_MAT_ID(ffn_moe_down-56): 186 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 441 us MUL_MAT_ID(ffn_moe_down-57): 201 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 434 us MUL_MAT_ID(ffn_moe_down-58): 255 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 434 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 438 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 442 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 401 us MUL_MAT_ID(ffn_moe_down-62): 312 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 360 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 424 us MUL_MAT_ID(ffn_moe_down-64): 213 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 437 us MUL_MAT_ID(ffn_moe_down-65): 443 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 350 us MUL_MAT_ID(ffn_moe_down-66): 211 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 448 us MUL_MAT_ID(ffn_moe_down-67): 247 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 437 us MUL_MAT_ID(ffn_moe_down-68): 216 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 648 us MUL_MAT_ID(ffn_moe_down-69): 331 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 340 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 378 us MUL_MAT_ID(ffn_moe_down-71): 213 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 352 us MUL_MAT_ID(ffn_moe_down-72): 185 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 263 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 391 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 405 us MUL_MAT_ID(ffn_moe_down-75): 204 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 407 us MUL_MAT_ID(ffn_moe_down-76): 215 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 469 us MUL_MAT_ID(ffn_moe_down-77): 193 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 414 us MUL_MAT_ID(ffn_moe_down-78): 269 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 430 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 373 us MUL_MAT_ID(ffn_moe_down-80): 269 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 569 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 389 us MUL_MAT_ID(ffn_moe_down-82): 216 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 395 us MUL_MAT_ID(ffn_moe_down-83): 231 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 388 us MUL_MAT_ID(ffn_moe_down-84): 267 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 410 us MUL_MAT_ID(ffn_moe_down-85): 230 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 404 us MUL_MAT_ID(ffn_moe_down-86): 182 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 429 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 343 us MUL_MAT_ID(ffn_moe_down-88): 187 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 382 us MUL_MAT_ID(ffn_moe_down-89): 366 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 462 us MUL_MAT_ID(ffn_moe_down-90): 282 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 441 us MUL_MAT_ID(ffn_moe_down-91): 252 us ggml_barrier(...): 203 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 380 us MUL_MAT_ID(ffn_moe_down-25): 211 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 429 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 378 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 339 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 420 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 388 us MUL_MAT_ID(ffn_moe_down-30): 434 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 394 us MUL_MAT_ID(ffn_moe_down-31): 187 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 359 us MUL_MAT_ID(ffn_moe_down-32): 205 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 436 us MUL_MAT_ID(ffn_moe_down-33): 586 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 363 us MUL_MAT_ID(ffn_moe_down-34): 203 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 386 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 377 us MUL_MAT_ID(ffn_moe_down-36): 561 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 346 us MUL_MAT_ID(ffn_moe_down-37): 248 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 379 us MUL_MAT_ID(ffn_moe_down-38): 387 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 395 us MUL_MAT_ID(ffn_moe_down-39): 698 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 392 us MUL_MAT_ID(ffn_moe_down-40): 661 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 393 us MUL_MAT_ID(ffn_moe_down-41): 390 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 389 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 354 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 390 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 341 us MUL_MAT_ID(ffn_moe_down-45): 378 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 379 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 385 us MUL_MAT_ID(ffn_moe_down-47): 203 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 367 us MUL_MAT_ID(ffn_moe_down-48): 200 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 390 us MUL_MAT_ID(ffn_moe_down-49): 203 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 362 us MUL_MAT_ID(ffn_moe_down-50): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 356 us MUL_MAT_ID(ffn_moe_down-51): 524 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 397 us MUL_MAT_ID(ffn_moe_down-52): 187 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 419 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 372 us MUL_MAT_ID(ffn_moe_down-54): 498 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 373 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 406 us MUL_MAT_ID(ffn_moe_down-56): 526 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 397 us MUL_MAT_ID(ffn_moe_down-57): 448 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 401 us MUL_MAT_ID(ffn_moe_down-58): 207 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 416 us MUL_MAT_ID(ffn_moe_down-59): 409 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 381 us MUL_MAT_ID(ffn_moe_down-60): 387 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 401 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 415 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 396 us MUL_MAT_ID(ffn_moe_down-63): 201 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 341 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 378 us MUL_MAT_ID(ffn_moe_down-65): 193 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 399 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 401 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 422 us MUL_MAT_ID(ffn_moe_down-68): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 427 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 363 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 374 us MUL_MAT_ID(ffn_moe_down-71): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 450 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 377 us MUL_MAT_ID(ffn_moe_down-73): 204 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 389 us MUL_MAT_ID(ffn_moe_down-74): 301 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 465 us MUL_MAT_ID(ffn_moe_down-75): 189 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 414 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 433 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 392 us MUL_MAT_ID(ffn_moe_down-78): 245 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 400 us MUL_MAT_ID(ffn_moe_down-79): 182 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 424 us MUL_MAT_ID(ffn_moe_down-80): 221 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 411 us MUL_MAT_ID(ffn_moe_down-81): 221 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 657 us MUL_MAT_ID(ffn_moe_down-82): 510 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 376 us MUL_MAT_ID(ffn_moe_down-83): 306 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 395 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 410 us MUL_MAT_ID(ffn_moe_down-85): 256 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 346 us MUL_MAT_ID(ffn_moe_down-86): 251 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 448 us MUL_MAT_ID(ffn_moe_down-87): 204 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 374 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 418 us MUL_MAT_ID(ffn_moe_down-89): 227 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 416 us MUL_MAT_ID(ffn_moe_down-90): 284 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 383 us MUL_MAT_ID(ffn_moe_down-91): 245 us ggml_barrier(...): 28 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 440 us MUL_MAT_ID(ffn_moe_down-25): 196 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 424 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 454 us MUL_MAT_ID(ffn_moe_down-27): 211 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 437 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 431 us MUL_MAT_ID(ffn_moe_down-29): 191 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 462 us MUL_MAT_ID(ffn_moe_down-30): 263 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 390 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 396 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 334 us MUL_MAT_ID(ffn_moe_down-33): 236 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 451 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 352 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 445 us MUL_MAT_ID(ffn_moe_down-36): 253 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 379 us MUL_MAT_ID(ffn_moe_down-37): 623 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 367 us MUL_MAT_ID(ffn_moe_down-38): 331 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 382 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 395 us MUL_MAT_ID(ffn_moe_down-40): 481 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 387 us MUL_MAT_ID(ffn_moe_down-41): 383 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 391 us MUL_MAT_ID(ffn_moe_down-42): 354 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 389 us MUL_MAT_ID(ffn_moe_down-43): 553 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 371 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 407 us MUL_MAT_ID(ffn_moe_down-45): 210 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 417 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 419 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 421 us MUL_MAT_ID(ffn_moe_down-48): 190 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 354 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 383 us MUL_MAT_ID(ffn_moe_down-50): 216 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 393 us MUL_MAT_ID(ffn_moe_down-51): 204 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 376 us MUL_MAT_ID(ffn_moe_down-52): 215 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 388 us MUL_MAT_ID(ffn_moe_down-53): 389 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 423 us MUL_MAT_ID(ffn_moe_down-54): 184 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 350 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 423 us MUL_MAT_ID(ffn_moe_down-56): 184 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 402 us MUL_MAT_ID(ffn_moe_down-57): 220 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 409 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 427 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 347 us MUL_MAT_ID(ffn_moe_down-60): 189 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 392 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 358 us MUL_MAT_ID(ffn_moe_down-62): 583 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 365 us MUL_MAT_ID(ffn_moe_down-63): 467 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 535 us MUL_MAT_ID(ffn_moe_down-64): 448 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 659 us MUL_MAT_ID(ffn_moe_down-65): 319 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 357 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 399 us MUL_MAT_ID(ffn_moe_down-67): 192 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 451 us MUL_MAT_ID(ffn_moe_down-68): 189 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 372 us MUL_MAT_ID(ffn_moe_down-69): 459 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 389 us MUL_MAT_ID(ffn_moe_down-70): 209 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 415 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 514 us MUL_MAT_ID(ffn_moe_down-72): 162 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 398 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 335 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 379 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 411 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 402 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 377 us MUL_MAT_ID(ffn_moe_down-78): 215 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 404 us MUL_MAT_ID(ffn_moe_down-79): 203 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 337 us MUL_MAT_ID(ffn_moe_down-80): 285 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 386 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 402 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 372 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 403 us MUL_MAT_ID(ffn_moe_down-84): 298 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 416 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 363 us MUL_MAT_ID(ffn_moe_down-86): 239 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 438 us MUL_MAT_ID(ffn_moe_down-87): 200 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 397 us MUL_MAT_ID(ffn_moe_down-88): 407 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 387 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 362 us MUL_MAT_ID(ffn_moe_down-90): 220 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 390 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 90 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 385 us MUL_MAT_ID(ffn_moe_down-25): 186 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 383 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 350 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 347 us MUL_MAT_ID(ffn_moe_down-28): 186 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 425 us MUL_MAT_ID(ffn_moe_down-29): 194 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 342 us MUL_MAT_ID(ffn_moe_down-30): 181 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 404 us MUL_MAT_ID(ffn_moe_down-31): 210 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 441 us MUL_MAT_ID(ffn_moe_down-32): 184 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 351 us MUL_MAT_ID(ffn_moe_down-33): 211 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 334 us MUL_MAT_ID(ffn_moe_down-34): 359 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 385 us MUL_MAT_ID(ffn_moe_down-35): 183 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 392 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 389 us MUL_MAT_ID(ffn_moe_down-37): 247 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 357 us MUL_MAT_ID(ffn_moe_down-38): 209 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 377 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 451 us MUL_MAT_ID(ffn_moe_down-40): 672 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 405 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 403 us MUL_MAT_ID(ffn_moe_down-42): 180 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 441 us MUL_MAT_ID(ffn_moe_down-43): 204 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 427 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 432 us MUL_MAT_ID(ffn_moe_down-45): 221 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 431 us MUL_MAT_ID(ffn_moe_down-46): 213 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 473 us MUL_MAT_ID(ffn_moe_down-47): 420 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 420 us MUL_MAT_ID(ffn_moe_down-48): 415 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 395 us MUL_MAT_ID(ffn_moe_down-49): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 408 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 400 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 432 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 388 us MUL_MAT_ID(ffn_moe_down-53): 185 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 512 us MUL_MAT_ID(ffn_moe_down-54): 245 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 427 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 453 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 411 us MUL_MAT_ID(ffn_moe_down-57): 282 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 409 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 435 us MUL_MAT_ID(ffn_moe_down-59): 210 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 435 us MUL_MAT_ID(ffn_moe_down-60): 388 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 398 us MUL_MAT_ID(ffn_moe_down-61): 251 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 452 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 390 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 428 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 391 us MUL_MAT_ID(ffn_moe_down-65): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 412 us MUL_MAT_ID(ffn_moe_down-66): 242 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 407 us MUL_MAT_ID(ffn_moe_down-67): 209 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 398 us MUL_MAT_ID(ffn_moe_down-68): 200 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 395 us MUL_MAT_ID(ffn_moe_down-69): 197 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 540 us MUL_MAT_ID(ffn_moe_down-70): 181 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 381 us MUL_MAT_ID(ffn_moe_down-71): 208 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 413 us MUL_MAT_ID(ffn_moe_down-72): 213 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 384 us MUL_MAT_ID(ffn_moe_down-73): 214 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 401 us MUL_MAT_ID(ffn_moe_down-74): 224 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 401 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 382 us MUL_MAT_ID(ffn_moe_down-76): 283 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 416 us MUL_MAT_ID(ffn_moe_down-77): 214 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 448 us MUL_MAT_ID(ffn_moe_down-78): 275 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 424 us MUL_MAT_ID(ffn_moe_down-79): 268 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 395 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 359 us MUL_MAT_ID(ffn_moe_down-81): 538 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 390 us MUL_MAT_ID(ffn_moe_down-82): 219 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 424 us MUL_MAT_ID(ffn_moe_down-83): 567 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 503 us MUL_MAT_ID(ffn_moe_down-84): 295 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 397 us MUL_MAT_ID(ffn_moe_down-85): 234 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 411 us MUL_MAT_ID(ffn_moe_down-86): 399 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 392 us MUL_MAT_ID(ffn_moe_down-87): 201 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 391 us MUL_MAT_ID(ffn_moe_down-88): 338 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 359 us MUL_MAT_ID(ffn_moe_down-89): 401 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 392 us MUL_MAT_ID(ffn_moe_down-90): 553 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 376 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 141 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 522 us MUL_MAT_ID(ffn_moe_down-25): 196 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 400 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 374 us MUL_MAT_ID(ffn_moe_down-27): 211 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 353 us MUL_MAT_ID(ffn_moe_down-28): 227 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 393 us MUL_MAT_ID(ffn_moe_down-29): 180 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 455 us MUL_MAT_ID(ffn_moe_down-30): 311 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 523 us MUL_MAT_ID(ffn_moe_down-31): 175 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 416 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 433 us MUL_MAT_ID(ffn_moe_down-33): 210 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 397 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 360 us MUL_MAT_ID(ffn_moe_down-35): 211 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 346 us MUL_MAT_ID(ffn_moe_down-36): 224 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 481 us MUL_MAT_ID(ffn_moe_down-37): 278 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 457 us MUL_MAT_ID(ffn_moe_down-38): 275 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 406 us MUL_MAT_ID(ffn_moe_down-39): 417 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 376 us MUL_MAT_ID(ffn_moe_down-40): 257 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 421 us MUL_MAT_ID(ffn_moe_down-41): 189 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 428 us MUL_MAT_ID(ffn_moe_down-42): 340 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 399 us MUL_MAT_ID(ffn_moe_down-43): 174 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 412 us MUL_MAT_ID(ffn_moe_down-44): 198 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 453 us MUL_MAT_ID(ffn_moe_down-45): 224 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 468 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 392 us MUL_MAT_ID(ffn_moe_down-47): 299 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 409 us MUL_MAT_ID(ffn_moe_down-48): 183 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 465 us MUL_MAT_ID(ffn_moe_down-49): 275 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 417 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 522 us MUL_MAT_ID(ffn_moe_down-51): 220 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 484 us MUL_MAT_ID(ffn_moe_down-52): 648 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 395 us MUL_MAT_ID(ffn_moe_down-53): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 366 us MUL_MAT_ID(ffn_moe_down-54): 433 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 378 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 474 us MUL_MAT_ID(ffn_moe_down-56): 515 us ggml_barrier(...): 343 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 455 us MUL_MAT_ID(ffn_moe_down-57): 196 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 472 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 428 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 438 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 356 us MUL_MAT_ID(ffn_moe_down-61): 187 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 429 us MUL_MAT_ID(ffn_moe_down-62): 215 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 412 us MUL_MAT_ID(ffn_moe_down-63): 385 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 406 us MUL_MAT_ID(ffn_moe_down-64): 208 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 414 us MUL_MAT_ID(ffn_moe_down-65): 193 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 431 us MUL_MAT_ID(ffn_moe_down-66): 197 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 456 us MUL_MAT_ID(ffn_moe_down-67): 270 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 415 us MUL_MAT_ID(ffn_moe_down-68): 212 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 387 us MUL_MAT_ID(ffn_moe_down-69): 197 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 393 us MUL_MAT_ID(ffn_moe_down-70): 185 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 435 us MUL_MAT_ID(ffn_moe_down-71): 259 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 376 us MUL_MAT_ID(ffn_moe_down-72): 185 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 415 us MUL_MAT_ID(ffn_moe_down-73): 242 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 420 us MUL_MAT_ID(ffn_moe_down-74): 221 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 407 us MUL_MAT_ID(ffn_moe_down-75): 180 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 478 us MUL_MAT_ID(ffn_moe_down-76): 198 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 389 us MUL_MAT_ID(ffn_moe_down-77): 189 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 425 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 365 us MUL_MAT_ID(ffn_moe_down-79): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 397 us MUL_MAT_ID(ffn_moe_down-80): 269 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 364 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 415 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 546 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 409 us MUL_MAT_ID(ffn_moe_down-84): 243 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 361 us MUL_MAT_ID(ffn_moe_down-85): 628 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 371 us MUL_MAT_ID(ffn_moe_down-86): 216 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 419 us MUL_MAT_ID(ffn_moe_down-87): 213 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 426 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 361 us MUL_MAT_ID(ffn_moe_down-89): 532 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 371 us MUL_MAT_ID(ffn_moe_down-90): 237 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 431 us MUL_MAT_ID(ffn_moe_down-91): 257 us ggml_barrier(...): 11 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 506 us MUL_MAT_ID(ffn_moe_down-25): 327 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 425 us MUL_MAT_ID(ffn_moe_down-26): 200 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 409 us MUL_MAT_ID(ffn_moe_down-27): 188 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 421 us MUL_MAT_ID(ffn_moe_down-28): 247 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 426 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 407 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 366 us MUL_MAT_ID(ffn_moe_down-31): 399 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 365 us MUL_MAT_ID(ffn_moe_down-32): 203 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 416 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 340 us MUL_MAT_ID(ffn_moe_down-34): 186 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 404 us MUL_MAT_ID(ffn_moe_down-35): 218 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 397 us MUL_MAT_ID(ffn_moe_down-36): 334 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 402 us MUL_MAT_ID(ffn_moe_down-37): 290 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 545 us MUL_MAT_ID(ffn_moe_down-38): 213 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 403 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 381 us MUL_MAT_ID(ffn_moe_down-40): 614 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 368 us MUL_MAT_ID(ffn_moe_down-41): 204 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 359 us MUL_MAT_ID(ffn_moe_down-42): 558 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 409 us MUL_MAT_ID(ffn_moe_down-43): 211 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 396 us MUL_MAT_ID(ffn_moe_down-44): 184 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 505 us MUL_MAT_ID(ffn_moe_down-45): 320 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 421 us MUL_MAT_ID(ffn_moe_down-46): 207 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 417 us MUL_MAT_ID(ffn_moe_down-47): 204 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 411 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 353 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 349 us MUL_MAT_ID(ffn_moe_down-50): 254 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 405 us MUL_MAT_ID(ffn_moe_down-51): 236 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 449 us MUL_MAT_ID(ffn_moe_down-52): 204 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 409 us MUL_MAT_ID(ffn_moe_down-53): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 431 us MUL_MAT_ID(ffn_moe_down-54): 204 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 365 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 448 us MUL_MAT_ID(ffn_moe_down-56): 203 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 447 us MUL_MAT_ID(ffn_moe_down-57): 567 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 455 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 428 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 353 us MUL_MAT_ID(ffn_moe_down-60): 234 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 396 us MUL_MAT_ID(ffn_moe_down-61): 203 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 381 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 551 us MUL_MAT_ID(ffn_moe_down-63): 222 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 399 us MUL_MAT_ID(ffn_moe_down-64): 594 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 367 us MUL_MAT_ID(ffn_moe_down-65): 172 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 388 us MUL_MAT_ID(ffn_moe_down-66): 434 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 424 us MUL_MAT_ID(ffn_moe_down-67): 219 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 405 us MUL_MAT_ID(ffn_moe_down-68): 205 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 380 us MUL_MAT_ID(ffn_moe_down-69): 313 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 401 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 404 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 417 us MUL_MAT_ID(ffn_moe_down-72): 224 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 366 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 354 us MUL_MAT_ID(ffn_moe_down-74): 285 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 451 us MUL_MAT_ID(ffn_moe_down-75): 213 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 416 us MUL_MAT_ID(ffn_moe_down-76): 185 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 391 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 415 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 372 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 410 us MUL_MAT_ID(ffn_moe_down-80): 209 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 413 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 430 us MUL_MAT_ID(ffn_moe_down-82): 183 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 418 us MUL_MAT_ID(ffn_moe_down-83): 225 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 441 us MUL_MAT_ID(ffn_moe_down-84): 255 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 414 us MUL_MAT_ID(ffn_moe_down-85): 695 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 388 us MUL_MAT_ID(ffn_moe_down-86): 208 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 386 us MUL_MAT_ID(ffn_moe_down-87): 216 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 427 us MUL_MAT_ID(ffn_moe_down-88): 181 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 422 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 378 us MUL_MAT_ID(ffn_moe_down-90): 230 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 435 us MUL_MAT_ID(ffn_moe_down-91): 244 us ggml_barrier(...): 38 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 364 us MUL_MAT_ID(ffn_moe_down-25): 198 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 389 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 384 us MUL_MAT_ID(ffn_moe_down-27): 291 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 454 us MUL_MAT_ID(ffn_moe_down-28): 187 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 364 us MUL_MAT_ID(ffn_moe_down-29): 214 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 431 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 430 us MUL_MAT_ID(ffn_moe_down-31): 364 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 397 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 395 us MUL_MAT_ID(ffn_moe_down-33): 202 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 406 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 387 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 400 us MUL_MAT_ID(ffn_moe_down-36): 222 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 391 us MUL_MAT_ID(ffn_moe_down-37): 224 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 404 us MUL_MAT_ID(ffn_moe_down-38): 192 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 330 us MUL_MAT_ID(ffn_moe_down-39): 734 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 438 us MUL_MAT_ID(ffn_moe_down-40): 239 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 670 us MUL_MAT_ID(ffn_moe_down-41): 309 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 361 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 398 us MUL_MAT_ID(ffn_moe_down-43): 199 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 379 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 401 us MUL_MAT_ID(ffn_moe_down-45): 242 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 435 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 439 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 358 us MUL_MAT_ID(ffn_moe_down-48): 196 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 211 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 428 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 399 us MUL_MAT_ID(ffn_moe_down-51): 201 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 466 us MUL_MAT_ID(ffn_moe_down-52): 602 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 652 us MUL_MAT_ID(ffn_moe_down-53): 228 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 390 us MUL_MAT_ID(ffn_moe_down-54): 204 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 401 us MUL_MAT_ID(ffn_moe_down-55): 202 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 391 us MUL_MAT_ID(ffn_moe_down-56): 186 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 360 us MUL_MAT_ID(ffn_moe_down-57): 222 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 423 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 406 us MUL_MAT_ID(ffn_moe_down-59): 210 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 350 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 387 us MUL_MAT_ID(ffn_moe_down-61): 217 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 443 us MUL_MAT_ID(ffn_moe_down-62): 201 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 376 us MUL_MAT_ID(ffn_moe_down-63): 201 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 353 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 384 us MUL_MAT_ID(ffn_moe_down-65): 199 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 448 us MUL_MAT_ID(ffn_moe_down-66): 401 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 407 us MUL_MAT_ID(ffn_moe_down-67): 307 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 458 us MUL_MAT_ID(ffn_moe_down-68): 167 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 392 us MUL_MAT_ID(ffn_moe_down-69): 189 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 358 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 394 us MUL_MAT_ID(ffn_moe_down-71): 207 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 345 us MUL_MAT_ID(ffn_moe_down-72): 220 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 418 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 361 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 427 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 358 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 425 us MUL_MAT_ID(ffn_moe_down-77): 206 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 411 us MUL_MAT_ID(ffn_moe_down-78): 218 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 404 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 421 us MUL_MAT_ID(ffn_moe_down-80): 187 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 415 us MUL_MAT_ID(ffn_moe_down-81): 210 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 380 us MUL_MAT_ID(ffn_moe_down-82): 181 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 409 us MUL_MAT_ID(ffn_moe_down-83): 189 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 380 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 403 us MUL_MAT_ID(ffn_moe_down-85): 241 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 509 us MUL_MAT_ID(ffn_moe_down-86): 258 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 463 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 390 us MUL_MAT_ID(ffn_moe_down-88): 193 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 365 us MUL_MAT_ID(ffn_moe_down-89): 280 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 369 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 363 us MUL_MAT_ID(ffn_moe_down-91): 242 us ggml_barrier(...): 55 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 359 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 444 us MUL_MAT_ID(ffn_moe_down-26): 348 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 426 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 359 us MUL_MAT_ID(ffn_moe_down-28): 200 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 416 us MUL_MAT_ID(ffn_moe_down-29): 203 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 488 us MUL_MAT_ID(ffn_moe_down-30): 175 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 387 us MUL_MAT_ID(ffn_moe_down-31): 176 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 389 us MUL_MAT_ID(ffn_moe_down-32): 220 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 374 us MUL_MAT_ID(ffn_moe_down-33): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 396 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 376 us MUL_MAT_ID(ffn_moe_down-35): 209 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 338 us MUL_MAT_ID(ffn_moe_down-36): 202 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 387 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 395 us MUL_MAT_ID(ffn_moe_down-38): 196 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 419 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 403 us MUL_MAT_ID(ffn_moe_down-40): 228 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 405 us MUL_MAT_ID(ffn_moe_down-41): 197 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 381 us MUL_MAT_ID(ffn_moe_down-42): 556 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 414 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 511 us MUL_MAT_ID(ffn_moe_down-44): 184 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 381 us MUL_MAT_ID(ffn_moe_down-45): 399 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 403 us MUL_MAT_ID(ffn_moe_down-46): 198 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 375 us MUL_MAT_ID(ffn_moe_down-47): 197 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 419 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 393 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 417 us MUL_MAT_ID(ffn_moe_down-51): 188 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 406 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 342 us MUL_MAT_ID(ffn_moe_down-53): 638 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 430 us MUL_MAT_ID(ffn_moe_down-54): 228 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 654 us MUL_MAT_ID(ffn_moe_down-55): 328 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 388 us MUL_MAT_ID(ffn_moe_down-56): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 403 us MUL_MAT_ID(ffn_moe_down-57): 212 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 435 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 404 us MUL_MAT_ID(ffn_moe_down-59): 255 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 350 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 426 us MUL_MAT_ID(ffn_moe_down-61): 277 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 444 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 388 us MUL_MAT_ID(ffn_moe_down-63): 201 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 360 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 537 us MUL_MAT_ID(ffn_moe_down-65): 183 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 501 us MUL_MAT_ID(ffn_moe_down-66): 333 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 431 us MUL_MAT_ID(ffn_moe_down-67): 468 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 415 us MUL_MAT_ID(ffn_moe_down-68): 343 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 412 us MUL_MAT_ID(ffn_moe_down-69): 280 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 377 us MUL_MAT_ID(ffn_moe_down-70): 219 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 369 us MUL_MAT_ID(ffn_moe_down-71): 369 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 422 us MUL_MAT_ID(ffn_moe_down-72): 204 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 426 us MUL_MAT_ID(ffn_moe_down-73): 282 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 445 us MUL_MAT_ID(ffn_moe_down-74): 233 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 429 us MUL_MAT_ID(ffn_moe_down-75): 401 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 390 us MUL_MAT_ID(ffn_moe_down-76): 345 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 387 us MUL_MAT_ID(ffn_moe_down-77): 200 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 449 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 406 us MUL_MAT_ID(ffn_moe_down-79): 219 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 440 us MUL_MAT_ID(ffn_moe_down-80): 207 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 429 us MUL_MAT_ID(ffn_moe_down-81): 187 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 426 us MUL_MAT_ID(ffn_moe_down-82): 338 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 447 us MUL_MAT_ID(ffn_moe_down-83): 309 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 412 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 403 us MUL_MAT_ID(ffn_moe_down-85): 226 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 407 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 414 us MUL_MAT_ID(ffn_moe_down-87): 343 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 460 us MUL_MAT_ID(ffn_moe_down-88): 213 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 410 us MUL_MAT_ID(ffn_moe_down-89): 209 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 478 us MUL_MAT_ID(ffn_moe_down-90): 398 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 518 us MUL_MAT_ID(ffn_moe_down-91): 752 us ggml_barrier(...): 139 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 425 us MUL_MAT_ID(ffn_moe_down-25): 212 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 395 us MUL_MAT_ID(ffn_moe_down-26): 210 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 405 us MUL_MAT_ID(ffn_moe_down-27): 207 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 477 us MUL_MAT_ID(ffn_moe_down-28): 242 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 418 us MUL_MAT_ID(ffn_moe_down-29): 219 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 429 us MUL_MAT_ID(ffn_moe_down-30): 375 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 406 us MUL_MAT_ID(ffn_moe_down-31): 202 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 399 us MUL_MAT_ID(ffn_moe_down-32): 222 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 402 us MUL_MAT_ID(ffn_moe_down-33): 214 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 443 us MUL_MAT_ID(ffn_moe_down-34): 226 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 406 us MUL_MAT_ID(ffn_moe_down-35): 576 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 395 us MUL_MAT_ID(ffn_moe_down-36): 214 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 417 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 392 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 403 us MUL_MAT_ID(ffn_moe_down-39): 242 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 391 us MUL_MAT_ID(ffn_moe_down-40): 401 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 430 us MUL_MAT_ID(ffn_moe_down-41): 209 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 412 us MUL_MAT_ID(ffn_moe_down-42): 203 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 401 us MUL_MAT_ID(ffn_moe_down-43): 332 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 395 us MUL_MAT_ID(ffn_moe_down-44): 202 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 399 us MUL_MAT_ID(ffn_moe_down-45): 589 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 406 us MUL_MAT_ID(ffn_moe_down-46): 203 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 390 us MUL_MAT_ID(ffn_moe_down-47): 206 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 400 us MUL_MAT_ID(ffn_moe_down-48): 206 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 401 us MUL_MAT_ID(ffn_moe_down-49): 208 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 395 us MUL_MAT_ID(ffn_moe_down-50): 205 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 398 us MUL_MAT_ID(ffn_moe_down-51): 205 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 410 us MUL_MAT_ID(ffn_moe_down-52): 216 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 415 us MUL_MAT_ID(ffn_moe_down-53): 210 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 412 us MUL_MAT_ID(ffn_moe_down-54): 203 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 404 us MUL_MAT_ID(ffn_moe_down-55): 207 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 455 us MUL_MAT_ID(ffn_moe_down-56): 276 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 398 us MUL_MAT_ID(ffn_moe_down-57): 214 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 414 us MUL_MAT_ID(ffn_moe_down-58): 211 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 402 us MUL_MAT_ID(ffn_moe_down-59): 215 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 208 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 406 us MUL_MAT_ID(ffn_moe_down-61): 209 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 538 us MUL_MAT_ID(ffn_moe_down-62): 230 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 405 us MUL_MAT_ID(ffn_moe_down-63): 214 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 446 us MUL_MAT_ID(ffn_moe_down-64): 245 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 433 us MUL_MAT_ID(ffn_moe_down-65): 230 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 395 us MUL_MAT_ID(ffn_moe_down-66): 217 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 391 us MUL_MAT_ID(ffn_moe_down-67): 217 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 436 us MUL_MAT_ID(ffn_moe_down-68): 441 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 407 us MUL_MAT_ID(ffn_moe_down-69): 205 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 421 us MUL_MAT_ID(ffn_moe_down-70): 445 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 406 us MUL_MAT_ID(ffn_moe_down-71): 222 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 386 us MUL_MAT_ID(ffn_moe_down-72): 219 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 401 us MUL_MAT_ID(ffn_moe_down-73): 210 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 463 us MUL_MAT_ID(ffn_moe_down-74): 304 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 404 us MUL_MAT_ID(ffn_moe_down-75): 217 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 397 us MUL_MAT_ID(ffn_moe_down-76): 224 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 430 us MUL_MAT_ID(ffn_moe_down-77): 234 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 422 us MUL_MAT_ID(ffn_moe_down-78): 231 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 426 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 437 us MUL_MAT_ID(ffn_moe_down-80): 233 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 405 us MUL_MAT_ID(ffn_moe_down-81): 211 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 420 us MUL_MAT_ID(ffn_moe_down-82): 255 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 453 us MUL_MAT_ID(ffn_moe_down-83): 231 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 411 us MUL_MAT_ID(ffn_moe_down-84): 243 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 435 us MUL_MAT_ID(ffn_moe_down-85): 257 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 396 us MUL_MAT_ID(ffn_moe_down-86): 205 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 432 us MUL_MAT_ID(ffn_moe_down-87): 214 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 407 us MUL_MAT_ID(ffn_moe_down-88): 232 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 417 us MUL_MAT_ID(ffn_moe_down-89): 425 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 423 us MUL_MAT_ID(ffn_moe_down-90): 230 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 386 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 15 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 380 us MUL_MAT_ID(ffn_moe_down-25): 214 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 410 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 472 us MUL_MAT_ID(ffn_moe_down-27): 232 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 409 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 392 us MUL_MAT_ID(ffn_moe_down-29): 203 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 408 us MUL_MAT_ID(ffn_moe_down-30): 227 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 436 us MUL_MAT_ID(ffn_moe_down-31): 417 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 399 us MUL_MAT_ID(ffn_moe_down-32): 206 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 495 us MUL_MAT_ID(ffn_moe_down-33): 300 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 393 us MUL_MAT_ID(ffn_moe_down-34): 275 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 392 us MUL_MAT_ID(ffn_moe_down-35): 215 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 404 us MUL_MAT_ID(ffn_moe_down-36): 218 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 415 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 386 us MUL_MAT_ID(ffn_moe_down-38): 215 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 371 us MUL_MAT_ID(ffn_moe_down-39): 244 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 379 us MUL_MAT_ID(ffn_moe_down-40): 456 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 390 us MUL_MAT_ID(ffn_moe_down-41): 420 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 428 us MUL_MAT_ID(ffn_moe_down-42): 223 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 422 us MUL_MAT_ID(ffn_moe_down-43): 219 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 386 us MUL_MAT_ID(ffn_moe_down-44): 204 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 486 us MUL_MAT_ID(ffn_moe_down-45): 299 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 394 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 385 us MUL_MAT_ID(ffn_moe_down-47): 211 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 396 us MUL_MAT_ID(ffn_moe_down-48): 204 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 395 us MUL_MAT_ID(ffn_moe_down-49): 464 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 439 us MUL_MAT_ID(ffn_moe_down-50): 217 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 403 us MUL_MAT_ID(ffn_moe_down-51): 233 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 435 us MUL_MAT_ID(ffn_moe_down-52): 334 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 434 us MUL_MAT_ID(ffn_moe_down-53): 235 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 393 us MUL_MAT_ID(ffn_moe_down-54): 206 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 395 us MUL_MAT_ID(ffn_moe_down-55): 220 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 578 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 424 us MUL_MAT_ID(ffn_moe_down-57): 227 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 410 us MUL_MAT_ID(ffn_moe_down-58): 210 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 390 us MUL_MAT_ID(ffn_moe_down-59): 212 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 381 us MUL_MAT_ID(ffn_moe_down-60): 219 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 431 us MUL_MAT_ID(ffn_moe_down-61): 225 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 382 us MUL_MAT_ID(ffn_moe_down-62): 209 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 453 us MUL_MAT_ID(ffn_moe_down-63): 475 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 497 us MUL_MAT_ID(ffn_moe_down-64): 200 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 411 us MUL_MAT_ID(ffn_moe_down-65): 391 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 404 us MUL_MAT_ID(ffn_moe_down-66): 210 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 477 us MUL_MAT_ID(ffn_moe_down-67): 438 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 404 us MUL_MAT_ID(ffn_moe_down-68): 202 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 409 us MUL_MAT_ID(ffn_moe_down-69): 225 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 392 us MUL_MAT_ID(ffn_moe_down-70): 231 us ggml_barrier(...): 357 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 405 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 425 us MUL_MAT_ID(ffn_moe_down-72): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 426 us MUL_MAT_ID(ffn_moe_down-73): 202 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 419 us MUL_MAT_ID(ffn_moe_down-74): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 398 us MUL_MAT_ID(ffn_moe_down-75): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 457 us MUL_MAT_ID(ffn_moe_down-76): 198 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 415 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 391 us MUL_MAT_ID(ffn_moe_down-78): 191 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 392 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 392 us MUL_MAT_ID(ffn_moe_down-80): 207 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 375 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 388 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 409 us MUL_MAT_ID(ffn_moe_down-83): 582 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 510 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 406 us MUL_MAT_ID(ffn_moe_down-85): 345 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 396 us MUL_MAT_ID(ffn_moe_down-86): 222 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 441 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 445 us MUL_MAT_ID(ffn_moe_down-88): 209 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 382 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 414 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 418 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 218 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 396 us MUL_MAT_ID(ffn_moe_down-25): 211 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 367 us MUL_MAT_ID(ffn_moe_down-26): 186 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 424 us MUL_MAT_ID(ffn_moe_down-27): 229 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 421 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 422 us MUL_MAT_ID(ffn_moe_down-29): 181 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 382 us MUL_MAT_ID(ffn_moe_down-30): 198 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 408 us MUL_MAT_ID(ffn_moe_down-31): 413 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 354 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 328 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 423 us MUL_MAT_ID(ffn_moe_down-34): 217 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 442 us MUL_MAT_ID(ffn_moe_down-35): 183 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 424 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 418 us MUL_MAT_ID(ffn_moe_down-37): 274 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 414 us MUL_MAT_ID(ffn_moe_down-38): 435 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 397 us MUL_MAT_ID(ffn_moe_down-39): 250 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 400 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 364 us MUL_MAT_ID(ffn_moe_down-41): 357 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 415 us MUL_MAT_ID(ffn_moe_down-42): 184 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 410 us MUL_MAT_ID(ffn_moe_down-43): 199 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 391 us MUL_MAT_ID(ffn_moe_down-44): 261 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 456 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 433 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 416 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 425 us MUL_MAT_ID(ffn_moe_down-48): 212 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 380 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 377 us MUL_MAT_ID(ffn_moe_down-50): 258 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 427 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 431 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 449 us MUL_MAT_ID(ffn_moe_down-53): 371 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 401 us MUL_MAT_ID(ffn_moe_down-54): 195 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 391 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 428 us MUL_MAT_ID(ffn_moe_down-56): 201 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 434 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 443 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 347 us MUL_MAT_ID(ffn_moe_down-59): 212 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 437 us MUL_MAT_ID(ffn_moe_down-60): 462 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 370 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 435 us MUL_MAT_ID(ffn_moe_down-62): 207 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 424 us MUL_MAT_ID(ffn_moe_down-63): 216 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 392 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 444 us MUL_MAT_ID(ffn_moe_down-65): 209 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 381 us MUL_MAT_ID(ffn_moe_down-66): 207 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 427 us MUL_MAT_ID(ffn_moe_down-67): 208 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 458 us MUL_MAT_ID(ffn_moe_down-68): 218 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 439 us MUL_MAT_ID(ffn_moe_down-69): 201 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 361 us MUL_MAT_ID(ffn_moe_down-70): 230 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 444 us MUL_MAT_ID(ffn_moe_down-71): 236 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 464 us MUL_MAT_ID(ffn_moe_down-72): 210 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 438 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 420 us MUL_MAT_ID(ffn_moe_down-74): 402 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 381 us MUL_MAT_ID(ffn_moe_down-75): 190 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 434 us MUL_MAT_ID(ffn_moe_down-76): 209 us ggml_barrier(...): 363 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 405 us MUL_MAT_ID(ffn_moe_down-77): 200 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 381 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 413 us MUL_MAT_ID(ffn_moe_down-79): 418 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 430 us MUL_MAT_ID(ffn_moe_down-80): 354 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 386 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 403 us MUL_MAT_ID(ffn_moe_down-82): 231 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 417 us MUL_MAT_ID(ffn_moe_down-83): 467 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 420 us MUL_MAT_ID(ffn_moe_down-84): 374 us ggml_barrier(...): 484 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 403 us MUL_MAT_ID(ffn_moe_down-85): 244 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 444 us MUL_MAT_ID(ffn_moe_down-86): 207 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 400 us MUL_MAT_ID(ffn_moe_down-87): 184 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 388 us MUL_MAT_ID(ffn_moe_down-88): 203 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 384 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 446 us MUL_MAT_ID(ffn_moe_down-90): 229 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 427 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 7 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 403 us MUL_MAT_ID(ffn_moe_down-25): 210 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 433 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 343 us MUL_MAT_ID(ffn_moe_down-27): 314 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 409 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 421 us MUL_MAT_ID(ffn_moe_down-29): 343 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 443 us MUL_MAT_ID(ffn_moe_down-30): 540 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 407 us MUL_MAT_ID(ffn_moe_down-31): 186 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 417 us MUL_MAT_ID(ffn_moe_down-32): 204 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 405 us MUL_MAT_ID(ffn_moe_down-33): 274 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 455 us MUL_MAT_ID(ffn_moe_down-34): 304 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 446 us MUL_MAT_ID(ffn_moe_down-35): 197 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 333 us MUL_MAT_ID(ffn_moe_down-36): 202 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 400 us MUL_MAT_ID(ffn_moe_down-37): 256 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 384 us MUL_MAT_ID(ffn_moe_down-38): 353 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 390 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 370 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 450 us MUL_MAT_ID(ffn_moe_down-41): 225 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 398 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 409 us MUL_MAT_ID(ffn_moe_down-43): 402 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 409 us MUL_MAT_ID(ffn_moe_down-44): 210 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 397 us MUL_MAT_ID(ffn_moe_down-45): 192 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 432 us MUL_MAT_ID(ffn_moe_down-46): 186 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 390 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 455 us MUL_MAT_ID(ffn_moe_down-48): 302 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 412 us MUL_MAT_ID(ffn_moe_down-49): 199 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 402 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 438 us MUL_MAT_ID(ffn_moe_down-51): 436 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 390 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 464 us MUL_MAT_ID(ffn_moe_down-53): 225 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 403 us MUL_MAT_ID(ffn_moe_down-54): 204 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 428 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 375 us MUL_MAT_ID(ffn_moe_down-56): 212 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 468 us MUL_MAT_ID(ffn_moe_down-57): 214 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 426 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 424 us MUL_MAT_ID(ffn_moe_down-59): 219 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 423 us MUL_MAT_ID(ffn_moe_down-60): 207 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 451 us MUL_MAT_ID(ffn_moe_down-61): 206 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 426 us MUL_MAT_ID(ffn_moe_down-62): 554 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 401 us MUL_MAT_ID(ffn_moe_down-63): 316 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 463 us MUL_MAT_ID(ffn_moe_down-64): 551 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 407 us MUL_MAT_ID(ffn_moe_down-65): 396 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 470 us MUL_MAT_ID(ffn_moe_down-66): 338 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 434 us MUL_MAT_ID(ffn_moe_down-67): 200 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 337 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 394 us MUL_MAT_ID(ffn_moe_down-69): 373 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 388 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 414 us MUL_MAT_ID(ffn_moe_down-71): 203 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 409 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 345 us MUL_MAT_ID(ffn_moe_down-73): 452 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 401 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 436 us MUL_MAT_ID(ffn_moe_down-75): 187 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 408 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 400 us MUL_MAT_ID(ffn_moe_down-77): 208 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 359 us MUL_MAT_ID(ffn_moe_down-78): 186 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 375 us MUL_MAT_ID(ffn_moe_down-79): 202 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 393 us MUL_MAT_ID(ffn_moe_down-80): 186 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 419 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 391 us MUL_MAT_ID(ffn_moe_down-82): 215 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 394 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 414 us MUL_MAT_ID(ffn_moe_down-84): 255 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 423 us MUL_MAT_ID(ffn_moe_down-85): 651 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 443 us MUL_MAT_ID(ffn_moe_down-86): 189 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 386 us MUL_MAT_ID(ffn_moe_down-87): 207 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 487 us MUL_MAT_ID(ffn_moe_down-88): 183 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 420 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 424 us MUL_MAT_ID(ffn_moe_down-90): 255 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 370 us MUL_MAT_ID(ffn_moe_down-91): 242 us ggml_barrier(...): 45 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 382 us MUL_MAT_ID(ffn_moe_down-25): 203 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 390 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 386 us MUL_MAT_ID(ffn_moe_down-27): 239 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 530 us MUL_MAT_ID(ffn_moe_down-28): 177 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 506 us MUL_MAT_ID(ffn_moe_down-29): 422 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 397 us MUL_MAT_ID(ffn_moe_down-30): 186 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 359 us MUL_MAT_ID(ffn_moe_down-31): 231 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 354 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 427 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 357 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 451 us MUL_MAT_ID(ffn_moe_down-35): 326 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 376 us MUL_MAT_ID(ffn_moe_down-36): 223 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 434 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 386 us MUL_MAT_ID(ffn_moe_down-38): 206 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 406 us MUL_MAT_ID(ffn_moe_down-39): 253 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 382 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 390 us MUL_MAT_ID(ffn_moe_down-41): 234 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 387 us MUL_MAT_ID(ffn_moe_down-42): 208 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 362 us MUL_MAT_ID(ffn_moe_down-43): 216 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 395 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 376 us MUL_MAT_ID(ffn_moe_down-45): 211 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 368 us MUL_MAT_ID(ffn_moe_down-46): 212 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 396 us MUL_MAT_ID(ffn_moe_down-47): 400 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 379 us MUL_MAT_ID(ffn_moe_down-48): 196 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 373 us MUL_MAT_ID(ffn_moe_down-49): 350 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 419 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 358 us MUL_MAT_ID(ffn_moe_down-51): 225 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 377 us MUL_MAT_ID(ffn_moe_down-52): 195 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 431 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 432 us MUL_MAT_ID(ffn_moe_down-54): 187 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 356 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 479 us MUL_MAT_ID(ffn_moe_down-56): 203 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 398 us MUL_MAT_ID(ffn_moe_down-57): 184 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 405 us MUL_MAT_ID(ffn_moe_down-58): 185 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 363 us MUL_MAT_ID(ffn_moe_down-59): 202 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 384 us MUL_MAT_ID(ffn_moe_down-60): 187 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 393 us MUL_MAT_ID(ffn_moe_down-61): 424 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 389 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 375 us MUL_MAT_ID(ffn_moe_down-63): 594 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 372 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 346 us MUL_MAT_ID(ffn_moe_down-65): 208 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 421 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 360 us MUL_MAT_ID(ffn_moe_down-67): 306 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 375 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 409 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 426 us MUL_MAT_ID(ffn_moe_down-70): 237 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 369 us MUL_MAT_ID(ffn_moe_down-71): 214 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 406 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 406 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 400 us MUL_MAT_ID(ffn_moe_down-74): 200 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 403 us MUL_MAT_ID(ffn_moe_down-75): 191 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 393 us MUL_MAT_ID(ffn_moe_down-76): 397 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 366 us MUL_MAT_ID(ffn_moe_down-77): 206 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 421 us MUL_MAT_ID(ffn_moe_down-78): 198 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 370 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 385 us MUL_MAT_ID(ffn_moe_down-80): 211 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 385 us MUL_MAT_ID(ffn_moe_down-81): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 563 us MUL_MAT_ID(ffn_moe_down-82): 207 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 405 us MUL_MAT_ID(ffn_moe_down-83): 597 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 427 us MUL_MAT_ID(ffn_moe_down-84): 245 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 405 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 410 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 414 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 407 us MUL_MAT_ID(ffn_moe_down-88): 212 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 341 us MUL_MAT_ID(ffn_moe_down-89): 433 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 393 us MUL_MAT_ID(ffn_moe_down-90): 690 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 423 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 19 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 514 us MUL_MAT_ID(ffn_moe_down-25): 219 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 411 us MUL_MAT_ID(ffn_moe_down-26): 188 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 514 us MUL_MAT_ID(ffn_moe_down-27): 189 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 418 us MUL_MAT_ID(ffn_moe_down-28): 231 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 437 us MUL_MAT_ID(ffn_moe_down-29): 203 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 436 us MUL_MAT_ID(ffn_moe_down-30): 433 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 368 us MUL_MAT_ID(ffn_moe_down-31): 182 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 417 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 384 us MUL_MAT_ID(ffn_moe_down-33): 240 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 449 us MUL_MAT_ID(ffn_moe_down-34): 215 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 418 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 710 us MUL_MAT_ID(ffn_moe_down-36): 350 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 366 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 414 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 430 us MUL_MAT_ID(ffn_moe_down-39): 239 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 435 us MUL_MAT_ID(ffn_moe_down-40): 227 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 435 us MUL_MAT_ID(ffn_moe_down-41): 207 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 419 us MUL_MAT_ID(ffn_moe_down-42): 475 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 388 us MUL_MAT_ID(ffn_moe_down-43): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 367 us MUL_MAT_ID(ffn_moe_down-44): 258 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 374 us MUL_MAT_ID(ffn_moe_down-45): 225 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 407 us MUL_MAT_ID(ffn_moe_down-46): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 398 us MUL_MAT_ID(ffn_moe_down-47): 363 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 375 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 401 us MUL_MAT_ID(ffn_moe_down-49): 192 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 368 us MUL_MAT_ID(ffn_moe_down-50): 192 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 384 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 421 us MUL_MAT_ID(ffn_moe_down-52): 206 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 376 us MUL_MAT_ID(ffn_moe_down-53): 198 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 431 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 381 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 406 us MUL_MAT_ID(ffn_moe_down-56): 208 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 410 us MUL_MAT_ID(ffn_moe_down-57): 478 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 409 us MUL_MAT_ID(ffn_moe_down-58): 208 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 422 us MUL_MAT_ID(ffn_moe_down-59): 190 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 384 us MUL_MAT_ID(ffn_moe_down-60): 230 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 390 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 376 us MUL_MAT_ID(ffn_moe_down-62): 616 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 258 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 378 us MUL_MAT_ID(ffn_moe_down-64): 200 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 400 us MUL_MAT_ID(ffn_moe_down-65): 454 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 389 us MUL_MAT_ID(ffn_moe_down-66): 183 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 407 us MUL_MAT_ID(ffn_moe_down-67): 229 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 412 us MUL_MAT_ID(ffn_moe_down-68): 245 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 566 us MUL_MAT_ID(ffn_moe_down-69): 326 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 668 us MUL_MAT_ID(ffn_moe_down-70): 295 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 367 us MUL_MAT_ID(ffn_moe_down-71): 208 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 415 us MUL_MAT_ID(ffn_moe_down-72): 205 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 420 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 413 us MUL_MAT_ID(ffn_moe_down-74): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 400 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 441 us MUL_MAT_ID(ffn_moe_down-76): 185 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 465 us MUL_MAT_ID(ffn_moe_down-77): 224 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 433 us MUL_MAT_ID(ffn_moe_down-78): 372 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 434 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 385 us MUL_MAT_ID(ffn_moe_down-80): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 394 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 386 us MUL_MAT_ID(ffn_moe_down-82): 192 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 477 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 402 us MUL_MAT_ID(ffn_moe_down-84): 244 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 403 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 442 us MUL_MAT_ID(ffn_moe_down-86): 198 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 431 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 398 us MUL_MAT_ID(ffn_moe_down-88): 205 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 455 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 390 us MUL_MAT_ID(ffn_moe_down-90): 225 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 407 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 6 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 365 us MUL_MAT_ID(ffn_moe_down-25): 207 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 428 us MUL_MAT_ID(ffn_moe_down-26): 197 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 428 us MUL_MAT_ID(ffn_moe_down-27): 206 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 375 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 369 us MUL_MAT_ID(ffn_moe_down-29): 223 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 402 us MUL_MAT_ID(ffn_moe_down-30): 229 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 414 us MUL_MAT_ID(ffn_moe_down-31): 203 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 436 us MUL_MAT_ID(ffn_moe_down-32): 188 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 409 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 406 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 360 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 384 us MUL_MAT_ID(ffn_moe_down-36): 255 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 407 us MUL_MAT_ID(ffn_moe_down-37): 605 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 428 us MUL_MAT_ID(ffn_moe_down-38): 248 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 424 us MUL_MAT_ID(ffn_moe_down-39): 266 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 656 us MUL_MAT_ID(ffn_moe_down-40): 292 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 430 us MUL_MAT_ID(ffn_moe_down-41): 199 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 527 us MUL_MAT_ID(ffn_moe_down-42): 205 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 416 us MUL_MAT_ID(ffn_moe_down-43): 189 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 413 us MUL_MAT_ID(ffn_moe_down-44): 282 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 460 us MUL_MAT_ID(ffn_moe_down-45): 299 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 387 us MUL_MAT_ID(ffn_moe_down-46): 195 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 542 us MUL_MAT_ID(ffn_moe_down-47): 186 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 382 us MUL_MAT_ID(ffn_moe_down-48): 209 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 436 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 443 us MUL_MAT_ID(ffn_moe_down-50): 227 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 570 us MUL_MAT_ID(ffn_moe_down-51): 178 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 352 us MUL_MAT_ID(ffn_moe_down-52): 290 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 384 us MUL_MAT_ID(ffn_moe_down-53): 240 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 404 us MUL_MAT_ID(ffn_moe_down-54): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 453 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 408 us MUL_MAT_ID(ffn_moe_down-56): 191 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 384 us MUL_MAT_ID(ffn_moe_down-57): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 377 us MUL_MAT_ID(ffn_moe_down-58): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 397 us MUL_MAT_ID(ffn_moe_down-59): 427 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 377 us MUL_MAT_ID(ffn_moe_down-60): 217 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 414 us MUL_MAT_ID(ffn_moe_down-61): 182 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 365 us MUL_MAT_ID(ffn_moe_down-62): 207 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 383 us MUL_MAT_ID(ffn_moe_down-63): 253 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 439 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 413 us MUL_MAT_ID(ffn_moe_down-65): 189 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 466 us MUL_MAT_ID(ffn_moe_down-66): 209 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 407 us MUL_MAT_ID(ffn_moe_down-67): 195 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 437 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 388 us MUL_MAT_ID(ffn_moe_down-69): 574 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 499 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 427 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 357 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 441 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 368 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 434 us MUL_MAT_ID(ffn_moe_down-75): 289 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 403 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 361 us MUL_MAT_ID(ffn_moe_down-77): 415 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 379 us MUL_MAT_ID(ffn_moe_down-78): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 343 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 389 us MUL_MAT_ID(ffn_moe_down-80): 267 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 196 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 405 us MUL_MAT_ID(ffn_moe_down-82): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 423 us MUL_MAT_ID(ffn_moe_down-83): 424 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 336 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 403 us MUL_MAT_ID(ffn_moe_down-85): 685 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 412 us MUL_MAT_ID(ffn_moe_down-86): 232 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 366 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 395 us MUL_MAT_ID(ffn_moe_down-88): 204 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 457 us MUL_MAT_ID(ffn_moe_down-89): 434 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 375 us MUL_MAT_ID(ffn_moe_down-90): 200 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 449 us MUL_MAT_ID(ffn_moe_down-91): 246 us ggml_barrier(...): 198 us GET_ROWS(inp_embd): 11 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 364 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 427 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 401 us MUL_MAT_ID(ffn_moe_down-27): 351 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 399 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 456 us MUL_MAT_ID(ffn_moe_down-29): 423 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 523 us MUL_MAT_ID(ffn_moe_down-30): 205 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 444 us MUL_MAT_ID(ffn_moe_down-31): 207 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 413 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 410 us MUL_MAT_ID(ffn_moe_down-33): 210 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 443 us MUL_MAT_ID(ffn_moe_down-34): 221 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 369 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 404 us MUL_MAT_ID(ffn_moe_down-36): 182 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 415 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 429 us MUL_MAT_ID(ffn_moe_down-38): 336 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 397 us MUL_MAT_ID(ffn_moe_down-39): 330 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 416 us MUL_MAT_ID(ffn_moe_down-40): 221 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 446 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 413 us MUL_MAT_ID(ffn_moe_down-42): 216 us ggml_barrier(...): 367 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 417 us MUL_MAT_ID(ffn_moe_down-43): 208 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 411 us MUL_MAT_ID(ffn_moe_down-44): 376 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 411 us MUL_MAT_ID(ffn_moe_down-45): 394 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 395 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 435 us MUL_MAT_ID(ffn_moe_down-47): 202 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 349 us MUL_MAT_ID(ffn_moe_down-48): 242 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 411 us MUL_MAT_ID(ffn_moe_down-49): 417 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 649 us MUL_MAT_ID(ffn_moe_down-50): 216 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 379 us MUL_MAT_ID(ffn_moe_down-51): 187 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 499 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 358 us MUL_MAT_ID(ffn_moe_down-53): 206 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 427 us MUL_MAT_ID(ffn_moe_down-54): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 373 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 358 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 418 us MUL_MAT_ID(ffn_moe_down-57): 202 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 436 us MUL_MAT_ID(ffn_moe_down-58): 174 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 403 us MUL_MAT_ID(ffn_moe_down-59): 213 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 426 us MUL_MAT_ID(ffn_moe_down-60): 247 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 654 us MUL_MAT_ID(ffn_moe_down-61): 278 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 356 us MUL_MAT_ID(ffn_moe_down-62): 319 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 406 us MUL_MAT_ID(ffn_moe_down-63): 202 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 369 us MUL_MAT_ID(ffn_moe_down-64): 201 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 375 us MUL_MAT_ID(ffn_moe_down-65): 204 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 347 us MUL_MAT_ID(ffn_moe_down-66): 211 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 435 us MUL_MAT_ID(ffn_moe_down-67): 182 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 434 us MUL_MAT_ID(ffn_moe_down-68): 209 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 466 us MUL_MAT_ID(ffn_moe_down-69): 172 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 390 us MUL_MAT_ID(ffn_moe_down-70): 215 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 384 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 345 us MUL_MAT_ID(ffn_moe_down-72): 585 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 429 us MUL_MAT_ID(ffn_moe_down-73): 304 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 400 us MUL_MAT_ID(ffn_moe_down-74): 185 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 402 us MUL_MAT_ID(ffn_moe_down-75): 337 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 399 us MUL_MAT_ID(ffn_moe_down-76): 188 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 381 us MUL_MAT_ID(ffn_moe_down-77): 180 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 381 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 400 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 451 us MUL_MAT_ID(ffn_moe_down-80): 307 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 373 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 419 us MUL_MAT_ID(ffn_moe_down-83): 207 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 402 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 473 us MUL_MAT_ID(ffn_moe_down-85): 446 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 389 us MUL_MAT_ID(ffn_moe_down-86): 189 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 399 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 403 us MUL_MAT_ID(ffn_moe_down-88): 205 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 374 us MUL_MAT_ID(ffn_moe_down-89): 215 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 400 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 412 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 19 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 377 us MUL_MAT_ID(ffn_moe_down-25): 199 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 424 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 360 us MUL_MAT_ID(ffn_moe_down-27): 200 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 352 us MUL_MAT_ID(ffn_moe_down-28): 214 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 424 us MUL_MAT_ID(ffn_moe_down-29): 213 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 441 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 384 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 379 us MUL_MAT_ID(ffn_moe_down-32): 219 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 400 us MUL_MAT_ID(ffn_moe_down-33): 205 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 371 us MUL_MAT_ID(ffn_moe_down-34): 302 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 426 us MUL_MAT_ID(ffn_moe_down-35): 208 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 364 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 402 us MUL_MAT_ID(ffn_moe_down-37): 254 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 379 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 438 us MUL_MAT_ID(ffn_moe_down-39): 688 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 341 us MUL_MAT_ID(ffn_moe_down-40): 246 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 385 us MUL_MAT_ID(ffn_moe_down-41): 199 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 392 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 445 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 410 us MUL_MAT_ID(ffn_moe_down-44): 223 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 384 us MUL_MAT_ID(ffn_moe_down-45): 225 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 453 us MUL_MAT_ID(ffn_moe_down-46): 236 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 400 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 368 us MUL_MAT_ID(ffn_moe_down-48): 222 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 397 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 387 us MUL_MAT_ID(ffn_moe_down-50): 217 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 377 us MUL_MAT_ID(ffn_moe_down-51): 204 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 410 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 375 us MUL_MAT_ID(ffn_moe_down-53): 196 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 391 us MUL_MAT_ID(ffn_moe_down-54): 218 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 432 us MUL_MAT_ID(ffn_moe_down-55): 259 us ggml_barrier(...): 384 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 420 us MUL_MAT_ID(ffn_moe_down-56): 208 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 413 us MUL_MAT_ID(ffn_moe_down-57): 233 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 399 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 417 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 496 us MUL_MAT_ID(ffn_moe_down-60): 259 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 512 us MUL_MAT_ID(ffn_moe_down-61): 347 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 511 us MUL_MAT_ID(ffn_moe_down-62): 276 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 473 us MUL_MAT_ID(ffn_moe_down-63): 202 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 392 us MUL_MAT_ID(ffn_moe_down-64): 218 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 381 us MUL_MAT_ID(ffn_moe_down-65): 199 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 430 us MUL_MAT_ID(ffn_moe_down-66): 205 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 389 us MUL_MAT_ID(ffn_moe_down-67): 210 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 403 us MUL_MAT_ID(ffn_moe_down-68): 205 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 380 us MUL_MAT_ID(ffn_moe_down-69): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 394 us MUL_MAT_ID(ffn_moe_down-70): 456 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 396 us MUL_MAT_ID(ffn_moe_down-71): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 226 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 624 us MUL_MAT_ID(ffn_moe_down-73): 205 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 392 us MUL_MAT_ID(ffn_moe_down-74): 375 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 443 us MUL_MAT_ID(ffn_moe_down-75): 194 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 382 us MUL_MAT_ID(ffn_moe_down-76): 252 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 394 us MUL_MAT_ID(ffn_moe_down-77): 225 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 400 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 421 us MUL_MAT_ID(ffn_moe_down-79): 190 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 366 us MUL_MAT_ID(ffn_moe_down-80): 385 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 433 us MUL_MAT_ID(ffn_moe_down-81): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 369 us MUL_MAT_ID(ffn_moe_down-82): 564 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 365 us MUL_MAT_ID(ffn_moe_down-83): 242 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 425 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 402 us MUL_MAT_ID(ffn_moe_down-85): 225 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 459 us MUL_MAT_ID(ffn_moe_down-86): 227 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 396 us MUL_MAT_ID(ffn_moe_down-87): 188 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 392 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 656 us MUL_MAT_ID(ffn_moe_down-89): 230 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 435 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 389 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 33 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 385 us MUL_MAT_ID(ffn_moe_down-25): 247 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 430 us MUL_MAT_ID(ffn_moe_down-26): 197 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 406 us MUL_MAT_ID(ffn_moe_down-27): 296 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 398 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 192 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 385 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 398 us MUL_MAT_ID(ffn_moe_down-31): 177 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 353 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 476 us MUL_MAT_ID(ffn_moe_down-33): 387 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 378 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 397 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 366 us MUL_MAT_ID(ffn_moe_down-36): 385 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 398 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 423 us MUL_MAT_ID(ffn_moe_down-38): 218 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 460 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 376 us MUL_MAT_ID(ffn_moe_down-40): 254 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 403 us MUL_MAT_ID(ffn_moe_down-41): 223 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 385 us MUL_MAT_ID(ffn_moe_down-42): 185 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 405 us MUL_MAT_ID(ffn_moe_down-43): 208 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 408 us MUL_MAT_ID(ffn_moe_down-44): 261 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 432 us MUL_MAT_ID(ffn_moe_down-45): 650 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 439 us MUL_MAT_ID(ffn_moe_down-46): 209 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 407 us MUL_MAT_ID(ffn_moe_down-47): 209 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 371 us MUL_MAT_ID(ffn_moe_down-48): 197 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 491 us MUL_MAT_ID(ffn_moe_down-49): 238 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 424 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 421 us MUL_MAT_ID(ffn_moe_down-51): 209 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 393 us MUL_MAT_ID(ffn_moe_down-52): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 394 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 499 us MUL_MAT_ID(ffn_moe_down-54): 214 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 396 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 465 us MUL_MAT_ID(ffn_moe_down-56): 392 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 428 us MUL_MAT_ID(ffn_moe_down-57): 214 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 432 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 451 us MUL_MAT_ID(ffn_moe_down-59): 270 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 444 us MUL_MAT_ID(ffn_moe_down-60): 270 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 469 us MUL_MAT_ID(ffn_moe_down-61): 307 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 435 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 382 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 476 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 441 us MUL_MAT_ID(ffn_moe_down-65): 184 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 382 us MUL_MAT_ID(ffn_moe_down-66): 305 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 391 us MUL_MAT_ID(ffn_moe_down-67): 191 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 459 us MUL_MAT_ID(ffn_moe_down-68): 210 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 404 us MUL_MAT_ID(ffn_moe_down-69): 293 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 458 us MUL_MAT_ID(ffn_moe_down-70): 238 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 900 us MUL_MAT_ID(ffn_moe_down-71): 357 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 529 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 274 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 387 us MUL_MAT_ID(ffn_moe_down-74): 558 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 384 us MUL_MAT_ID(ffn_moe_down-75): 308 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 448 us MUL_MAT_ID(ffn_moe_down-76): 200 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 488 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 519 us MUL_MAT_ID(ffn_moe_down-78): 201 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 395 us MUL_MAT_ID(ffn_moe_down-79): 210 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 540 us MUL_MAT_ID(ffn_moe_down-80): 208 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 435 us MUL_MAT_ID(ffn_moe_down-81): 337 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 447 us MUL_MAT_ID(ffn_moe_down-82): 194 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 381 us MUL_MAT_ID(ffn_moe_down-83): 207 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 372 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 539 us MUL_MAT_ID(ffn_moe_down-85): 279 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 392 us MUL_MAT_ID(ffn_moe_down-86): 173 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 377 us MUL_MAT_ID(ffn_moe_down-87): 206 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 371 us MUL_MAT_ID(ffn_moe_down-88): 228 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 351 us MUL_MAT_ID(ffn_moe_down-89): 195 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 403 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 384 us MUL_MAT_ID(ffn_moe_down-91): 229 us ggml_barrier(...): 214 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 441 us MUL_MAT_ID(ffn_moe_down-25): 229 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 393 us MUL_MAT_ID(ffn_moe_down-26): 345 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 336 us MUL_MAT_ID(ffn_moe_down-27): 184 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 380 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 212 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 407 us MUL_MAT_ID(ffn_moe_down-30): 187 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 374 us MUL_MAT_ID(ffn_moe_down-31): 318 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 379 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 399 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 346 us MUL_MAT_ID(ffn_moe_down-34): 554 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 358 us MUL_MAT_ID(ffn_moe_down-35): 201 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 438 us MUL_MAT_ID(ffn_moe_down-36): 608 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 396 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 205 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 435 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 385 us MUL_MAT_ID(ffn_moe_down-40): 700 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 354 us MUL_MAT_ID(ffn_moe_down-41): 205 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 397 us MUL_MAT_ID(ffn_moe_down-42): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 406 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 334 us MUL_MAT_ID(ffn_moe_down-44): 194 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 391 us MUL_MAT_ID(ffn_moe_down-45): 226 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 408 us MUL_MAT_ID(ffn_moe_down-46): 195 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 407 us MUL_MAT_ID(ffn_moe_down-47): 187 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 345 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 405 us MUL_MAT_ID(ffn_moe_down-49): 473 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 408 us MUL_MAT_ID(ffn_moe_down-50): 211 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 391 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 421 us MUL_MAT_ID(ffn_moe_down-52): 495 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 376 us MUL_MAT_ID(ffn_moe_down-53): 276 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 351 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 378 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 398 us MUL_MAT_ID(ffn_moe_down-56): 293 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 407 us MUL_MAT_ID(ffn_moe_down-57): 204 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 441 us MUL_MAT_ID(ffn_moe_down-58): 369 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 409 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 420 us MUL_MAT_ID(ffn_moe_down-60): 209 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 421 us MUL_MAT_ID(ffn_moe_down-61): 199 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 429 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 399 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 368 us MUL_MAT_ID(ffn_moe_down-64): 183 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 386 us MUL_MAT_ID(ffn_moe_down-65): 199 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 584 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 427 us MUL_MAT_ID(ffn_moe_down-67): 182 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 424 us MUL_MAT_ID(ffn_moe_down-68): 196 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 536 us MUL_MAT_ID(ffn_moe_down-69): 176 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 428 us MUL_MAT_ID(ffn_moe_down-70): 185 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 396 us MUL_MAT_ID(ffn_moe_down-71): 185 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 443 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 405 us MUL_MAT_ID(ffn_moe_down-73): 357 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 408 us MUL_MAT_ID(ffn_moe_down-74): 204 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 450 us MUL_MAT_ID(ffn_moe_down-75): 228 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 387 us MUL_MAT_ID(ffn_moe_down-76): 179 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 358 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 513 us MUL_MAT_ID(ffn_moe_down-78): 396 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 423 us MUL_MAT_ID(ffn_moe_down-79): 477 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 625 us MUL_MAT_ID(ffn_moe_down-80): 218 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 401 us MUL_MAT_ID(ffn_moe_down-81): 212 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 421 us MUL_MAT_ID(ffn_moe_down-82): 288 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 409 us MUL_MAT_ID(ffn_moe_down-83): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 398 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 392 us MUL_MAT_ID(ffn_moe_down-85): 409 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 391 us MUL_MAT_ID(ffn_moe_down-86): 196 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 404 us MUL_MAT_ID(ffn_moe_down-87): 223 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 370 us MUL_MAT_ID(ffn_moe_down-88): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 357 us MUL_MAT_ID(ffn_moe_down-89): 177 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 394 us MUL_MAT_ID(ffn_moe_down-90): 659 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 561 us MUL_MAT_ID(ffn_moe_down-91): 421 us ggml_barrier(...): 20 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 599 us MUL_MAT_ID(ffn_moe_down-25): 313 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 380 us MUL_MAT_ID(ffn_moe_down-26): 544 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 428 us MUL_MAT_ID(ffn_moe_down-27): 198 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 442 us MUL_MAT_ID(ffn_moe_down-28): 182 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 413 us MUL_MAT_ID(ffn_moe_down-29): 384 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 362 us MUL_MAT_ID(ffn_moe_down-30): 219 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 394 us MUL_MAT_ID(ffn_moe_down-31): 185 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 427 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 343 us MUL_MAT_ID(ffn_moe_down-33): 344 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 427 us MUL_MAT_ID(ffn_moe_down-35): 189 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 478 us MUL_MAT_ID(ffn_moe_down-36): 390 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 371 us MUL_MAT_ID(ffn_moe_down-37): 579 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 397 us MUL_MAT_ID(ffn_moe_down-38): 373 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 393 us MUL_MAT_ID(ffn_moe_down-39): 248 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 372 us MUL_MAT_ID(ffn_moe_down-40): 238 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 404 us MUL_MAT_ID(ffn_moe_down-41): 226 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 404 us MUL_MAT_ID(ffn_moe_down-42): 204 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 376 us MUL_MAT_ID(ffn_moe_down-43): 233 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 377 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 642 us MUL_MAT_ID(ffn_moe_down-45): 261 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 395 us MUL_MAT_ID(ffn_moe_down-46): 414 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 398 us MUL_MAT_ID(ffn_moe_down-47): 402 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 371 us MUL_MAT_ID(ffn_moe_down-48): 372 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 378 us MUL_MAT_ID(ffn_moe_down-49): 207 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 506 us MUL_MAT_ID(ffn_moe_down-50): 177 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 394 us MUL_MAT_ID(ffn_moe_down-51): 207 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 408 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 380 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 403 us MUL_MAT_ID(ffn_moe_down-54): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 358 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 412 us MUL_MAT_ID(ffn_moe_down-56): 369 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 387 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 388 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 431 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 407 us MUL_MAT_ID(ffn_moe_down-60): 331 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 399 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 494 us MUL_MAT_ID(ffn_moe_down-62): 201 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 401 us MUL_MAT_ID(ffn_moe_down-63): 207 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 426 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 389 us MUL_MAT_ID(ffn_moe_down-65): 267 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 402 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 399 us MUL_MAT_ID(ffn_moe_down-67): 213 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 403 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 387 us MUL_MAT_ID(ffn_moe_down-69): 201 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 433 us MUL_MAT_ID(ffn_moe_down-70): 205 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 396 us MUL_MAT_ID(ffn_moe_down-71): 209 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 351 us MUL_MAT_ID(ffn_moe_down-72): 188 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 368 us MUL_MAT_ID(ffn_moe_down-73): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 340 us MUL_MAT_ID(ffn_moe_down-74): 206 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 419 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 408 us MUL_MAT_ID(ffn_moe_down-76): 200 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 452 us MUL_MAT_ID(ffn_moe_down-77): 199 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 374 us MUL_MAT_ID(ffn_moe_down-78): 223 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 451 us MUL_MAT_ID(ffn_moe_down-79): 235 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 421 us MUL_MAT_ID(ffn_moe_down-80): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 434 us MUL_MAT_ID(ffn_moe_down-81): 182 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 465 us MUL_MAT_ID(ffn_moe_down-82): 364 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 475 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 350 us MUL_MAT_ID(ffn_moe_down-84): 660 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 418 us MUL_MAT_ID(ffn_moe_down-85): 243 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 388 us MUL_MAT_ID(ffn_moe_down-86): 198 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 419 us MUL_MAT_ID(ffn_moe_down-87): 201 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 448 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 421 us MUL_MAT_ID(ffn_moe_down-89): 179 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 516 us MUL_MAT_ID(ffn_moe_down-90): 686 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 653 us MUL_MAT_ID(ffn_moe_down-91): 357 us ggml_barrier(...): 4 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 397 us MUL_MAT_ID(ffn_moe_down-25): 225 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 437 us MUL_MAT_ID(ffn_moe_down-26): 202 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 453 us MUL_MAT_ID(ffn_moe_down-27): 189 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 423 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 353 us MUL_MAT_ID(ffn_moe_down-29): 208 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 451 us MUL_MAT_ID(ffn_moe_down-30): 198 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 356 us MUL_MAT_ID(ffn_moe_down-31): 207 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 415 us MUL_MAT_ID(ffn_moe_down-32): 281 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 417 us MUL_MAT_ID(ffn_moe_down-33): 202 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 203 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 382 us MUL_MAT_ID(ffn_moe_down-35): 185 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 426 us MUL_MAT_ID(ffn_moe_down-36): 564 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 387 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 399 us MUL_MAT_ID(ffn_moe_down-38): 288 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 373 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 450 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 438 us MUL_MAT_ID(ffn_moe_down-41): 186 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 359 us MUL_MAT_ID(ffn_moe_down-42): 203 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 428 us MUL_MAT_ID(ffn_moe_down-43): 186 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 395 us MUL_MAT_ID(ffn_moe_down-44): 212 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 436 us MUL_MAT_ID(ffn_moe_down-45): 357 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 376 us MUL_MAT_ID(ffn_moe_down-46): 398 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 445 us MUL_MAT_ID(ffn_moe_down-47): 236 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 390 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 342 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 427 us MUL_MAT_ID(ffn_moe_down-50): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 462 us MUL_MAT_ID(ffn_moe_down-51): 207 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 436 us MUL_MAT_ID(ffn_moe_down-52): 211 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 397 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 452 us MUL_MAT_ID(ffn_moe_down-54): 224 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 457 us MUL_MAT_ID(ffn_moe_down-55): 208 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 398 us MUL_MAT_ID(ffn_moe_down-56): 426 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 384 us MUL_MAT_ID(ffn_moe_down-57): 189 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 413 us MUL_MAT_ID(ffn_moe_down-58): 417 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 395 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 442 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 384 us MUL_MAT_ID(ffn_moe_down-61): 181 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 428 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 420 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 509 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 424 us MUL_MAT_ID(ffn_moe_down-65): 213 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 441 us MUL_MAT_ID(ffn_moe_down-66): 185 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 395 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 349 us MUL_MAT_ID(ffn_moe_down-68): 592 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 403 us MUL_MAT_ID(ffn_moe_down-69): 334 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 390 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 384 us MUL_MAT_ID(ffn_moe_down-71): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 443 us MUL_MAT_ID(ffn_moe_down-72): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 410 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 416 us MUL_MAT_ID(ffn_moe_down-74): 203 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 438 us MUL_MAT_ID(ffn_moe_down-75): 377 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 424 us MUL_MAT_ID(ffn_moe_down-76): 182 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 209 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 380 us MUL_MAT_ID(ffn_moe_down-78): 362 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 374 us MUL_MAT_ID(ffn_moe_down-79): 187 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 407 us MUL_MAT_ID(ffn_moe_down-80): 251 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 400 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 368 us MUL_MAT_ID(ffn_moe_down-82): 426 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 421 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 367 us MUL_MAT_ID(ffn_moe_down-84): 201 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 418 us MUL_MAT_ID(ffn_moe_down-85): 768 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 370 us MUL_MAT_ID(ffn_moe_down-86): 362 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 415 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 350 us MUL_MAT_ID(ffn_moe_down-88): 200 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 427 us MUL_MAT_ID(ffn_moe_down-89): 205 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 468 us MUL_MAT_ID(ffn_moe_down-90): 226 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 428 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 29 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 360 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 370 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 363 us MUL_MAT_ID(ffn_moe_down-27): 191 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 442 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 526 us MUL_MAT_ID(ffn_moe_down-29): 476 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 383 us MUL_MAT_ID(ffn_moe_down-30): 197 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 391 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 417 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 485 us MUL_MAT_ID(ffn_moe_down-33): 197 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 425 us MUL_MAT_ID(ffn_moe_down-34): 286 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 390 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 484 us MUL_MAT_ID(ffn_moe_down-36): 288 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 549 us MUL_MAT_ID(ffn_moe_down-37): 241 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 429 us MUL_MAT_ID(ffn_moe_down-38): 187 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 411 us MUL_MAT_ID(ffn_moe_down-39): 243 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 419 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 441 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 461 us MUL_MAT_ID(ffn_moe_down-42): 471 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 393 us MUL_MAT_ID(ffn_moe_down-43): 214 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 382 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 441 us MUL_MAT_ID(ffn_moe_down-45): 500 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 399 us MUL_MAT_ID(ffn_moe_down-46): 564 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 367 us MUL_MAT_ID(ffn_moe_down-47): 200 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 370 us MUL_MAT_ID(ffn_moe_down-48): 262 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 394 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 352 us MUL_MAT_ID(ffn_moe_down-50): 205 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 404 us MUL_MAT_ID(ffn_moe_down-51): 199 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 420 us MUL_MAT_ID(ffn_moe_down-52): 202 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 413 us MUL_MAT_ID(ffn_moe_down-53): 217 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 447 us MUL_MAT_ID(ffn_moe_down-54): 362 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 378 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 431 us MUL_MAT_ID(ffn_moe_down-56): 205 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 388 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 409 us MUL_MAT_ID(ffn_moe_down-58): 189 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 389 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 373 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 368 us MUL_MAT_ID(ffn_moe_down-61): 588 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 410 us MUL_MAT_ID(ffn_moe_down-62): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 347 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 359 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 385 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 369 us MUL_MAT_ID(ffn_moe_down-66): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 358 us MUL_MAT_ID(ffn_moe_down-67): 206 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 569 us MUL_MAT_ID(ffn_moe_down-68): 577 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 393 us MUL_MAT_ID(ffn_moe_down-69): 219 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 413 us MUL_MAT_ID(ffn_moe_down-70): 189 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 391 us MUL_MAT_ID(ffn_moe_down-71): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 402 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 392 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 382 us MUL_MAT_ID(ffn_moe_down-74): 188 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 415 us MUL_MAT_ID(ffn_moe_down-75): 186 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 391 us MUL_MAT_ID(ffn_moe_down-76): 197 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 369 us MUL_MAT_ID(ffn_moe_down-77): 187 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 420 us MUL_MAT_ID(ffn_moe_down-78): 186 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 526 us MUL_MAT_ID(ffn_moe_down-79): 176 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 411 us MUL_MAT_ID(ffn_moe_down-80): 213 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 414 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 399 us MUL_MAT_ID(ffn_moe_down-82): 209 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 397 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 383 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 426 us MUL_MAT_ID(ffn_moe_down-85): 239 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 360 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 416 us MUL_MAT_ID(ffn_moe_down-87): 189 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 375 us MUL_MAT_ID(ffn_moe_down-88): 181 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 360 us MUL_MAT_ID(ffn_moe_down-89): 202 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 374 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 405 us MUL_MAT_ID(ffn_moe_down-91): 246 us ggml_barrier(...): 12 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 358 us MUL_MAT_ID(ffn_moe_down-25): 215 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 392 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 434 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 413 us MUL_MAT_ID(ffn_moe_down-28): 276 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 450 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 476 us MUL_MAT_ID(ffn_moe_down-30): 365 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 429 us MUL_MAT_ID(ffn_moe_down-31): 241 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 703 us MUL_MAT_ID(ffn_moe_down-32): 329 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 435 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 473 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 390 us MUL_MAT_ID(ffn_moe_down-35): 210 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 386 us MUL_MAT_ID(ffn_moe_down-36): 213 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 351 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 398 us MUL_MAT_ID(ffn_moe_down-38): 202 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 397 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 412 us MUL_MAT_ID(ffn_moe_down-40): 439 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 362 us MUL_MAT_ID(ffn_moe_down-41): 212 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 406 us MUL_MAT_ID(ffn_moe_down-42): 397 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 396 us MUL_MAT_ID(ffn_moe_down-43): 225 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 634 us MUL_MAT_ID(ffn_moe_down-44): 246 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 387 us MUL_MAT_ID(ffn_moe_down-45): 307 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 416 us MUL_MAT_ID(ffn_moe_down-46): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 401 us MUL_MAT_ID(ffn_moe_down-47): 554 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 410 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 361 us MUL_MAT_ID(ffn_moe_down-49): 206 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 384 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 397 us MUL_MAT_ID(ffn_moe_down-51): 203 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 446 us MUL_MAT_ID(ffn_moe_down-52): 216 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 432 us MUL_MAT_ID(ffn_moe_down-53): 213 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 446 us MUL_MAT_ID(ffn_moe_down-54): 218 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 399 us MUL_MAT_ID(ffn_moe_down-55): 222 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 443 us MUL_MAT_ID(ffn_moe_down-56): 331 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 457 us MUL_MAT_ID(ffn_moe_down-57): 208 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 408 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 464 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 447 us MUL_MAT_ID(ffn_moe_down-60): 234 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 347 us MUL_MAT_ID(ffn_moe_down-61): 226 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 409 us MUL_MAT_ID(ffn_moe_down-62): 187 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 387 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 417 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 386 us MUL_MAT_ID(ffn_moe_down-65): 486 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 398 us MUL_MAT_ID(ffn_moe_down-66): 202 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 412 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 386 us MUL_MAT_ID(ffn_moe_down-68): 189 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 435 us MUL_MAT_ID(ffn_moe_down-69): 205 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 428 us MUL_MAT_ID(ffn_moe_down-70): 437 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 350 us MUL_MAT_ID(ffn_moe_down-71): 432 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 421 us MUL_MAT_ID(ffn_moe_down-72): 218 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 410 us MUL_MAT_ID(ffn_moe_down-73): 532 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 504 us MUL_MAT_ID(ffn_moe_down-74): 190 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 441 us MUL_MAT_ID(ffn_moe_down-75): 213 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 389 us MUL_MAT_ID(ffn_moe_down-76): 352 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 355 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 476 us MUL_MAT_ID(ffn_moe_down-78): 278 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 421 us MUL_MAT_ID(ffn_moe_down-79): 262 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 405 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 384 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 353 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 413 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 401 us MUL_MAT_ID(ffn_moe_down-84): 224 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 391 us MUL_MAT_ID(ffn_moe_down-85): 245 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 372 us MUL_MAT_ID(ffn_moe_down-86): 217 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 421 us MUL_MAT_ID(ffn_moe_down-87): 331 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 430 us MUL_MAT_ID(ffn_moe_down-88): 206 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 394 us MUL_MAT_ID(ffn_moe_down-89): 203 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 423 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 430 us MUL_MAT_ID(ffn_moe_down-91): 256 us ggml_barrier(...): 14 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 372 us MUL_MAT_ID(ffn_moe_down-25): 205 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 373 us MUL_MAT_ID(ffn_moe_down-26): 186 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 391 us MUL_MAT_ID(ffn_moe_down-27): 279 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 426 us MUL_MAT_ID(ffn_moe_down-28): 199 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 395 us MUL_MAT_ID(ffn_moe_down-29): 449 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 389 us MUL_MAT_ID(ffn_moe_down-30): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 361 us MUL_MAT_ID(ffn_moe_down-31): 192 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 397 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 367 us MUL_MAT_ID(ffn_moe_down-33): 422 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 392 us MUL_MAT_ID(ffn_moe_down-34): 456 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 391 us MUL_MAT_ID(ffn_moe_down-35): 211 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 380 us MUL_MAT_ID(ffn_moe_down-36): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 364 us MUL_MAT_ID(ffn_moe_down-37): 238 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 370 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 402 us MUL_MAT_ID(ffn_moe_down-39): 299 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 353 us MUL_MAT_ID(ffn_moe_down-40): 434 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 366 us MUL_MAT_ID(ffn_moe_down-41): 186 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 400 us MUL_MAT_ID(ffn_moe_down-42): 185 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 373 us MUL_MAT_ID(ffn_moe_down-43): 459 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 370 us MUL_MAT_ID(ffn_moe_down-44): 224 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 421 us MUL_MAT_ID(ffn_moe_down-45): 237 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 673 us MUL_MAT_ID(ffn_moe_down-46): 355 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 353 us MUL_MAT_ID(ffn_moe_down-47): 202 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 394 us MUL_MAT_ID(ffn_moe_down-48): 345 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 397 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 423 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 357 us MUL_MAT_ID(ffn_moe_down-51): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 374 us MUL_MAT_ID(ffn_moe_down-52): 447 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 362 us MUL_MAT_ID(ffn_moe_down-53): 198 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 399 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 460 us MUL_MAT_ID(ffn_moe_down-55): 296 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 368 us MUL_MAT_ID(ffn_moe_down-56): 187 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 452 us MUL_MAT_ID(ffn_moe_down-57): 224 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 617 us MUL_MAT_ID(ffn_moe_down-58): 372 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 409 us MUL_MAT_ID(ffn_moe_down-59): 188 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 424 us MUL_MAT_ID(ffn_moe_down-60): 590 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 407 us MUL_MAT_ID(ffn_moe_down-61): 212 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 382 us MUL_MAT_ID(ffn_moe_down-62): 365 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 433 us MUL_MAT_ID(ffn_moe_down-63): 200 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 335 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 448 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 448 us MUL_MAT_ID(ffn_moe_down-66): 187 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 402 us MUL_MAT_ID(ffn_moe_down-67): 215 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 454 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 384 us MUL_MAT_ID(ffn_moe_down-69): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 405 us MUL_MAT_ID(ffn_moe_down-70): 186 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 375 us MUL_MAT_ID(ffn_moe_down-71): 187 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 428 us MUL_MAT_ID(ffn_moe_down-72): 187 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 342 us MUL_MAT_ID(ffn_moe_down-73): 201 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 492 us MUL_MAT_ID(ffn_moe_down-74): 179 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 425 us MUL_MAT_ID(ffn_moe_down-75): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 405 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 442 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 505 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 425 us MUL_MAT_ID(ffn_moe_down-79): 213 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 400 us MUL_MAT_ID(ffn_moe_down-80): 322 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 416 us MUL_MAT_ID(ffn_moe_down-81): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 395 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 417 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 438 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 389 us MUL_MAT_ID(ffn_moe_down-85): 258 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 389 us MUL_MAT_ID(ffn_moe_down-86): 417 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 380 us MUL_MAT_ID(ffn_moe_down-87): 570 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 416 us MUL_MAT_ID(ffn_moe_down-88): 426 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 368 us MUL_MAT_ID(ffn_moe_down-89): 204 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 454 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 378 us MUL_MAT_ID(ffn_moe_down-91): 677 us ggml_barrier(...): 52 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 401 us MUL_MAT_ID(ffn_moe_down-25): 221 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 421 us MUL_MAT_ID(ffn_moe_down-26): 197 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 415 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 338 us MUL_MAT_ID(ffn_moe_down-28): 200 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 461 us MUL_MAT_ID(ffn_moe_down-29): 219 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 364 us MUL_MAT_ID(ffn_moe_down-30): 189 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 424 us MUL_MAT_ID(ffn_moe_down-31): 183 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 451 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 376 us MUL_MAT_ID(ffn_moe_down-33): 220 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 418 us MUL_MAT_ID(ffn_moe_down-34): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 353 us MUL_MAT_ID(ffn_moe_down-35): 473 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 372 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 355 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 431 us MUL_MAT_ID(ffn_moe_down-38): 335 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 401 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 387 us MUL_MAT_ID(ffn_moe_down-40): 216 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 414 us MUL_MAT_ID(ffn_moe_down-41): 561 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 416 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 418 us MUL_MAT_ID(ffn_moe_down-43): 204 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 339 us MUL_MAT_ID(ffn_moe_down-44): 194 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 400 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 408 us MUL_MAT_ID(ffn_moe_down-46): 188 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 406 us MUL_MAT_ID(ffn_moe_down-47): 384 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 403 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 403 us MUL_MAT_ID(ffn_moe_down-49): 202 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 390 us MUL_MAT_ID(ffn_moe_down-50): 214 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 428 us MUL_MAT_ID(ffn_moe_down-51): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 457 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 402 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 423 us MUL_MAT_ID(ffn_moe_down-54): 186 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 397 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 348 us MUL_MAT_ID(ffn_moe_down-56): 221 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 433 us MUL_MAT_ID(ffn_moe_down-57): 188 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 502 us MUL_MAT_ID(ffn_moe_down-58): 241 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 700 us MUL_MAT_ID(ffn_moe_down-59): 375 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 444 us MUL_MAT_ID(ffn_moe_down-60): 554 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 344 us MUL_MAT_ID(ffn_moe_down-61): 200 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 371 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 370 us MUL_MAT_ID(ffn_moe_down-63): 194 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 374 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 396 us MUL_MAT_ID(ffn_moe_down-65): 199 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 357 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 412 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 403 us MUL_MAT_ID(ffn_moe_down-68): 201 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 394 us MUL_MAT_ID(ffn_moe_down-69): 289 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 625 us MUL_MAT_ID(ffn_moe_down-70): 318 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 371 us MUL_MAT_ID(ffn_moe_down-71): 203 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 427 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 343 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 422 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 398 us MUL_MAT_ID(ffn_moe_down-75): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 372 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 419 us MUL_MAT_ID(ffn_moe_down-77): 307 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 465 us MUL_MAT_ID(ffn_moe_down-78): 343 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 414 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 433 us MUL_MAT_ID(ffn_moe_down-80): 194 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 408 us MUL_MAT_ID(ffn_moe_down-81): 220 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 435 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 420 us MUL_MAT_ID(ffn_moe_down-83): 579 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 380 us MUL_MAT_ID(ffn_moe_down-84): 250 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 450 us MUL_MAT_ID(ffn_moe_down-85): 375 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 375 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 398 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 369 us MUL_MAT_ID(ffn_moe_down-88): 202 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 397 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 361 us MUL_MAT_ID(ffn_moe_down-90): 251 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 382 us MUL_MAT_ID(ffn_moe_down-91): 242 us ggml_barrier(...): 48 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 378 us MUL_MAT_ID(ffn_moe_down-25): 425 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 393 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 397 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 350 us MUL_MAT_ID(ffn_moe_down-28): 209 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 398 us MUL_MAT_ID(ffn_moe_down-29): 221 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 396 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 392 us MUL_MAT_ID(ffn_moe_down-31): 177 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 364 us MUL_MAT_ID(ffn_moe_down-32): 187 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 396 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 397 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 339 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 356 us MUL_MAT_ID(ffn_moe_down-36): 222 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 393 us MUL_MAT_ID(ffn_moe_down-37): 238 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 415 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 348 us MUL_MAT_ID(ffn_moe_down-39): 241 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 397 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 378 us MUL_MAT_ID(ffn_moe_down-41): 231 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 412 us MUL_MAT_ID(ffn_moe_down-42): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 402 us MUL_MAT_ID(ffn_moe_down-43): 196 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 351 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 387 us MUL_MAT_ID(ffn_moe_down-45): 224 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 384 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 370 us MUL_MAT_ID(ffn_moe_down-47): 413 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 388 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 408 us MUL_MAT_ID(ffn_moe_down-49): 217 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 437 us MUL_MAT_ID(ffn_moe_down-50): 217 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 451 us MUL_MAT_ID(ffn_moe_down-51): 201 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 361 us MUL_MAT_ID(ffn_moe_down-52): 206 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 397 us MUL_MAT_ID(ffn_moe_down-53): 230 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 426 us MUL_MAT_ID(ffn_moe_down-54): 214 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 484 us MUL_MAT_ID(ffn_moe_down-55): 171 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 191 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 375 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 415 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 429 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 378 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 422 us MUL_MAT_ID(ffn_moe_down-61): 249 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 396 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 350 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 380 us MUL_MAT_ID(ffn_moe_down-64): 507 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 415 us MUL_MAT_ID(ffn_moe_down-65): 205 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 341 us MUL_MAT_ID(ffn_moe_down-66): 187 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 391 us MUL_MAT_ID(ffn_moe_down-67): 192 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 341 us MUL_MAT_ID(ffn_moe_down-68): 251 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 414 us MUL_MAT_ID(ffn_moe_down-69): 216 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 424 us MUL_MAT_ID(ffn_moe_down-70): 197 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 389 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 429 us MUL_MAT_ID(ffn_moe_down-72): 220 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 529 us MUL_MAT_ID(ffn_moe_down-73): 240 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 423 us MUL_MAT_ID(ffn_moe_down-74): 274 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 634 us MUL_MAT_ID(ffn_moe_down-75): 217 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 369 us MUL_MAT_ID(ffn_moe_down-76): 245 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 404 us MUL_MAT_ID(ffn_moe_down-77): 314 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 416 us MUL_MAT_ID(ffn_moe_down-78): 442 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 385 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 391 us MUL_MAT_ID(ffn_moe_down-80): 203 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 419 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 392 us MUL_MAT_ID(ffn_moe_down-82): 203 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 445 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 418 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 420 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 630 us MUL_MAT_ID(ffn_moe_down-86): 324 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 389 us MUL_MAT_ID(ffn_moe_down-87): 198 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 417 us MUL_MAT_ID(ffn_moe_down-88): 212 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 389 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 409 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 421 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 14 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 405 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 400 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 417 us MUL_MAT_ID(ffn_moe_down-27): 208 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 407 us MUL_MAT_ID(ffn_moe_down-28): 207 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 372 us MUL_MAT_ID(ffn_moe_down-29): 232 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 333 us MUL_MAT_ID(ffn_moe_down-31): 192 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 381 us MUL_MAT_ID(ffn_moe_down-32): 207 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 409 us MUL_MAT_ID(ffn_moe_down-33): 211 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 446 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 505 us MUL_MAT_ID(ffn_moe_down-35): 175 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 425 us MUL_MAT_ID(ffn_moe_down-36): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 456 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 377 us MUL_MAT_ID(ffn_moe_down-38): 243 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 418 us MUL_MAT_ID(ffn_moe_down-39): 541 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 389 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 414 us MUL_MAT_ID(ffn_moe_down-41): 411 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 367 us MUL_MAT_ID(ffn_moe_down-42): 190 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 387 us MUL_MAT_ID(ffn_moe_down-43): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 340 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 446 us MUL_MAT_ID(ffn_moe_down-45): 208 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 418 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 432 us MUL_MAT_ID(ffn_moe_down-47): 182 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 436 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 474 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 435 us MUL_MAT_ID(ffn_moe_down-50): 285 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 418 us MUL_MAT_ID(ffn_moe_down-51): 208 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 427 us MUL_MAT_ID(ffn_moe_down-52): 187 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 444 us MUL_MAT_ID(ffn_moe_down-53): 208 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 469 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 411 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 446 us MUL_MAT_ID(ffn_moe_down-56): 217 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 472 us MUL_MAT_ID(ffn_moe_down-57): 204 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 404 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 444 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 417 us MUL_MAT_ID(ffn_moe_down-60): 203 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 449 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 438 us MUL_MAT_ID(ffn_moe_down-62): 214 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 449 us MUL_MAT_ID(ffn_moe_down-63): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 465 us MUL_MAT_ID(ffn_moe_down-64): 201 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 434 us MUL_MAT_ID(ffn_moe_down-65): 196 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 349 us MUL_MAT_ID(ffn_moe_down-67): 270 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 419 us MUL_MAT_ID(ffn_moe_down-68): 349 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 414 us MUL_MAT_ID(ffn_moe_down-69): 204 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 392 us MUL_MAT_ID(ffn_moe_down-70): 213 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 349 us MUL_MAT_ID(ffn_moe_down-71): 206 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 442 us MUL_MAT_ID(ffn_moe_down-72): 187 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 418 us MUL_MAT_ID(ffn_moe_down-73): 196 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 443 us MUL_MAT_ID(ffn_moe_down-74): 223 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 447 us MUL_MAT_ID(ffn_moe_down-75): 273 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 439 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 448 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 381 us MUL_MAT_ID(ffn_moe_down-78): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 415 us MUL_MAT_ID(ffn_moe_down-79): 210 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 440 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 404 us MUL_MAT_ID(ffn_moe_down-81): 187 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 434 us MUL_MAT_ID(ffn_moe_down-82): 592 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 408 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 408 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 442 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 422 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 379 us MUL_MAT_ID(ffn_moe_down-87): 214 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 401 us MUL_MAT_ID(ffn_moe_down-88): 215 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 682 us MUL_MAT_ID(ffn_moe_down-89): 533 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 383 us MUL_MAT_ID(ffn_moe_down-90): 549 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 410 us MUL_MAT_ID(ffn_moe_down-91): 246 us ggml_barrier(...): 15 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 395 us MUL_MAT_ID(ffn_moe_down-25): 240 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 577 us MUL_MAT_ID(ffn_moe_down-26): 166 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 390 us MUL_MAT_ID(ffn_moe_down-27): 224 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 354 us MUL_MAT_ID(ffn_moe_down-28): 184 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 416 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 408 us MUL_MAT_ID(ffn_moe_down-30): 221 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 417 us MUL_MAT_ID(ffn_moe_down-31): 420 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 409 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 431 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 405 us MUL_MAT_ID(ffn_moe_down-34): 203 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 382 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 375 us MUL_MAT_ID(ffn_moe_down-36): 191 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 431 us MUL_MAT_ID(ffn_moe_down-37): 246 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 377 us MUL_MAT_ID(ffn_moe_down-38): 235 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 457 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 398 us MUL_MAT_ID(ffn_moe_down-40): 229 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 415 us MUL_MAT_ID(ffn_moe_down-41): 206 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 511 us MUL_MAT_ID(ffn_moe_down-42): 181 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 436 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 401 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 448 us MUL_MAT_ID(ffn_moe_down-45): 202 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 413 us MUL_MAT_ID(ffn_moe_down-46): 206 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 403 us MUL_MAT_ID(ffn_moe_down-47): 285 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 449 us MUL_MAT_ID(ffn_moe_down-48): 572 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 407 us MUL_MAT_ID(ffn_moe_down-49): 181 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 365 us MUL_MAT_ID(ffn_moe_down-50): 442 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 397 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 1073 us MUL_MAT_ID(ffn_moe_down-52): 385 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 457 us MUL_MAT_ID(ffn_moe_down-53): 195 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 356 us MUL_MAT_ID(ffn_moe_down-54): 448 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 365 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 422 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 387 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 408 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 403 us MUL_MAT_ID(ffn_moe_down-59): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 358 us MUL_MAT_ID(ffn_moe_down-60): 218 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 402 us MUL_MAT_ID(ffn_moe_down-61): 200 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 400 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 366 us MUL_MAT_ID(ffn_moe_down-63): 452 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 369 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 380 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 549 us MUL_MAT_ID(ffn_moe_down-66): 775 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 557 us MUL_MAT_ID(ffn_moe_down-67): 229 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 404 us MUL_MAT_ID(ffn_moe_down-68): 262 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 342 us MUL_MAT_ID(ffn_moe_down-69): 205 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 376 us MUL_MAT_ID(ffn_moe_down-70): 191 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 435 us MUL_MAT_ID(ffn_moe_down-71): 215 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 343 us MUL_MAT_ID(ffn_moe_down-72): 389 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 381 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 416 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 411 us MUL_MAT_ID(ffn_moe_down-76): 206 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 427 us MUL_MAT_ID(ffn_moe_down-77): 433 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 364 us MUL_MAT_ID(ffn_moe_down-78): 369 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 402 us MUL_MAT_ID(ffn_moe_down-79): 202 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 455 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 466 us MUL_MAT_ID(ffn_moe_down-81): 186 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 388 us MUL_MAT_ID(ffn_moe_down-82): 211 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 413 us MUL_MAT_ID(ffn_moe_down-83): 280 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 794 us MUL_MAT_ID(ffn_moe_down-84): 373 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 378 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 389 us MUL_MAT_ID(ffn_moe_down-86): 418 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 400 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 442 us MUL_MAT_ID(ffn_moe_down-88): 199 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 379 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 402 us MUL_MAT_ID(ffn_moe_down-90): 241 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 406 us MUL_MAT_ID(ffn_moe_down-91): 244 us ggml_barrier(...): 41 us GET_ROWS(inp_embd): 22 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 382 us MUL_MAT_ID(ffn_moe_down-25): 215 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 401 us MUL_MAT_ID(ffn_moe_down-26): 199 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 382 us MUL_MAT_ID(ffn_moe_down-27): 204 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 366 us MUL_MAT_ID(ffn_moe_down-28): 188 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 355 us MUL_MAT_ID(ffn_moe_down-29): 209 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 376 us MUL_MAT_ID(ffn_moe_down-30): 183 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 350 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 357 us MUL_MAT_ID(ffn_moe_down-32): 213 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 389 us MUL_MAT_ID(ffn_moe_down-33): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 343 us MUL_MAT_ID(ffn_moe_down-34): 425 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 397 us MUL_MAT_ID(ffn_moe_down-35): 225 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 526 us MUL_MAT_ID(ffn_moe_down-36): 365 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 380 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 403 us MUL_MAT_ID(ffn_moe_down-38): 227 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 387 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 418 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 400 us MUL_MAT_ID(ffn_moe_down-41): 190 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 407 us MUL_MAT_ID(ffn_moe_down-42): 187 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 392 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 651 us MUL_MAT_ID(ffn_moe_down-44): 528 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 517 us MUL_MAT_ID(ffn_moe_down-45): 213 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 382 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 413 us MUL_MAT_ID(ffn_moe_down-47): 203 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 382 us MUL_MAT_ID(ffn_moe_down-48): 202 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 352 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 519 us MUL_MAT_ID(ffn_moe_down-50): 229 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 394 us MUL_MAT_ID(ffn_moe_down-51): 562 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 390 us MUL_MAT_ID(ffn_moe_down-52): 382 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 375 us MUL_MAT_ID(ffn_moe_down-53): 260 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 413 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 392 us MUL_MAT_ID(ffn_moe_down-55): 532 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 385 us MUL_MAT_ID(ffn_moe_down-56): 300 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 387 us MUL_MAT_ID(ffn_moe_down-57): 214 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 431 us MUL_MAT_ID(ffn_moe_down-58): 254 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 644 us MUL_MAT_ID(ffn_moe_down-59): 295 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 375 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 480 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 414 us MUL_MAT_ID(ffn_moe_down-62): 207 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 526 us MUL_MAT_ID(ffn_moe_down-63): 305 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 423 us MUL_MAT_ID(ffn_moe_down-64): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 431 us MUL_MAT_ID(ffn_moe_down-65): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 406 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 347 us MUL_MAT_ID(ffn_moe_down-67): 549 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 387 us MUL_MAT_ID(ffn_moe_down-68): 206 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 410 us MUL_MAT_ID(ffn_moe_down-69): 203 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 341 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 433 us MUL_MAT_ID(ffn_moe_down-71): 234 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 479 us MUL_MAT_ID(ffn_moe_down-72): 1341 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 786 us MUL_MAT_ID(ffn_moe_down-73): 375 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 857 us MUL_MAT_ID(ffn_moe_down-74): 472 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 648 us MUL_MAT_ID(ffn_moe_down-75): 246 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 416 us MUL_MAT_ID(ffn_moe_down-76): 426 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 427 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 494 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 477 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 412 us MUL_MAT_ID(ffn_moe_down-81): 187 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 388 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 404 us MUL_MAT_ID(ffn_moe_down-83): 223 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 456 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 471 us MUL_MAT_ID(ffn_moe_down-85): 538 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 396 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 424 us MUL_MAT_ID(ffn_moe_down-87): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 414 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 356 us MUL_MAT_ID(ffn_moe_down-89): 239 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 435 us MUL_MAT_ID(ffn_moe_down-90): 215 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 400 us MUL_MAT_ID(ffn_moe_down-91): 246 us ggml_barrier(...): 37 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 402 us MUL_MAT_ID(ffn_moe_down-25): 238 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 446 us MUL_MAT_ID(ffn_moe_down-26): 183 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 364 us MUL_MAT_ID(ffn_moe_down-27): 352 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 355 us MUL_MAT_ID(ffn_moe_down-28): 217 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 179 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 570 us MUL_MAT_ID(ffn_moe_down-30): 327 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 359 us MUL_MAT_ID(ffn_moe_down-31): 201 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 387 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 370 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 417 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 435 us MUL_MAT_ID(ffn_moe_down-35): 257 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 425 us MUL_MAT_ID(ffn_moe_down-36): 239 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 407 us MUL_MAT_ID(ffn_moe_down-37): 296 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 418 us MUL_MAT_ID(ffn_moe_down-38): 164 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 396 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 396 us MUL_MAT_ID(ffn_moe_down-40): 188 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 384 us MUL_MAT_ID(ffn_moe_down-41): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 339 us MUL_MAT_ID(ffn_moe_down-42): 188 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 413 us MUL_MAT_ID(ffn_moe_down-43): 517 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 378 us MUL_MAT_ID(ffn_moe_down-44): 221 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 421 us MUL_MAT_ID(ffn_moe_down-45): 556 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 395 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 380 us MUL_MAT_ID(ffn_moe_down-47): 209 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 388 us MUL_MAT_ID(ffn_moe_down-48): 204 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 409 us MUL_MAT_ID(ffn_moe_down-49): 338 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 371 us MUL_MAT_ID(ffn_moe_down-50): 193 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 404 us MUL_MAT_ID(ffn_moe_down-51): 183 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 379 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 376 us MUL_MAT_ID(ffn_moe_down-53): 187 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 407 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 388 us MUL_MAT_ID(ffn_moe_down-55): 310 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 422 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 384 us MUL_MAT_ID(ffn_moe_down-57): 201 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 435 us MUL_MAT_ID(ffn_moe_down-58): 202 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 411 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 365 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 465 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 393 us MUL_MAT_ID(ffn_moe_down-62): 187 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 372 us MUL_MAT_ID(ffn_moe_down-63): 200 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 363 us MUL_MAT_ID(ffn_moe_down-64): 358 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 384 us MUL_MAT_ID(ffn_moe_down-65): 254 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 430 us MUL_MAT_ID(ffn_moe_down-66): 237 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 398 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 408 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 361 us MUL_MAT_ID(ffn_moe_down-69): 207 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 388 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 356 us MUL_MAT_ID(ffn_moe_down-71): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 379 us MUL_MAT_ID(ffn_moe_down-72): 182 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 376 us MUL_MAT_ID(ffn_moe_down-73): 280 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 385 us MUL_MAT_ID(ffn_moe_down-74): 209 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 367 us MUL_MAT_ID(ffn_moe_down-75): 202 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 371 us MUL_MAT_ID(ffn_moe_down-76): 513 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 448 us MUL_MAT_ID(ffn_moe_down-77): 207 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 449 us MUL_MAT_ID(ffn_moe_down-78): 212 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 381 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 379 us MUL_MAT_ID(ffn_moe_down-80): 300 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 478 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 344 us MUL_MAT_ID(ffn_moe_down-82): 179 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 383 us MUL_MAT_ID(ffn_moe_down-83): 258 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 419 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 419 us MUL_MAT_ID(ffn_moe_down-85): 207 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 426 us MUL_MAT_ID(ffn_moe_down-86): 252 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 430 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 390 us MUL_MAT_ID(ffn_moe_down-88): 219 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 433 us MUL_MAT_ID(ffn_moe_down-89): 212 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 408 us MUL_MAT_ID(ffn_moe_down-90): 245 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 413 us MUL_MAT_ID(ffn_moe_down-91): 261 us ggml_barrier(...): 8 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 408 us MUL_MAT_ID(ffn_moe_down-26): 205 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 443 us MUL_MAT_ID(ffn_moe_down-27): 209 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 354 us MUL_MAT_ID(ffn_moe_down-28): 220 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 391 us MUL_MAT_ID(ffn_moe_down-29): 629 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 404 us MUL_MAT_ID(ffn_moe_down-30): 216 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 678 us MUL_MAT_ID(ffn_moe_down-31): 328 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 427 us MUL_MAT_ID(ffn_moe_down-32): 460 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 405 us MUL_MAT_ID(ffn_moe_down-33): 603 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 398 us MUL_MAT_ID(ffn_moe_down-34): 186 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 392 us MUL_MAT_ID(ffn_moe_down-35): 399 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 392 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 576 us MUL_MAT_ID(ffn_moe_down-37): 189 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 378 us MUL_MAT_ID(ffn_moe_down-38): 203 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 352 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 367 us MUL_MAT_ID(ffn_moe_down-40): 266 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 445 us MUL_MAT_ID(ffn_moe_down-41): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 424 us MUL_MAT_ID(ffn_moe_down-42): 488 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 381 us MUL_MAT_ID(ffn_moe_down-43): 217 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 417 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 439 us MUL_MAT_ID(ffn_moe_down-45): 225 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 374 us MUL_MAT_ID(ffn_moe_down-46): 232 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 379 us MUL_MAT_ID(ffn_moe_down-47): 206 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 380 us MUL_MAT_ID(ffn_moe_down-48): 431 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 375 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 406 us MUL_MAT_ID(ffn_moe_down-50): 597 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 388 us MUL_MAT_ID(ffn_moe_down-51): 210 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 423 us MUL_MAT_ID(ffn_moe_down-52): 214 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 377 us MUL_MAT_ID(ffn_moe_down-53): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 395 us MUL_MAT_ID(ffn_moe_down-54): 225 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 433 us MUL_MAT_ID(ffn_moe_down-55): 215 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 423 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 451 us MUL_MAT_ID(ffn_moe_down-57): 196 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 467 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 397 us MUL_MAT_ID(ffn_moe_down-59): 207 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 599 us MUL_MAT_ID(ffn_moe_down-60): 333 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 386 us MUL_MAT_ID(ffn_moe_down-61): 186 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 420 us MUL_MAT_ID(ffn_moe_down-62): 228 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 417 us MUL_MAT_ID(ffn_moe_down-63): 194 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 467 us MUL_MAT_ID(ffn_moe_down-64): 286 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 409 us MUL_MAT_ID(ffn_moe_down-65): 200 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 398 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 418 us MUL_MAT_ID(ffn_moe_down-67): 191 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 411 us MUL_MAT_ID(ffn_moe_down-68): 190 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 416 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 376 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 407 us MUL_MAT_ID(ffn_moe_down-71): 529 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 351 us MUL_MAT_ID(ffn_moe_down-72): 234 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 428 us MUL_MAT_ID(ffn_moe_down-73): 207 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 391 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 375 us MUL_MAT_ID(ffn_moe_down-75): 190 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 455 us MUL_MAT_ID(ffn_moe_down-76): 185 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 373 us MUL_MAT_ID(ffn_moe_down-77): 318 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 376 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 405 us MUL_MAT_ID(ffn_moe_down-79): 187 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 410 us MUL_MAT_ID(ffn_moe_down-80): 232 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 396 us MUL_MAT_ID(ffn_moe_down-81): 184 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 384 us MUL_MAT_ID(ffn_moe_down-82): 606 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 412 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 413 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 404 us MUL_MAT_ID(ffn_moe_down-85): 463 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 391 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 451 us MUL_MAT_ID(ffn_moe_down-87): 181 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 442 us MUL_MAT_ID(ffn_moe_down-88): 216 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 374 us MUL_MAT_ID(ffn_moe_down-89): 285 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 540 us MUL_MAT_ID(ffn_moe_down-90): 246 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 390 us MUL_MAT_ID(ffn_moe_down-91): 241 us ggml_barrier(...): 173 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 354 us MUL_MAT_ID(ffn_moe_down-25): 569 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 351 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 361 us MUL_MAT_ID(ffn_moe_down-27): 206 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 386 us MUL_MAT_ID(ffn_moe_down-28): 398 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 377 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 438 us MUL_MAT_ID(ffn_moe_down-30): 537 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 416 us MUL_MAT_ID(ffn_moe_down-31): 204 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 531 us MUL_MAT_ID(ffn_moe_down-32): 438 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 400 us MUL_MAT_ID(ffn_moe_down-33): 197 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 408 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 414 us MUL_MAT_ID(ffn_moe_down-35): 189 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 357 us MUL_MAT_ID(ffn_moe_down-36): 414 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 381 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 430 us MUL_MAT_ID(ffn_moe_down-38): 228 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 425 us MUL_MAT_ID(ffn_moe_down-39): 256 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 527 us MUL_MAT_ID(ffn_moe_down-40): 687 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 719 us MUL_MAT_ID(ffn_moe_down-41): 523 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 403 us MUL_MAT_ID(ffn_moe_down-42): 207 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 405 us MUL_MAT_ID(ffn_moe_down-43): 338 us ggml_barrier(...): 371 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 376 us MUL_MAT_ID(ffn_moe_down-44): 267 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 387 us MUL_MAT_ID(ffn_moe_down-45): 208 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 417 us MUL_MAT_ID(ffn_moe_down-46): 401 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 412 us MUL_MAT_ID(ffn_moe_down-47): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 412 us MUL_MAT_ID(ffn_moe_down-48): 581 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 413 us MUL_MAT_ID(ffn_moe_down-49): 204 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 411 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 527 us MUL_MAT_ID(ffn_moe_down-51): 326 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 362 us MUL_MAT_ID(ffn_moe_down-52): 201 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 384 us MUL_MAT_ID(ffn_moe_down-53): 182 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 395 us MUL_MAT_ID(ffn_moe_down-54): 219 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 204 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 517 us MUL_MAT_ID(ffn_moe_down-56): 182 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 400 us MUL_MAT_ID(ffn_moe_down-57): 200 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 427 us MUL_MAT_ID(ffn_moe_down-58): 242 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 378 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 369 us MUL_MAT_ID(ffn_moe_down-60): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 412 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 399 us MUL_MAT_ID(ffn_moe_down-62): 223 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 340 us MUL_MAT_ID(ffn_moe_down-63): 210 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 373 us MUL_MAT_ID(ffn_moe_down-64): 199 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 356 us MUL_MAT_ID(ffn_moe_down-65): 196 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 377 us MUL_MAT_ID(ffn_moe_down-66): 197 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 352 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 370 us MUL_MAT_ID(ffn_moe_down-68): 207 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 349 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 387 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 405 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 430 us MUL_MAT_ID(ffn_moe_down-72): 309 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 410 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 343 us MUL_MAT_ID(ffn_moe_down-74): 213 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 402 us MUL_MAT_ID(ffn_moe_down-75): 230 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 416 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 399 us MUL_MAT_ID(ffn_moe_down-77): 258 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 441 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 360 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 398 us MUL_MAT_ID(ffn_moe_down-80): 460 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 370 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 402 us MUL_MAT_ID(ffn_moe_down-82): 497 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 398 us MUL_MAT_ID(ffn_moe_down-83): 212 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 353 us MUL_MAT_ID(ffn_moe_down-84): 723 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 361 us MUL_MAT_ID(ffn_moe_down-85): 243 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 416 us MUL_MAT_ID(ffn_moe_down-86): 576 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 388 us MUL_MAT_ID(ffn_moe_down-87): 227 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 476 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 463 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 421 us MUL_MAT_ID(ffn_moe_down-90): 224 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 411 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 156 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 389 us MUL_MAT_ID(ffn_moe_down-25): 207 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 434 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 415 us MUL_MAT_ID(ffn_moe_down-27): 315 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 385 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 522 us MUL_MAT_ID(ffn_moe_down-29): 197 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 406 us MUL_MAT_ID(ffn_moe_down-30): 184 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 395 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 394 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 369 us MUL_MAT_ID(ffn_moe_down-33): 181 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 368 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 463 us MUL_MAT_ID(ffn_moe_down-35): 208 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 442 us MUL_MAT_ID(ffn_moe_down-36): 303 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 393 us MUL_MAT_ID(ffn_moe_down-37): 345 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 390 us MUL_MAT_ID(ffn_moe_down-38): 571 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 428 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 435 us MUL_MAT_ID(ffn_moe_down-40): 221 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 356 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 432 us MUL_MAT_ID(ffn_moe_down-42): 223 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 441 us MUL_MAT_ID(ffn_moe_down-43): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 363 us MUL_MAT_ID(ffn_moe_down-44): 185 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 378 us MUL_MAT_ID(ffn_moe_down-45): 436 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 406 us MUL_MAT_ID(ffn_moe_down-46): 368 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 468 us MUL_MAT_ID(ffn_moe_down-47): 183 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 451 us MUL_MAT_ID(ffn_moe_down-48): 289 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 465 us MUL_MAT_ID(ffn_moe_down-49): 234 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 430 us MUL_MAT_ID(ffn_moe_down-50): 231 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 648 us MUL_MAT_ID(ffn_moe_down-51): 251 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 395 us MUL_MAT_ID(ffn_moe_down-52): 260 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 418 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 429 us MUL_MAT_ID(ffn_moe_down-54): 201 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 688 us MUL_MAT_ID(ffn_moe_down-55): 329 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 405 us MUL_MAT_ID(ffn_moe_down-56): 218 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 399 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 380 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 527 us MUL_MAT_ID(ffn_moe_down-59): 177 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 382 us MUL_MAT_ID(ffn_moe_down-60): 208 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 441 us MUL_MAT_ID(ffn_moe_down-61): 219 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 423 us MUL_MAT_ID(ffn_moe_down-62): 318 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 377 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 495 us MUL_MAT_ID(ffn_moe_down-64): 181 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 398 us MUL_MAT_ID(ffn_moe_down-65): 405 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 298 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 410 us MUL_MAT_ID(ffn_moe_down-67): 215 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 473 us MUL_MAT_ID(ffn_moe_down-68): 291 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 378 us MUL_MAT_ID(ffn_moe_down-69): 223 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 427 us MUL_MAT_ID(ffn_moe_down-70): 277 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 424 us MUL_MAT_ID(ffn_moe_down-71): 182 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 430 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 414 us MUL_MAT_ID(ffn_moe_down-73): 193 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 440 us MUL_MAT_ID(ffn_moe_down-74): 249 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 559 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 407 us MUL_MAT_ID(ffn_moe_down-76): 196 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 466 us MUL_MAT_ID(ffn_moe_down-77): 235 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 435 us MUL_MAT_ID(ffn_moe_down-78): 339 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 426 us MUL_MAT_ID(ffn_moe_down-79): 202 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 398 us MUL_MAT_ID(ffn_moe_down-80): 208 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 404 us MUL_MAT_ID(ffn_moe_down-81): 370 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 409 us MUL_MAT_ID(ffn_moe_down-82): 210 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 463 us MUL_MAT_ID(ffn_moe_down-83): 277 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 387 us MUL_MAT_ID(ffn_moe_down-84): 240 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 383 us MUL_MAT_ID(ffn_moe_down-85): 245 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 411 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 394 us MUL_MAT_ID(ffn_moe_down-87): 175 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 360 us MUL_MAT_ID(ffn_moe_down-88): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 527 us MUL_MAT_ID(ffn_moe_down-89): 216 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 396 us MUL_MAT_ID(ffn_moe_down-90): 650 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 396 us MUL_MAT_ID(ffn_moe_down-91): 634 us ggml_barrier(...): 13 us GET_ROWS(inp_embd): 23 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 405 us MUL_MAT_ID(ffn_moe_down-25): 228 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 652 us MUL_MAT_ID(ffn_moe_down-26): 242 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 401 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 468 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 415 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 430 us MUL_MAT_ID(ffn_moe_down-30): 189 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 421 us MUL_MAT_ID(ffn_moe_down-31): 208 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 391 us MUL_MAT_ID(ffn_moe_down-32): 215 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 424 us MUL_MAT_ID(ffn_moe_down-33): 273 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 420 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 605 us MUL_MAT_ID(ffn_moe_down-35): 363 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 416 us MUL_MAT_ID(ffn_moe_down-36): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 443 us MUL_MAT_ID(ffn_moe_down-37): 249 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 385 us MUL_MAT_ID(ffn_moe_down-38): 191 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 416 us MUL_MAT_ID(ffn_moe_down-39): 253 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 416 us MUL_MAT_ID(ffn_moe_down-40): 525 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 426 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 369 us MUL_MAT_ID(ffn_moe_down-42): 215 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 432 us MUL_MAT_ID(ffn_moe_down-43): 208 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 443 us MUL_MAT_ID(ffn_moe_down-44): 411 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 400 us MUL_MAT_ID(ffn_moe_down-45): 587 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 364 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 358 us MUL_MAT_ID(ffn_moe_down-47): 188 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 444 us MUL_MAT_ID(ffn_moe_down-48): 204 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 414 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 426 us MUL_MAT_ID(ffn_moe_down-50): 217 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 377 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 419 us MUL_MAT_ID(ffn_moe_down-52): 379 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 384 us MUL_MAT_ID(ffn_moe_down-53): 185 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 371 us MUL_MAT_ID(ffn_moe_down-54): 186 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 422 us MUL_MAT_ID(ffn_moe_down-55): 262 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 641 us MUL_MAT_ID(ffn_moe_down-56): 359 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 462 us MUL_MAT_ID(ffn_moe_down-57): 249 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 666 us MUL_MAT_ID(ffn_moe_down-58): 188 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 399 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 439 us MUL_MAT_ID(ffn_moe_down-60): 183 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 387 us MUL_MAT_ID(ffn_moe_down-61): 181 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 395 us MUL_MAT_ID(ffn_moe_down-62): 218 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 422 us MUL_MAT_ID(ffn_moe_down-63): 235 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 419 us MUL_MAT_ID(ffn_moe_down-64): 213 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 406 us MUL_MAT_ID(ffn_moe_down-65): 287 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 379 us MUL_MAT_ID(ffn_moe_down-66): 227 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 402 us MUL_MAT_ID(ffn_moe_down-67): 184 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 418 us MUL_MAT_ID(ffn_moe_down-68): 206 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 387 us MUL_MAT_ID(ffn_moe_down-69): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 366 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 407 us MUL_MAT_ID(ffn_moe_down-71): 208 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 368 us MUL_MAT_ID(ffn_moe_down-72): 292 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 438 us MUL_MAT_ID(ffn_moe_down-73): 362 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 371 us MUL_MAT_ID(ffn_moe_down-74): 192 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 375 us MUL_MAT_ID(ffn_moe_down-75): 366 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 396 us MUL_MAT_ID(ffn_moe_down-76): 215 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 411 us MUL_MAT_ID(ffn_moe_down-77): 189 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 405 us MUL_MAT_ID(ffn_moe_down-78): 191 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 387 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 368 us MUL_MAT_ID(ffn_moe_down-80): 228 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 406 us MUL_MAT_ID(ffn_moe_down-81): 169 us ggml_barrier(...): 336 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 377 us MUL_MAT_ID(ffn_moe_down-82): 196 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 376 us MUL_MAT_ID(ffn_moe_down-83): 185 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 378 us MUL_MAT_ID(ffn_moe_down-84): 240 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 399 us MUL_MAT_ID(ffn_moe_down-85): 509 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 644 us MUL_MAT_ID(ffn_moe_down-86): 332 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 377 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 389 us MUL_MAT_ID(ffn_moe_down-88): 225 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 390 us MUL_MAT_ID(ffn_moe_down-89): 195 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 393 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 440 us MUL_MAT_ID(ffn_moe_down-91): 241 us ggml_barrier(...): 8 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 368 us MUL_MAT_ID(ffn_moe_down-25): 213 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 419 us MUL_MAT_ID(ffn_moe_down-26): 206 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 484 us MUL_MAT_ID(ffn_moe_down-27): 294 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 380 us MUL_MAT_ID(ffn_moe_down-28): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 354 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 400 us MUL_MAT_ID(ffn_moe_down-30): 234 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 421 us MUL_MAT_ID(ffn_moe_down-31): 575 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 420 us MUL_MAT_ID(ffn_moe_down-32): 249 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 377 us MUL_MAT_ID(ffn_moe_down-33): 209 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 611 us MUL_MAT_ID(ffn_moe_down-34): 340 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 366 us MUL_MAT_ID(ffn_moe_down-35): 339 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 409 us MUL_MAT_ID(ffn_moe_down-36): 215 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 411 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 399 us MUL_MAT_ID(ffn_moe_down-38): 571 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 406 us MUL_MAT_ID(ffn_moe_down-39): 257 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 431 us MUL_MAT_ID(ffn_moe_down-40): 522 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 394 us MUL_MAT_ID(ffn_moe_down-41): 300 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 437 us MUL_MAT_ID(ffn_moe_down-42): 401 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 367 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 381 us MUL_MAT_ID(ffn_moe_down-44): 220 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 411 us MUL_MAT_ID(ffn_moe_down-45): 204 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 473 us MUL_MAT_ID(ffn_moe_down-46): 182 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 437 us MUL_MAT_ID(ffn_moe_down-47): 225 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 438 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 477 us MUL_MAT_ID(ffn_moe_down-49): 270 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 427 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 428 us MUL_MAT_ID(ffn_moe_down-51): 322 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 429 us MUL_MAT_ID(ffn_moe_down-52): 249 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 409 us MUL_MAT_ID(ffn_moe_down-53): 190 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 489 us MUL_MAT_ID(ffn_moe_down-54): 435 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 422 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 399 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 437 us MUL_MAT_ID(ffn_moe_down-57): 204 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 469 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 605 us MUL_MAT_ID(ffn_moe_down-59): 460 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 403 us MUL_MAT_ID(ffn_moe_down-60): 203 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 388 us MUL_MAT_ID(ffn_moe_down-61): 560 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 637 us MUL_MAT_ID(ffn_moe_down-62): 224 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 394 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 469 us MUL_MAT_ID(ffn_moe_down-64): 180 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 417 us MUL_MAT_ID(ffn_moe_down-65): 208 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 396 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 406 us MUL_MAT_ID(ffn_moe_down-67): 196 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 418 us MUL_MAT_ID(ffn_moe_down-68): 211 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 833 us MUL_MAT_ID(ffn_moe_down-69): 337 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 351 us MUL_MAT_ID(ffn_moe_down-70): 213 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 394 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 355 us MUL_MAT_ID(ffn_moe_down-72): 273 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 370 us MUL_MAT_ID(ffn_moe_down-73): 244 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 555 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 399 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 421 us MUL_MAT_ID(ffn_moe_down-76): 239 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 379 us MUL_MAT_ID(ffn_moe_down-77): 197 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 391 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 357 us MUL_MAT_ID(ffn_moe_down-79): 168 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 496 us MUL_MAT_ID(ffn_moe_down-80): 325 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 360 us MUL_MAT_ID(ffn_moe_down-81): 180 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 400 us MUL_MAT_ID(ffn_moe_down-82): 311 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 441 us MUL_MAT_ID(ffn_moe_down-83): 382 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 404 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 528 us MUL_MAT_ID(ffn_moe_down-85): 791 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 413 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 458 us MUL_MAT_ID(ffn_moe_down-87): 178 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 580 us MUL_MAT_ID(ffn_moe_down-88): 327 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 367 us MUL_MAT_ID(ffn_moe_down-89): 212 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 378 us MUL_MAT_ID(ffn_moe_down-90): 249 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 374 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 141 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 400 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 381 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 341 us MUL_MAT_ID(ffn_moe_down-27): 209 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 376 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 357 us MUL_MAT_ID(ffn_moe_down-29): 196 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 405 us MUL_MAT_ID(ffn_moe_down-30): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 376 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 355 us MUL_MAT_ID(ffn_moe_down-32): 290 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 435 us MUL_MAT_ID(ffn_moe_down-33): 211 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 357 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 433 us MUL_MAT_ID(ffn_moe_down-36): 206 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 410 us MUL_MAT_ID(ffn_moe_down-37): 350 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 459 us MUL_MAT_ID(ffn_moe_down-38): 206 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 416 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 386 us MUL_MAT_ID(ffn_moe_down-40): 225 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 400 us MUL_MAT_ID(ffn_moe_down-41): 222 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 407 us MUL_MAT_ID(ffn_moe_down-42): 205 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 397 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 387 us MUL_MAT_ID(ffn_moe_down-44): 198 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 350 us MUL_MAT_ID(ffn_moe_down-45): 213 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 381 us MUL_MAT_ID(ffn_moe_down-46): 223 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 447 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 370 us MUL_MAT_ID(ffn_moe_down-48): 197 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 382 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 391 us MUL_MAT_ID(ffn_moe_down-50): 264 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 424 us MUL_MAT_ID(ffn_moe_down-51): 205 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 396 us MUL_MAT_ID(ffn_moe_down-52): 561 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 435 us MUL_MAT_ID(ffn_moe_down-53): 328 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 429 us MUL_MAT_ID(ffn_moe_down-54): 340 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 415 us MUL_MAT_ID(ffn_moe_down-55): 282 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 427 us MUL_MAT_ID(ffn_moe_down-56): 199 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 421 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 435 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 398 us MUL_MAT_ID(ffn_moe_down-59): 206 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 426 us MUL_MAT_ID(ffn_moe_down-60): 204 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 528 us MUL_MAT_ID(ffn_moe_down-61): 176 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 411 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 398 us MUL_MAT_ID(ffn_moe_down-63): 211 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 400 us MUL_MAT_ID(ffn_moe_down-64): 217 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 392 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 215 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 410 us MUL_MAT_ID(ffn_moe_down-67): 216 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 418 us MUL_MAT_ID(ffn_moe_down-68): 334 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 476 us MUL_MAT_ID(ffn_moe_down-69): 204 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 360 us MUL_MAT_ID(ffn_moe_down-70): 189 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 397 us MUL_MAT_ID(ffn_moe_down-71): 225 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 366 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 360 us MUL_MAT_ID(ffn_moe_down-73): 220 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 447 us MUL_MAT_ID(ffn_moe_down-74): 233 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 621 us MUL_MAT_ID(ffn_moe_down-75): 331 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 381 us MUL_MAT_ID(ffn_moe_down-76): 272 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 417 us MUL_MAT_ID(ffn_moe_down-77): 534 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 410 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 429 us MUL_MAT_ID(ffn_moe_down-79): 346 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 445 us MUL_MAT_ID(ffn_moe_down-80): 246 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 554 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 433 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 431 us MUL_MAT_ID(ffn_moe_down-83): 188 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 366 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 464 us MUL_MAT_ID(ffn_moe_down-85): 246 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 402 us MUL_MAT_ID(ffn_moe_down-86): 285 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 435 us MUL_MAT_ID(ffn_moe_down-87): 234 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 383 us MUL_MAT_ID(ffn_moe_down-88): 443 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 667 us MUL_MAT_ID(ffn_moe_down-89): 230 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 412 us MUL_MAT_ID(ffn_moe_down-90): 768 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 502 us MUL_MAT_ID(ffn_moe_down-91): 360 us ggml_barrier(...): 3 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 396 us MUL_MAT_ID(ffn_moe_down-25): 203 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 421 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 444 us MUL_MAT_ID(ffn_moe_down-27): 287 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 379 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 403 us MUL_MAT_ID(ffn_moe_down-29): 184 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 533 us MUL_MAT_ID(ffn_moe_down-30): 320 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 374 us MUL_MAT_ID(ffn_moe_down-31): 490 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 415 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 349 us MUL_MAT_ID(ffn_moe_down-33): 216 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 388 us MUL_MAT_ID(ffn_moe_down-34): 205 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 357 us MUL_MAT_ID(ffn_moe_down-35): 211 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 392 us MUL_MAT_ID(ffn_moe_down-36): 185 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 384 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 356 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 399 us MUL_MAT_ID(ffn_moe_down-39): 251 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 407 us MUL_MAT_ID(ffn_moe_down-40): 226 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 363 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 378 us MUL_MAT_ID(ffn_moe_down-42): 190 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 445 us MUL_MAT_ID(ffn_moe_down-43): 199 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 385 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 390 us MUL_MAT_ID(ffn_moe_down-45): 229 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 471 us MUL_MAT_ID(ffn_moe_down-46): 343 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 400 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 370 us MUL_MAT_ID(ffn_moe_down-48): 392 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 372 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 446 us MUL_MAT_ID(ffn_moe_down-50): 530 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 358 us MUL_MAT_ID(ffn_moe_down-51): 206 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 365 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 345 us MUL_MAT_ID(ffn_moe_down-53): 268 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 395 us MUL_MAT_ID(ffn_moe_down-54): 191 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 429 us MUL_MAT_ID(ffn_moe_down-55): 186 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 410 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 402 us MUL_MAT_ID(ffn_moe_down-57): 203 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 438 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 379 us MUL_MAT_ID(ffn_moe_down-59): 420 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 376 us MUL_MAT_ID(ffn_moe_down-60): 202 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 419 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 436 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 402 us MUL_MAT_ID(ffn_moe_down-63): 525 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 385 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 398 us MUL_MAT_ID(ffn_moe_down-65): 542 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 416 us MUL_MAT_ID(ffn_moe_down-66): 205 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 418 us MUL_MAT_ID(ffn_moe_down-67): 228 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 441 us MUL_MAT_ID(ffn_moe_down-68): 349 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 392 us MUL_MAT_ID(ffn_moe_down-69): 213 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 426 us MUL_MAT_ID(ffn_moe_down-70): 205 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 429 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 411 us MUL_MAT_ID(ffn_moe_down-72): 203 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 429 us MUL_MAT_ID(ffn_moe_down-73): 533 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 397 us MUL_MAT_ID(ffn_moe_down-74): 254 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 452 us MUL_MAT_ID(ffn_moe_down-75): 305 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 469 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 415 us MUL_MAT_ID(ffn_moe_down-77): 367 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 416 us MUL_MAT_ID(ffn_moe_down-78): 316 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 408 us MUL_MAT_ID(ffn_moe_down-79): 229 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 395 us MUL_MAT_ID(ffn_moe_down-80): 202 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 441 us MUL_MAT_ID(ffn_moe_down-81): 186 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 387 us MUL_MAT_ID(ffn_moe_down-82): 333 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 454 us MUL_MAT_ID(ffn_moe_down-83): 477 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 383 us MUL_MAT_ID(ffn_moe_down-84): 245 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 406 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 394 us MUL_MAT_ID(ffn_moe_down-86): 397 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 437 us MUL_MAT_ID(ffn_moe_down-87): 345 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 445 us MUL_MAT_ID(ffn_moe_down-88): 697 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 426 us MUL_MAT_ID(ffn_moe_down-89): 485 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 416 us MUL_MAT_ID(ffn_moe_down-90): 514 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 527 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 13 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 359 us MUL_MAT_ID(ffn_moe_down-25): 351 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 381 us MUL_MAT_ID(ffn_moe_down-26): 247 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 372 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 412 us MUL_MAT_ID(ffn_moe_down-28): 208 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 403 us MUL_MAT_ID(ffn_moe_down-29): 346 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 384 us MUL_MAT_ID(ffn_moe_down-30): 203 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 410 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 411 us MUL_MAT_ID(ffn_moe_down-32): 212 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 458 us MUL_MAT_ID(ffn_moe_down-33): 217 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 445 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 364 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 379 us MUL_MAT_ID(ffn_moe_down-36): 220 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 361 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 390 us MUL_MAT_ID(ffn_moe_down-38): 404 us ggml_barrier(...): 367 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 388 us MUL_MAT_ID(ffn_moe_down-39): 242 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 399 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 400 us MUL_MAT_ID(ffn_moe_down-41): 545 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 414 us MUL_MAT_ID(ffn_moe_down-42): 219 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 425 us MUL_MAT_ID(ffn_moe_down-43): 243 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 434 us MUL_MAT_ID(ffn_moe_down-44): 568 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 417 us MUL_MAT_ID(ffn_moe_down-45): 201 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 402 us MUL_MAT_ID(ffn_moe_down-46): 221 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 409 us MUL_MAT_ID(ffn_moe_down-47): 197 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 386 us MUL_MAT_ID(ffn_moe_down-48): 471 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 377 us MUL_MAT_ID(ffn_moe_down-49): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 367 us MUL_MAT_ID(ffn_moe_down-50): 208 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 385 us MUL_MAT_ID(ffn_moe_down-51): 191 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 351 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 404 us MUL_MAT_ID(ffn_moe_down-53): 198 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 378 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 394 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 386 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 392 us MUL_MAT_ID(ffn_moe_down-57): 207 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 410 us MUL_MAT_ID(ffn_moe_down-58): 219 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 427 us MUL_MAT_ID(ffn_moe_down-59): 211 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 374 us MUL_MAT_ID(ffn_moe_down-60): 221 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 393 us MUL_MAT_ID(ffn_moe_down-61): 213 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 382 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 367 us MUL_MAT_ID(ffn_moe_down-63): 195 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 371 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 335 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 397 us MUL_MAT_ID(ffn_moe_down-66): 272 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 465 us MUL_MAT_ID(ffn_moe_down-67): 279 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 408 us MUL_MAT_ID(ffn_moe_down-68): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 374 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 353 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 412 us MUL_MAT_ID(ffn_moe_down-71): 209 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 421 us MUL_MAT_ID(ffn_moe_down-72): 182 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 412 us MUL_MAT_ID(ffn_moe_down-73): 229 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 404 us MUL_MAT_ID(ffn_moe_down-74): 208 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 423 us MUL_MAT_ID(ffn_moe_down-75): 196 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 370 us MUL_MAT_ID(ffn_moe_down-76): 274 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 437 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 382 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 372 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 394 us MUL_MAT_ID(ffn_moe_down-80): 190 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 364 us MUL_MAT_ID(ffn_moe_down-81): 316 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 429 us MUL_MAT_ID(ffn_moe_down-82): 183 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 354 us MUL_MAT_ID(ffn_moe_down-83): 190 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 403 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 406 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 419 us MUL_MAT_ID(ffn_moe_down-86): 390 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 390 us MUL_MAT_ID(ffn_moe_down-87): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 345 us MUL_MAT_ID(ffn_moe_down-88): 189 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 388 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 369 us MUL_MAT_ID(ffn_moe_down-90): 229 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 395 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 42 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 419 us MUL_MAT_ID(ffn_moe_down-25): 211 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 448 us MUL_MAT_ID(ffn_moe_down-26): 200 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 389 us MUL_MAT_ID(ffn_moe_down-27): 198 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 461 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 436 us MUL_MAT_ID(ffn_moe_down-29): 353 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 444 us MUL_MAT_ID(ffn_moe_down-30): 216 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 421 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 385 us MUL_MAT_ID(ffn_moe_down-32): 205 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 452 us MUL_MAT_ID(ffn_moe_down-33): 218 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 401 us MUL_MAT_ID(ffn_moe_down-34): 229 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 447 us MUL_MAT_ID(ffn_moe_down-35): 200 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 384 us MUL_MAT_ID(ffn_moe_down-36): 195 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 358 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 349 us MUL_MAT_ID(ffn_moe_down-38): 371 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 381 us MUL_MAT_ID(ffn_moe_down-39): 253 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 500 us MUL_MAT_ID(ffn_moe_down-40): 265 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 399 us MUL_MAT_ID(ffn_moe_down-41): 198 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 423 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 380 us MUL_MAT_ID(ffn_moe_down-43): 198 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 424 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 400 us MUL_MAT_ID(ffn_moe_down-45): 217 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 421 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 402 us MUL_MAT_ID(ffn_moe_down-47): 212 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 437 us MUL_MAT_ID(ffn_moe_down-48): 211 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 359 us MUL_MAT_ID(ffn_moe_down-49): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 407 us MUL_MAT_ID(ffn_moe_down-50): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 363 us MUL_MAT_ID(ffn_moe_down-51): 203 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 393 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 665 us MUL_MAT_ID(ffn_moe_down-53): 250 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 337 us MUL_MAT_ID(ffn_moe_down-54): 187 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 382 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 363 us MUL_MAT_ID(ffn_moe_down-56): 205 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 379 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 397 us MUL_MAT_ID(ffn_moe_down-58): 202 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 405 us MUL_MAT_ID(ffn_moe_down-59): 206 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 390 us MUL_MAT_ID(ffn_moe_down-60): 186 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 339 us MUL_MAT_ID(ffn_moe_down-61): 187 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 379 us MUL_MAT_ID(ffn_moe_down-62): 184 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 358 us MUL_MAT_ID(ffn_moe_down-63): 204 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 348 us MUL_MAT_ID(ffn_moe_down-64): 392 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 408 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 389 us MUL_MAT_ID(ffn_moe_down-67): 216 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 363 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 378 us MUL_MAT_ID(ffn_moe_down-69): 187 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 389 us MUL_MAT_ID(ffn_moe_down-70): 585 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 403 us MUL_MAT_ID(ffn_moe_down-71): 189 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 407 us MUL_MAT_ID(ffn_moe_down-72): 362 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 381 us MUL_MAT_ID(ffn_moe_down-73): 184 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 362 us MUL_MAT_ID(ffn_moe_down-74): 586 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 356 us MUL_MAT_ID(ffn_moe_down-75): 188 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 381 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 351 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 407 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 391 us MUL_MAT_ID(ffn_moe_down-79): 185 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 348 us MUL_MAT_ID(ffn_moe_down-80): 211 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 335 us MUL_MAT_ID(ffn_moe_down-81): 216 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 383 us MUL_MAT_ID(ffn_moe_down-82): 424 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 398 us MUL_MAT_ID(ffn_moe_down-83): 219 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 357 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 418 us MUL_MAT_ID(ffn_moe_down-85): 246 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 397 us MUL_MAT_ID(ffn_moe_down-86): 202 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 390 us MUL_MAT_ID(ffn_moe_down-87): 254 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 629 us MUL_MAT_ID(ffn_moe_down-88): 214 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 380 us MUL_MAT_ID(ffn_moe_down-89): 181 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 381 us MUL_MAT_ID(ffn_moe_down-90): 834 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 384 us MUL_MAT_ID(ffn_moe_down-91): 233 us ggml_barrier(...): 84 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 372 us MUL_MAT_ID(ffn_moe_down-25): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 443 us MUL_MAT_ID(ffn_moe_down-26): 221 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 404 us MUL_MAT_ID(ffn_moe_down-27): 500 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 413 us MUL_MAT_ID(ffn_moe_down-28): 372 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 369 us MUL_MAT_ID(ffn_moe_down-29): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 408 us MUL_MAT_ID(ffn_moe_down-30): 184 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 381 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 382 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 393 us MUL_MAT_ID(ffn_moe_down-33): 230 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 406 us MUL_MAT_ID(ffn_moe_down-34): 187 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 409 us MUL_MAT_ID(ffn_moe_down-35): 207 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 371 us MUL_MAT_ID(ffn_moe_down-36): 184 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 485 us MUL_MAT_ID(ffn_moe_down-37): 262 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 459 us MUL_MAT_ID(ffn_moe_down-38): 513 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 520 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 422 us MUL_MAT_ID(ffn_moe_down-40): 218 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 415 us MUL_MAT_ID(ffn_moe_down-41): 596 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 374 us MUL_MAT_ID(ffn_moe_down-42): 217 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 379 us MUL_MAT_ID(ffn_moe_down-43): 214 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 407 us MUL_MAT_ID(ffn_moe_down-44): 556 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 335 us MUL_MAT_ID(ffn_moe_down-45): 204 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 419 us MUL_MAT_ID(ffn_moe_down-46): 183 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 357 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 378 us MUL_MAT_ID(ffn_moe_down-48): 194 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 436 us MUL_MAT_ID(ffn_moe_down-49): 232 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 678 us MUL_MAT_ID(ffn_moe_down-50): 272 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 399 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 393 us MUL_MAT_ID(ffn_moe_down-52): 182 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 433 us MUL_MAT_ID(ffn_moe_down-53): 382 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 405 us MUL_MAT_ID(ffn_moe_down-54): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 449 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 438 us MUL_MAT_ID(ffn_moe_down-56): 469 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 438 us MUL_MAT_ID(ffn_moe_down-57): 204 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 364 us MUL_MAT_ID(ffn_moe_down-58): 235 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 479 us MUL_MAT_ID(ffn_moe_down-59): 202 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 389 us MUL_MAT_ID(ffn_moe_down-60): 186 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 388 us MUL_MAT_ID(ffn_moe_down-61): 466 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 360 us MUL_MAT_ID(ffn_moe_down-62): 451 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 385 us MUL_MAT_ID(ffn_moe_down-63): 198 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 424 us MUL_MAT_ID(ffn_moe_down-64): 222 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 450 us MUL_MAT_ID(ffn_moe_down-65): 232 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 442 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 410 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 386 us MUL_MAT_ID(ffn_moe_down-68): 263 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 450 us MUL_MAT_ID(ffn_moe_down-69): 181 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 392 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 437 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 427 us MUL_MAT_ID(ffn_moe_down-72): 506 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 403 us MUL_MAT_ID(ffn_moe_down-73): 394 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 382 us MUL_MAT_ID(ffn_moe_down-74): 572 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 390 us MUL_MAT_ID(ffn_moe_down-75): 189 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 411 us MUL_MAT_ID(ffn_moe_down-76): 172 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 430 us MUL_MAT_ID(ffn_moe_down-78): 188 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 408 us MUL_MAT_ID(ffn_moe_down-79): 189 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 391 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 365 us MUL_MAT_ID(ffn_moe_down-81): 185 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 362 us MUL_MAT_ID(ffn_moe_down-82): 183 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 372 us MUL_MAT_ID(ffn_moe_down-83): 217 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 382 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 412 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 503 us MUL_MAT_ID(ffn_moe_down-86): 168 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 452 us MUL_MAT_ID(ffn_moe_down-87): 200 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 396 us MUL_MAT_ID(ffn_moe_down-88): 349 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 379 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 354 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 415 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 41 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 370 us MUL_MAT_ID(ffn_moe_down-25): 213 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 382 us MUL_MAT_ID(ffn_moe_down-26): 327 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 450 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 388 us MUL_MAT_ID(ffn_moe_down-28): 196 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 418 us MUL_MAT_ID(ffn_moe_down-29): 188 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 344 us MUL_MAT_ID(ffn_moe_down-30): 189 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 355 us MUL_MAT_ID(ffn_moe_down-31): 178 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 450 us MUL_MAT_ID(ffn_moe_down-32): 203 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 411 us MUL_MAT_ID(ffn_moe_down-33): 217 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 407 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 399 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 502 us MUL_MAT_ID(ffn_moe_down-36): 287 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 378 us MUL_MAT_ID(ffn_moe_down-37): 262 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 420 us MUL_MAT_ID(ffn_moe_down-38): 210 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 402 us MUL_MAT_ID(ffn_moe_down-39): 239 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 402 us MUL_MAT_ID(ffn_moe_down-40): 219 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 408 us MUL_MAT_ID(ffn_moe_down-41): 202 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 405 us MUL_MAT_ID(ffn_moe_down-42): 212 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 397 us MUL_MAT_ID(ffn_moe_down-43): 202 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 419 us MUL_MAT_ID(ffn_moe_down-44): 208 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 377 us MUL_MAT_ID(ffn_moe_down-45): 196 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 401 us MUL_MAT_ID(ffn_moe_down-46): 204 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 399 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 417 us MUL_MAT_ID(ffn_moe_down-48): 184 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 344 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 406 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 478 us MUL_MAT_ID(ffn_moe_down-51): 604 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 649 us MUL_MAT_ID(ffn_moe_down-52): 328 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 370 us MUL_MAT_ID(ffn_moe_down-53): 396 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 363 us MUL_MAT_ID(ffn_moe_down-54): 209 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 409 us MUL_MAT_ID(ffn_moe_down-55): 216 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 405 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 427 us MUL_MAT_ID(ffn_moe_down-57): 456 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 389 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 364 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 471 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 404 us MUL_MAT_ID(ffn_moe_down-61): 187 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 363 us MUL_MAT_ID(ffn_moe_down-62): 203 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 430 us MUL_MAT_ID(ffn_moe_down-63): 714 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 438 us MUL_MAT_ID(ffn_moe_down-64): 217 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 389 us MUL_MAT_ID(ffn_moe_down-65): 206 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 394 us MUL_MAT_ID(ffn_moe_down-66): 535 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 386 us MUL_MAT_ID(ffn_moe_down-67): 207 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 440 us MUL_MAT_ID(ffn_moe_down-68): 202 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 388 us MUL_MAT_ID(ffn_moe_down-69): 188 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 394 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 418 us MUL_MAT_ID(ffn_moe_down-71): 225 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 449 us MUL_MAT_ID(ffn_moe_down-72): 322 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 406 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 336 us MUL_MAT_ID(ffn_moe_down-74): 207 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 446 us MUL_MAT_ID(ffn_moe_down-75): 204 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 411 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 413 us MUL_MAT_ID(ffn_moe_down-77): 394 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 395 us MUL_MAT_ID(ffn_moe_down-78): 195 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 436 us MUL_MAT_ID(ffn_moe_down-79): 197 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 419 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 448 us MUL_MAT_ID(ffn_moe_down-81): 187 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 352 us MUL_MAT_ID(ffn_moe_down-82): 485 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 407 us MUL_MAT_ID(ffn_moe_down-83): 221 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 418 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 427 us MUL_MAT_ID(ffn_moe_down-85): 245 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 436 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 350 us MUL_MAT_ID(ffn_moe_down-87): 204 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 381 us MUL_MAT_ID(ffn_moe_down-88): 223 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 421 us MUL_MAT_ID(ffn_moe_down-89): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 380 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 423 us MUL_MAT_ID(ffn_moe_down-91): 246 us ggml_barrier(...): 21 us GET_ROWS(inp_embd): 4 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 368 us MUL_MAT_ID(ffn_moe_down-25): 202 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 420 us MUL_MAT_ID(ffn_moe_down-26): 205 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 432 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 548 us MUL_MAT_ID(ffn_moe_down-28): 193 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 429 us MUL_MAT_ID(ffn_moe_down-29): 215 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 504 us MUL_MAT_ID(ffn_moe_down-30): 198 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 419 us MUL_MAT_ID(ffn_moe_down-31): 423 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 400 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 371 us MUL_MAT_ID(ffn_moe_down-33): 191 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 394 us MUL_MAT_ID(ffn_moe_down-34): 210 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 417 us MUL_MAT_ID(ffn_moe_down-35): 221 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 396 us MUL_MAT_ID(ffn_moe_down-36): 225 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 383 us MUL_MAT_ID(ffn_moe_down-37): 250 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 437 us MUL_MAT_ID(ffn_moe_down-38): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 431 us MUL_MAT_ID(ffn_moe_down-39): 228 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 370 us MUL_MAT_ID(ffn_moe_down-40): 310 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 439 us MUL_MAT_ID(ffn_moe_down-41): 425 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 418 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 376 us MUL_MAT_ID(ffn_moe_down-43): 293 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 412 us MUL_MAT_ID(ffn_moe_down-44): 200 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 410 us MUL_MAT_ID(ffn_moe_down-45): 191 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 445 us MUL_MAT_ID(ffn_moe_down-46): 232 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 604 us MUL_MAT_ID(ffn_moe_down-47): 222 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 383 us MUL_MAT_ID(ffn_moe_down-48): 219 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 405 us MUL_MAT_ID(ffn_moe_down-49): 196 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 444 us MUL_MAT_ID(ffn_moe_down-50): 352 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 394 us MUL_MAT_ID(ffn_moe_down-51): 188 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 439 us MUL_MAT_ID(ffn_moe_down-52): 452 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 401 us MUL_MAT_ID(ffn_moe_down-53): 192 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 416 us MUL_MAT_ID(ffn_moe_down-54): 220 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 393 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 368 us MUL_MAT_ID(ffn_moe_down-56): 244 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 410 us MUL_MAT_ID(ffn_moe_down-57): 202 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 351 us MUL_MAT_ID(ffn_moe_down-58): 176 us ggml_barrier(...): 344 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 381 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 370 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 376 us MUL_MAT_ID(ffn_moe_down-61): 202 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 342 us MUL_MAT_ID(ffn_moe_down-62): 192 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 379 us MUL_MAT_ID(ffn_moe_down-63): 237 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 421 us MUL_MAT_ID(ffn_moe_down-64): 236 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 377 us MUL_MAT_ID(ffn_moe_down-65): 253 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 397 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 382 us MUL_MAT_ID(ffn_moe_down-67): 332 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 409 us MUL_MAT_ID(ffn_moe_down-68): 180 us ggml_barrier(...): 360 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 398 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 392 us MUL_MAT_ID(ffn_moe_down-70): 254 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 470 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 420 us MUL_MAT_ID(ffn_moe_down-72): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 433 us MUL_MAT_ID(ffn_moe_down-73): 197 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 439 us MUL_MAT_ID(ffn_moe_down-74): 391 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 414 us MUL_MAT_ID(ffn_moe_down-75): 205 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 442 us MUL_MAT_ID(ffn_moe_down-76): 205 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 408 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 421 us MUL_MAT_ID(ffn_moe_down-78): 213 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 462 us MUL_MAT_ID(ffn_moe_down-79): 205 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 437 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 404 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 417 us MUL_MAT_ID(ffn_moe_down-82): 212 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 420 us MUL_MAT_ID(ffn_moe_down-83): 424 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 628 us MUL_MAT_ID(ffn_moe_down-84): 248 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 397 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 514 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 399 us MUL_MAT_ID(ffn_moe_down-87): 197 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 368 us MUL_MAT_ID(ffn_moe_down-88): 218 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 401 us MUL_MAT_ID(ffn_moe_down-89): 185 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 422 us MUL_MAT_ID(ffn_moe_down-90): 587 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 403 us MUL_MAT_ID(ffn_moe_down-91): 246 us ggml_barrier(...): 18 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 414 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 391 us MUL_MAT_ID(ffn_moe_down-26): 206 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 402 us MUL_MAT_ID(ffn_moe_down-27): 301 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 448 us MUL_MAT_ID(ffn_moe_down-28): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 405 us MUL_MAT_ID(ffn_moe_down-29): 186 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 383 us MUL_MAT_ID(ffn_moe_down-30): 191 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 395 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 337 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 410 us MUL_MAT_ID(ffn_moe_down-33): 430 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 367 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 385 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 354 us MUL_MAT_ID(ffn_moe_down-36): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 434 us MUL_MAT_ID(ffn_moe_down-37): 274 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 398 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 361 us MUL_MAT_ID(ffn_moe_down-39): 671 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 388 us MUL_MAT_ID(ffn_moe_down-40): 187 us ggml_barrier(...): 432 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 523 us MUL_MAT_ID(ffn_moe_down-41): 298 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 367 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 386 us MUL_MAT_ID(ffn_moe_down-43): 416 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 393 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 387 us MUL_MAT_ID(ffn_moe_down-45): 223 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 385 us MUL_MAT_ID(ffn_moe_down-46): 203 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 386 us MUL_MAT_ID(ffn_moe_down-47): 277 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 367 us MUL_MAT_ID(ffn_moe_down-48): 225 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 419 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 338 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 365 us MUL_MAT_ID(ffn_moe_down-51): 205 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 404 us MUL_MAT_ID(ffn_moe_down-52): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 384 us MUL_MAT_ID(ffn_moe_down-53): 182 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 441 us MUL_MAT_ID(ffn_moe_down-54): 205 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 446 us MUL_MAT_ID(ffn_moe_down-55): 217 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 406 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 477 us MUL_MAT_ID(ffn_moe_down-57): 174 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 415 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 393 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 349 us MUL_MAT_ID(ffn_moe_down-60): 213 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 383 us MUL_MAT_ID(ffn_moe_down-61): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 375 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 378 us MUL_MAT_ID(ffn_moe_down-63): 208 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 409 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 353 us MUL_MAT_ID(ffn_moe_down-65): 205 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 374 us MUL_MAT_ID(ffn_moe_down-66): 222 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 448 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 415 us MUL_MAT_ID(ffn_moe_down-68): 191 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 408 us MUL_MAT_ID(ffn_moe_down-69): 206 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 518 us MUL_MAT_ID(ffn_moe_down-70): 175 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 400 us MUL_MAT_ID(ffn_moe_down-71): 202 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 404 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 351 us MUL_MAT_ID(ffn_moe_down-73): 206 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 380 us MUL_MAT_ID(ffn_moe_down-74): 244 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 434 us MUL_MAT_ID(ffn_moe_down-75): 228 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 428 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 402 us MUL_MAT_ID(ffn_moe_down-77): 253 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 419 us MUL_MAT_ID(ffn_moe_down-78): 212 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 678 us MUL_MAT_ID(ffn_moe_down-79): 342 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 344 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 399 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 462 us MUL_MAT_ID(ffn_moe_down-82): 261 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 394 us MUL_MAT_ID(ffn_moe_down-83): 543 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 399 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 429 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 420 us MUL_MAT_ID(ffn_moe_down-86): 328 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 388 us MUL_MAT_ID(ffn_moe_down-87): 185 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 387 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 395 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 494 us MUL_MAT_ID(ffn_moe_down-90): 737 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 346 us MUL_MAT_ID(ffn_moe_down-91): 242 us ggml_barrier(...): 76 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 384 us MUL_MAT_ID(ffn_moe_down-25): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 385 us MUL_MAT_ID(ffn_moe_down-26): 204 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 349 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 349 us MUL_MAT_ID(ffn_moe_down-28): 219 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 430 us MUL_MAT_ID(ffn_moe_down-29): 306 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 425 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 389 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 405 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 379 us MUL_MAT_ID(ffn_moe_down-33): 212 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 395 us MUL_MAT_ID(ffn_moe_down-34): 221 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 397 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 598 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 390 us MUL_MAT_ID(ffn_moe_down-37): 389 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 373 us MUL_MAT_ID(ffn_moe_down-38): 225 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 381 us MUL_MAT_ID(ffn_moe_down-39): 436 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 390 us MUL_MAT_ID(ffn_moe_down-40): 377 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 385 us MUL_MAT_ID(ffn_moe_down-41): 210 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 346 us MUL_MAT_ID(ffn_moe_down-42): 208 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 381 us MUL_MAT_ID(ffn_moe_down-43): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 362 us MUL_MAT_ID(ffn_moe_down-44): 290 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 375 us MUL_MAT_ID(ffn_moe_down-45): 210 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 380 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 386 us MUL_MAT_ID(ffn_moe_down-47): 186 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 385 us MUL_MAT_ID(ffn_moe_down-48): 187 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 382 us MUL_MAT_ID(ffn_moe_down-50): 207 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 402 us MUL_MAT_ID(ffn_moe_down-51): 327 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 397 us MUL_MAT_ID(ffn_moe_down-52): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 397 us MUL_MAT_ID(ffn_moe_down-53): 330 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 418 us MUL_MAT_ID(ffn_moe_down-54): 188 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 416 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 382 us MUL_MAT_ID(ffn_moe_down-56): 210 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 443 us MUL_MAT_ID(ffn_moe_down-57): 189 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 441 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 358 us MUL_MAT_ID(ffn_moe_down-59): 318 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 396 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 359 us MUL_MAT_ID(ffn_moe_down-61): 203 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 407 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 455 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 411 us MUL_MAT_ID(ffn_moe_down-64): 349 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 398 us MUL_MAT_ID(ffn_moe_down-65): 372 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 408 us MUL_MAT_ID(ffn_moe_down-66): 275 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 420 us MUL_MAT_ID(ffn_moe_down-67): 195 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 210 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 393 us MUL_MAT_ID(ffn_moe_down-69): 540 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 421 us MUL_MAT_ID(ffn_moe_down-70): 207 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 394 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 353 us MUL_MAT_ID(ffn_moe_down-72): 388 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 380 us MUL_MAT_ID(ffn_moe_down-73): 583 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 411 us MUL_MAT_ID(ffn_moe_down-74): 208 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 362 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 404 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 384 us MUL_MAT_ID(ffn_moe_down-77): 365 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 390 us MUL_MAT_ID(ffn_moe_down-78): 201 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 374 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 413 us MUL_MAT_ID(ffn_moe_down-80): 290 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 358 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 401 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 482 us MUL_MAT_ID(ffn_moe_down-83): 289 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 538 us MUL_MAT_ID(ffn_moe_down-84): 473 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 397 us MUL_MAT_ID(ffn_moe_down-85): 227 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 396 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 358 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 412 us MUL_MAT_ID(ffn_moe_down-88): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 391 us MUL_MAT_ID(ffn_moe_down-89): 243 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 482 us MUL_MAT_ID(ffn_moe_down-90): 259 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 425 us MUL_MAT_ID(ffn_moe_down-91): 269 us ggml_barrier(...): 4 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 382 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 410 us MUL_MAT_ID(ffn_moe_down-26): 184 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 372 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 456 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 364 us MUL_MAT_ID(ffn_moe_down-29): 208 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 427 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 423 us MUL_MAT_ID(ffn_moe_down-31): 186 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 455 us MUL_MAT_ID(ffn_moe_down-32): 286 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 382 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 385 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 337 us MUL_MAT_ID(ffn_moe_down-35): 189 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 368 us MUL_MAT_ID(ffn_moe_down-36): 187 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 378 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 556 us MUL_MAT_ID(ffn_moe_down-38): 320 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 365 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 370 us MUL_MAT_ID(ffn_moe_down-40): 241 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 450 us MUL_MAT_ID(ffn_moe_down-41): 186 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 417 us MUL_MAT_ID(ffn_moe_down-42): 183 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 420 us MUL_MAT_ID(ffn_moe_down-43): 208 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 331 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 402 us MUL_MAT_ID(ffn_moe_down-45): 342 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 371 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 442 us MUL_MAT_ID(ffn_moe_down-47): 324 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 420 us MUL_MAT_ID(ffn_moe_down-48): 446 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 425 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 377 us MUL_MAT_ID(ffn_moe_down-50): 198 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 384 us MUL_MAT_ID(ffn_moe_down-51): 320 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 378 us MUL_MAT_ID(ffn_moe_down-52): 208 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 422 us MUL_MAT_ID(ffn_moe_down-53): 397 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 388 us MUL_MAT_ID(ffn_moe_down-54): 214 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 387 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 446 us MUL_MAT_ID(ffn_moe_down-56): 377 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 384 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 467 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 432 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 423 us MUL_MAT_ID(ffn_moe_down-60): 202 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 437 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 418 us MUL_MAT_ID(ffn_moe_down-62): 204 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 382 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 439 us MUL_MAT_ID(ffn_moe_down-64): 199 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 401 us MUL_MAT_ID(ffn_moe_down-65): 203 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 461 us MUL_MAT_ID(ffn_moe_down-66): 401 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 383 us MUL_MAT_ID(ffn_moe_down-67): 206 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 403 us MUL_MAT_ID(ffn_moe_down-68): 204 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 449 us MUL_MAT_ID(ffn_moe_down-69): 186 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 386 us MUL_MAT_ID(ffn_moe_down-70): 230 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 375 us MUL_MAT_ID(ffn_moe_down-71): 199 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 431 us MUL_MAT_ID(ffn_moe_down-72): 209 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 208 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 428 us MUL_MAT_ID(ffn_moe_down-74): 203 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 439 us MUL_MAT_ID(ffn_moe_down-75): 206 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 419 us MUL_MAT_ID(ffn_moe_down-76): 204 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 459 us MUL_MAT_ID(ffn_moe_down-77): 200 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 394 us MUL_MAT_ID(ffn_moe_down-78): 184 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 370 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 362 us MUL_MAT_ID(ffn_moe_down-80): 201 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 380 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 397 us MUL_MAT_ID(ffn_moe_down-82): 216 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 374 us MUL_MAT_ID(ffn_moe_down-83): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 423 us MUL_MAT_ID(ffn_moe_down-84): 345 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 394 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 421 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 411 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 422 us MUL_MAT_ID(ffn_moe_down-88): 198 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 446 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 356 us MUL_MAT_ID(ffn_moe_down-90): 632 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 371 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 36 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 375 us MUL_MAT_ID(ffn_moe_down-25): 181 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 409 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 371 us MUL_MAT_ID(ffn_moe_down-27): 421 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 368 us MUL_MAT_ID(ffn_moe_down-28): 200 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 442 us MUL_MAT_ID(ffn_moe_down-29): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 400 us MUL_MAT_ID(ffn_moe_down-30): 204 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 474 us MUL_MAT_ID(ffn_moe_down-31): 224 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 502 us MUL_MAT_ID(ffn_moe_down-32): 313 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 209 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 411 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 428 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 365 us MUL_MAT_ID(ffn_moe_down-36): 447 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 406 us MUL_MAT_ID(ffn_moe_down-37): 257 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 413 us MUL_MAT_ID(ffn_moe_down-38): 220 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 408 us MUL_MAT_ID(ffn_moe_down-39): 771 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 394 us MUL_MAT_ID(ffn_moe_down-40): 197 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 400 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 415 us MUL_MAT_ID(ffn_moe_down-42): 183 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 377 us MUL_MAT_ID(ffn_moe_down-43): 329 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 410 us MUL_MAT_ID(ffn_moe_down-44): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 431 us MUL_MAT_ID(ffn_moe_down-45): 263 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 411 us MUL_MAT_ID(ffn_moe_down-46): 198 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 413 us MUL_MAT_ID(ffn_moe_down-47): 238 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 397 us MUL_MAT_ID(ffn_moe_down-48): 239 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 536 us MUL_MAT_ID(ffn_moe_down-49): 250 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 410 us MUL_MAT_ID(ffn_moe_down-50): 367 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 409 us MUL_MAT_ID(ffn_moe_down-51): 227 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 400 us MUL_MAT_ID(ffn_moe_down-52): 186 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 365 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 382 us MUL_MAT_ID(ffn_moe_down-54): 195 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 424 us MUL_MAT_ID(ffn_moe_down-55): 216 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 387 us MUL_MAT_ID(ffn_moe_down-56): 616 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 412 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 402 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 438 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 341 us MUL_MAT_ID(ffn_moe_down-60): 205 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 395 us MUL_MAT_ID(ffn_moe_down-61): 603 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 378 us MUL_MAT_ID(ffn_moe_down-62): 204 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 401 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 429 us MUL_MAT_ID(ffn_moe_down-64): 183 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 363 us MUL_MAT_ID(ffn_moe_down-65): 205 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 380 us MUL_MAT_ID(ffn_moe_down-66): 448 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 418 us MUL_MAT_ID(ffn_moe_down-67): 230 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 410 us MUL_MAT_ID(ffn_moe_down-68): 208 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 377 us MUL_MAT_ID(ffn_moe_down-69): 478 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 415 us MUL_MAT_ID(ffn_moe_down-70): 220 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 406 us MUL_MAT_ID(ffn_moe_down-71): 225 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 413 us MUL_MAT_ID(ffn_moe_down-72): 295 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 396 us MUL_MAT_ID(ffn_moe_down-73): 199 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 362 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 418 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 424 us MUL_MAT_ID(ffn_moe_down-76): 203 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 437 us MUL_MAT_ID(ffn_moe_down-77): 212 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 444 us MUL_MAT_ID(ffn_moe_down-78): 507 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 401 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 397 us MUL_MAT_ID(ffn_moe_down-80): 206 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 378 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 373 us MUL_MAT_ID(ffn_moe_down-82): 215 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 386 us MUL_MAT_ID(ffn_moe_down-83): 203 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 385 us MUL_MAT_ID(ffn_moe_down-84): 249 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 408 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 393 us MUL_MAT_ID(ffn_moe_down-86): 202 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 420 us MUL_MAT_ID(ffn_moe_down-87): 186 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 393 us MUL_MAT_ID(ffn_moe_down-88): 247 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 406 us MUL_MAT_ID(ffn_moe_down-89): 212 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 387 us MUL_MAT_ID(ffn_moe_down-90): 246 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 379 us MUL_MAT_ID(ffn_moe_down-91): 285 us ggml_barrier(...): 23 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 368 us MUL_MAT_ID(ffn_moe_down-25): 210 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 387 us MUL_MAT_ID(ffn_moe_down-26): 200 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 423 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 489 us MUL_MAT_ID(ffn_moe_down-28): 222 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 427 us MUL_MAT_ID(ffn_moe_down-29): 359 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 423 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 386 us MUL_MAT_ID(ffn_moe_down-31): 282 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 404 us MUL_MAT_ID(ffn_moe_down-32): 228 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 699 us MUL_MAT_ID(ffn_moe_down-33): 280 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 398 us MUL_MAT_ID(ffn_moe_down-34): 245 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 393 us MUL_MAT_ID(ffn_moe_down-35): 351 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 404 us MUL_MAT_ID(ffn_moe_down-36): 199 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 416 us MUL_MAT_ID(ffn_moe_down-37): 267 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 228 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 418 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 445 us MUL_MAT_ID(ffn_moe_down-40): 201 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 427 us MUL_MAT_ID(ffn_moe_down-41): 225 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 419 us MUL_MAT_ID(ffn_moe_down-42): 218 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 642 us MUL_MAT_ID(ffn_moe_down-43): 211 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 373 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 397 us MUL_MAT_ID(ffn_moe_down-45): 451 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 442 us MUL_MAT_ID(ffn_moe_down-46): 247 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 388 us MUL_MAT_ID(ffn_moe_down-47): 204 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 406 us MUL_MAT_ID(ffn_moe_down-48): 514 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 414 us MUL_MAT_ID(ffn_moe_down-49): 212 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 427 us MUL_MAT_ID(ffn_moe_down-50): 288 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 408 us MUL_MAT_ID(ffn_moe_down-51): 202 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 386 us MUL_MAT_ID(ffn_moe_down-52): 200 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 535 us MUL_MAT_ID(ffn_moe_down-53): 181 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 381 us MUL_MAT_ID(ffn_moe_down-54): 354 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 406 us MUL_MAT_ID(ffn_moe_down-55): 294 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 425 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 428 us MUL_MAT_ID(ffn_moe_down-57): 201 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 420 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 416 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 437 us MUL_MAT_ID(ffn_moe_down-60): 231 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 437 us MUL_MAT_ID(ffn_moe_down-61): 221 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 408 us MUL_MAT_ID(ffn_moe_down-62): 203 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 469 us MUL_MAT_ID(ffn_moe_down-63): 222 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 405 us MUL_MAT_ID(ffn_moe_down-64): 391 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 440 us MUL_MAT_ID(ffn_moe_down-65): 185 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 395 us MUL_MAT_ID(ffn_moe_down-66): 251 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 355 us MUL_MAT_ID(ffn_moe_down-67): 214 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 388 us MUL_MAT_ID(ffn_moe_down-68): 209 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 416 us MUL_MAT_ID(ffn_moe_down-69): 188 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 373 us MUL_MAT_ID(ffn_moe_down-70): 439 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 395 us MUL_MAT_ID(ffn_moe_down-71): 213 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 446 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 438 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 399 us MUL_MAT_ID(ffn_moe_down-74): 213 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 428 us MUL_MAT_ID(ffn_moe_down-75): 187 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 397 us MUL_MAT_ID(ffn_moe_down-76): 196 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 428 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 428 us MUL_MAT_ID(ffn_moe_down-78): 189 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 455 us MUL_MAT_ID(ffn_moe_down-79): 197 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 416 us MUL_MAT_ID(ffn_moe_down-80): 212 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 413 us MUL_MAT_ID(ffn_moe_down-81): 276 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 406 us MUL_MAT_ID(ffn_moe_down-82): 194 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 404 us MUL_MAT_ID(ffn_moe_down-83): 183 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 418 us MUL_MAT_ID(ffn_moe_down-84): 403 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 397 us MUL_MAT_ID(ffn_moe_down-85): 230 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 419 us MUL_MAT_ID(ffn_moe_down-86): 531 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 517 us MUL_MAT_ID(ffn_moe_down-87): 208 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 383 us MUL_MAT_ID(ffn_moe_down-88): 205 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 431 us MUL_MAT_ID(ffn_moe_down-89): 195 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 437 us MUL_MAT_ID(ffn_moe_down-90): 565 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 451 us MUL_MAT_ID(ffn_moe_down-91): 275 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 391 us MUL_MAT_ID(ffn_moe_down-25): 317 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 422 us MUL_MAT_ID(ffn_moe_down-26): 205 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 390 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 390 us MUL_MAT_ID(ffn_moe_down-28): 214 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 418 us MUL_MAT_ID(ffn_moe_down-29): 185 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 379 us MUL_MAT_ID(ffn_moe_down-30): 223 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 418 us MUL_MAT_ID(ffn_moe_down-31): 185 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 382 us MUL_MAT_ID(ffn_moe_down-32): 216 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 433 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 402 us MUL_MAT_ID(ffn_moe_down-34): 211 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 388 us MUL_MAT_ID(ffn_moe_down-35): 225 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 433 us MUL_MAT_ID(ffn_moe_down-36): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 366 us MUL_MAT_ID(ffn_moe_down-37): 241 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 424 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 357 us MUL_MAT_ID(ffn_moe_down-39): 271 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 407 us MUL_MAT_ID(ffn_moe_down-40): 254 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 652 us MUL_MAT_ID(ffn_moe_down-41): 269 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 357 us MUL_MAT_ID(ffn_moe_down-42): 216 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 412 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 436 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 433 us MUL_MAT_ID(ffn_moe_down-45): 208 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 341 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 434 us MUL_MAT_ID(ffn_moe_down-47): 207 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 416 us MUL_MAT_ID(ffn_moe_down-48): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 384 us MUL_MAT_ID(ffn_moe_down-49): 201 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 371 us MUL_MAT_ID(ffn_moe_down-50): 442 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 370 us MUL_MAT_ID(ffn_moe_down-51): 201 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 411 us MUL_MAT_ID(ffn_moe_down-52): 249 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 611 us MUL_MAT_ID(ffn_moe_down-53): 213 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 368 us MUL_MAT_ID(ffn_moe_down-54): 182 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 372 us MUL_MAT_ID(ffn_moe_down-55): 265 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 447 us MUL_MAT_ID(ffn_moe_down-56): 315 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 380 us MUL_MAT_ID(ffn_moe_down-57): 201 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 431 us MUL_MAT_ID(ffn_moe_down-58): 286 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 371 us MUL_MAT_ID(ffn_moe_down-59): 218 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 368 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 394 us MUL_MAT_ID(ffn_moe_down-61): 205 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 426 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 345 us MUL_MAT_ID(ffn_moe_down-63): 266 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 429 us MUL_MAT_ID(ffn_moe_down-64): 231 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 393 us MUL_MAT_ID(ffn_moe_down-65): 366 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 401 us MUL_MAT_ID(ffn_moe_down-66): 184 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 351 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 383 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 357 us MUL_MAT_ID(ffn_moe_down-69): 232 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 428 us MUL_MAT_ID(ffn_moe_down-70): 198 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 450 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 400 us MUL_MAT_ID(ffn_moe_down-72): 205 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 374 us MUL_MAT_ID(ffn_moe_down-73): 203 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 392 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 381 us MUL_MAT_ID(ffn_moe_down-75): 220 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 409 us MUL_MAT_ID(ffn_moe_down-76): 206 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 421 us MUL_MAT_ID(ffn_moe_down-77): 330 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 376 us MUL_MAT_ID(ffn_moe_down-78): 188 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 349 us MUL_MAT_ID(ffn_moe_down-79): 202 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 401 us MUL_MAT_ID(ffn_moe_down-80): 210 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 449 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 363 us MUL_MAT_ID(ffn_moe_down-83): 436 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 355 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 393 us MUL_MAT_ID(ffn_moe_down-85): 674 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 406 us MUL_MAT_ID(ffn_moe_down-86): 210 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 331 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 461 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 377 us MUL_MAT_ID(ffn_moe_down-89): 216 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 557 us MUL_MAT_ID(ffn_moe_down-90): 677 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 397 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 310 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 384 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 401 us MUL_MAT_ID(ffn_moe_down-26): 203 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 429 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 425 us MUL_MAT_ID(ffn_moe_down-28): 595 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 415 us MUL_MAT_ID(ffn_moe_down-29): 205 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 405 us MUL_MAT_ID(ffn_moe_down-30): 186 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 428 us MUL_MAT_ID(ffn_moe_down-31): 391 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 392 us MUL_MAT_ID(ffn_moe_down-32): 370 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 381 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 532 us MUL_MAT_ID(ffn_moe_down-34): 259 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 402 us MUL_MAT_ID(ffn_moe_down-35): 206 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 395 us MUL_MAT_ID(ffn_moe_down-36): 585 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 389 us MUL_MAT_ID(ffn_moe_down-37): 256 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 436 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 386 us MUL_MAT_ID(ffn_moe_down-39): 493 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 426 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 367 us MUL_MAT_ID(ffn_moe_down-41): 206 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 403 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 368 us MUL_MAT_ID(ffn_moe_down-43): 220 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 392 us MUL_MAT_ID(ffn_moe_down-44): 185 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 416 us MUL_MAT_ID(ffn_moe_down-45): 353 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 347 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 430 us MUL_MAT_ID(ffn_moe_down-47): 211 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 452 us MUL_MAT_ID(ffn_moe_down-48): 185 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 428 us MUL_MAT_ID(ffn_moe_down-49): 268 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 379 us MUL_MAT_ID(ffn_moe_down-50): 185 us ggml_barrier(...): 348 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 444 us MUL_MAT_ID(ffn_moe_down-51): 185 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 419 us MUL_MAT_ID(ffn_moe_down-52): 223 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 416 us MUL_MAT_ID(ffn_moe_down-53): 323 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 648 us MUL_MAT_ID(ffn_moe_down-54): 333 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 394 us MUL_MAT_ID(ffn_moe_down-55): 203 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 409 us MUL_MAT_ID(ffn_moe_down-56): 277 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 441 us MUL_MAT_ID(ffn_moe_down-57): 203 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 399 us MUL_MAT_ID(ffn_moe_down-58): 189 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 371 us MUL_MAT_ID(ffn_moe_down-59): 189 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 387 us MUL_MAT_ID(ffn_moe_down-60): 185 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 348 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 374 us MUL_MAT_ID(ffn_moe_down-62): 224 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 378 us MUL_MAT_ID(ffn_moe_down-63): 562 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 397 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 374 us MUL_MAT_ID(ffn_moe_down-65): 216 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 571 us MUL_MAT_ID(ffn_moe_down-66): 214 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 404 us MUL_MAT_ID(ffn_moe_down-67): 196 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 380 us MUL_MAT_ID(ffn_moe_down-68): 198 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 349 us MUL_MAT_ID(ffn_moe_down-69): 170 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 359 us MUL_MAT_ID(ffn_moe_down-70): 209 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 429 us MUL_MAT_ID(ffn_moe_down-71): 189 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 395 us MUL_MAT_ID(ffn_moe_down-72): 540 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 344 us MUL_MAT_ID(ffn_moe_down-73): 275 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 428 us MUL_MAT_ID(ffn_moe_down-74): 500 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 387 us MUL_MAT_ID(ffn_moe_down-75): 224 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 354 us MUL_MAT_ID(ffn_moe_down-76): 205 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 433 us MUL_MAT_ID(ffn_moe_down-77): 431 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 418 us MUL_MAT_ID(ffn_moe_down-78): 271 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 397 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 366 us MUL_MAT_ID(ffn_moe_down-80): 185 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 343 us MUL_MAT_ID(ffn_moe_down-81): 213 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 452 us MUL_MAT_ID(ffn_moe_down-82): 207 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 377 us MUL_MAT_ID(ffn_moe_down-83): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 381 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 402 us MUL_MAT_ID(ffn_moe_down-85): 244 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 415 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 396 us MUL_MAT_ID(ffn_moe_down-87): 200 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 392 us MUL_MAT_ID(ffn_moe_down-88): 178 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 374 us MUL_MAT_ID(ffn_moe_down-89): 214 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 395 us MUL_MAT_ID(ffn_moe_down-90): 264 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 383 us MUL_MAT_ID(ffn_moe_down-91): 249 us ggml_barrier(...): 19 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 407 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 414 us MUL_MAT_ID(ffn_moe_down-26): 216 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 377 us MUL_MAT_ID(ffn_moe_down-27): 190 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 351 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 413 us MUL_MAT_ID(ffn_moe_down-29): 201 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 450 us MUL_MAT_ID(ffn_moe_down-30): 274 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 418 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 338 us MUL_MAT_ID(ffn_moe_down-32): 190 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 414 us MUL_MAT_ID(ffn_moe_down-33): 197 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 344 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 402 us MUL_MAT_ID(ffn_moe_down-35): 183 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 401 us MUL_MAT_ID(ffn_moe_down-36): 197 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 420 us MUL_MAT_ID(ffn_moe_down-37): 246 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 387 us MUL_MAT_ID(ffn_moe_down-38): 223 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 379 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 390 us MUL_MAT_ID(ffn_moe_down-40): 223 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 409 us MUL_MAT_ID(ffn_moe_down-41): 209 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 387 us MUL_MAT_ID(ffn_moe_down-42): 218 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 400 us MUL_MAT_ID(ffn_moe_down-43): 199 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 332 us MUL_MAT_ID(ffn_moe_down-44): 253 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 433 us MUL_MAT_ID(ffn_moe_down-45): 216 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 389 us MUL_MAT_ID(ffn_moe_down-46): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 357 us MUL_MAT_ID(ffn_moe_down-47): 205 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 424 us MUL_MAT_ID(ffn_moe_down-48): 250 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 392 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 391 us MUL_MAT_ID(ffn_moe_down-50): 344 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 382 us MUL_MAT_ID(ffn_moe_down-51): 331 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 399 us MUL_MAT_ID(ffn_moe_down-52): 237 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 406 us MUL_MAT_ID(ffn_moe_down-53): 190 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 357 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 431 us MUL_MAT_ID(ffn_moe_down-55): 208 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 390 us MUL_MAT_ID(ffn_moe_down-56): 204 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 418 us MUL_MAT_ID(ffn_moe_down-57): 225 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 410 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 397 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 374 us MUL_MAT_ID(ffn_moe_down-60): 195 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 410 us MUL_MAT_ID(ffn_moe_down-61): 215 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 589 us MUL_MAT_ID(ffn_moe_down-62): 224 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 370 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 387 us MUL_MAT_ID(ffn_moe_down-64): 335 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 393 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 353 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 403 us MUL_MAT_ID(ffn_moe_down-67): 303 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 456 us MUL_MAT_ID(ffn_moe_down-68): 275 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 442 us MUL_MAT_ID(ffn_moe_down-69): 216 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 628 us MUL_MAT_ID(ffn_moe_down-70): 212 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 417 us MUL_MAT_ID(ffn_moe_down-71): 205 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 385 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 259 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 426 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 419 us MUL_MAT_ID(ffn_moe_down-75): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 450 us MUL_MAT_ID(ffn_moe_down-76): 197 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 411 us MUL_MAT_ID(ffn_moe_down-77): 252 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 426 us MUL_MAT_ID(ffn_moe_down-78): 208 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 421 us MUL_MAT_ID(ffn_moe_down-79): 225 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 420 us MUL_MAT_ID(ffn_moe_down-80): 250 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 429 us MUL_MAT_ID(ffn_moe_down-81): 206 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 465 us MUL_MAT_ID(ffn_moe_down-82): 394 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 419 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 425 us MUL_MAT_ID(ffn_moe_down-84): 226 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 391 us MUL_MAT_ID(ffn_moe_down-85): 228 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 422 us MUL_MAT_ID(ffn_moe_down-86): 281 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 415 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 444 us MUL_MAT_ID(ffn_moe_down-88): 203 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 453 us MUL_MAT_ID(ffn_moe_down-89): 185 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 418 us MUL_MAT_ID(ffn_moe_down-90): 240 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 416 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 17 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 366 us MUL_MAT_ID(ffn_moe_down-25): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 403 us MUL_MAT_ID(ffn_moe_down-26): 216 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 379 us MUL_MAT_ID(ffn_moe_down-27): 190 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 363 us MUL_MAT_ID(ffn_moe_down-28): 199 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 433 us MUL_MAT_ID(ffn_moe_down-29): 210 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 464 us MUL_MAT_ID(ffn_moe_down-30): 644 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 354 us MUL_MAT_ID(ffn_moe_down-31): 187 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 403 us MUL_MAT_ID(ffn_moe_down-32): 203 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 348 us MUL_MAT_ID(ffn_moe_down-33): 251 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 362 us MUL_MAT_ID(ffn_moe_down-34): 190 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 381 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 339 us MUL_MAT_ID(ffn_moe_down-36): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 356 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 384 us MUL_MAT_ID(ffn_moe_down-38): 197 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 367 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 386 us MUL_MAT_ID(ffn_moe_down-40): 225 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 438 us MUL_MAT_ID(ffn_moe_down-41): 198 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 391 us MUL_MAT_ID(ffn_moe_down-42): 208 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 419 us MUL_MAT_ID(ffn_moe_down-43): 284 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 401 us MUL_MAT_ID(ffn_moe_down-44): 393 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 373 us MUL_MAT_ID(ffn_moe_down-45): 239 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 421 us MUL_MAT_ID(ffn_moe_down-46): 377 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 404 us MUL_MAT_ID(ffn_moe_down-47): 184 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 340 us MUL_MAT_ID(ffn_moe_down-48): 219 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 450 us MUL_MAT_ID(ffn_moe_down-49): 508 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 390 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 388 us MUL_MAT_ID(ffn_moe_down-51): 219 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 358 us MUL_MAT_ID(ffn_moe_down-52): 184 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 380 us MUL_MAT_ID(ffn_moe_down-53): 192 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 376 us MUL_MAT_ID(ffn_moe_down-54): 216 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 415 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 413 us MUL_MAT_ID(ffn_moe_down-56): 335 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 431 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 384 us MUL_MAT_ID(ffn_moe_down-58): 270 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 384 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 409 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 394 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 391 us MUL_MAT_ID(ffn_moe_down-62): 242 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 413 us MUL_MAT_ID(ffn_moe_down-63): 208 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 351 us MUL_MAT_ID(ffn_moe_down-64): 204 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 411 us MUL_MAT_ID(ffn_moe_down-65): 460 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 494 us MUL_MAT_ID(ffn_moe_down-66): 505 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 354 us MUL_MAT_ID(ffn_moe_down-67): 210 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 425 us MUL_MAT_ID(ffn_moe_down-68): 211 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 387 us MUL_MAT_ID(ffn_moe_down-69): 384 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 415 us MUL_MAT_ID(ffn_moe_down-70): 341 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 393 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 397 us MUL_MAT_ID(ffn_moe_down-72): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 478 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 385 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 426 us MUL_MAT_ID(ffn_moe_down-75): 187 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 425 us MUL_MAT_ID(ffn_moe_down-76): 180 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 527 us MUL_MAT_ID(ffn_moe_down-77): 188 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 414 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 401 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 359 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 407 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 446 us MUL_MAT_ID(ffn_moe_down-82): 246 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 590 us MUL_MAT_ID(ffn_moe_down-83): 779 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 365 us MUL_MAT_ID(ffn_moe_down-84): 249 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 424 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 368 us MUL_MAT_ID(ffn_moe_down-86): 209 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 421 us MUL_MAT_ID(ffn_moe_down-87): 279 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 397 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 539 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 415 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 422 us MUL_MAT_ID(ffn_moe_down-91): 242 us ggml_barrier(...): 27 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 356 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 389 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 391 us MUL_MAT_ID(ffn_moe_down-27): 214 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 472 us MUL_MAT_ID(ffn_moe_down-28): 218 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 421 us MUL_MAT_ID(ffn_moe_down-29): 226 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 438 us MUL_MAT_ID(ffn_moe_down-30): 257 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 429 us MUL_MAT_ID(ffn_moe_down-31): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 348 us MUL_MAT_ID(ffn_moe_down-32): 217 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 386 us MUL_MAT_ID(ffn_moe_down-33): 187 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 202 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 354 us MUL_MAT_ID(ffn_moe_down-35): 205 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 424 us MUL_MAT_ID(ffn_moe_down-36): 293 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 411 us MUL_MAT_ID(ffn_moe_down-37): 241 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 378 us MUL_MAT_ID(ffn_moe_down-38): 191 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 375 us MUL_MAT_ID(ffn_moe_down-39): 456 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 386 us MUL_MAT_ID(ffn_moe_down-40): 219 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 419 us MUL_MAT_ID(ffn_moe_down-41): 272 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 381 us MUL_MAT_ID(ffn_moe_down-42): 233 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 432 us MUL_MAT_ID(ffn_moe_down-43): 210 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 392 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 361 us MUL_MAT_ID(ffn_moe_down-45): 218 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 447 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 417 us MUL_MAT_ID(ffn_moe_down-47): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 368 us MUL_MAT_ID(ffn_moe_down-48): 238 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 351 us MUL_MAT_ID(ffn_moe_down-49): 217 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 406 us MUL_MAT_ID(ffn_moe_down-50): 350 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 418 us MUL_MAT_ID(ffn_moe_down-51): 189 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 444 us MUL_MAT_ID(ffn_moe_down-52): 391 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 435 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 414 us MUL_MAT_ID(ffn_moe_down-54): 551 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 364 us MUL_MAT_ID(ffn_moe_down-55): 184 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 440 us MUL_MAT_ID(ffn_moe_down-56): 450 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 383 us MUL_MAT_ID(ffn_moe_down-57): 310 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 406 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 413 us MUL_MAT_ID(ffn_moe_down-59): 417 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 371 us MUL_MAT_ID(ffn_moe_down-60): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 370 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 438 us MUL_MAT_ID(ffn_moe_down-62): 209 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 426 us MUL_MAT_ID(ffn_moe_down-63): 231 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 419 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 359 us MUL_MAT_ID(ffn_moe_down-65): 303 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 409 us MUL_MAT_ID(ffn_moe_down-66): 204 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 432 us MUL_MAT_ID(ffn_moe_down-67): 202 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 389 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 376 us MUL_MAT_ID(ffn_moe_down-69): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 342 us MUL_MAT_ID(ffn_moe_down-70): 230 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 401 us MUL_MAT_ID(ffn_moe_down-71): 202 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 420 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 419 us MUL_MAT_ID(ffn_moe_down-73): 209 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 382 us MUL_MAT_ID(ffn_moe_down-74): 205 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 429 us MUL_MAT_ID(ffn_moe_down-75): 226 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 371 us MUL_MAT_ID(ffn_moe_down-76): 462 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 392 us MUL_MAT_ID(ffn_moe_down-77): 465 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 367 us MUL_MAT_ID(ffn_moe_down-78): 206 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 424 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 370 us MUL_MAT_ID(ffn_moe_down-80): 241 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 471 us MUL_MAT_ID(ffn_moe_down-81): 234 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 435 us MUL_MAT_ID(ffn_moe_down-82): 477 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 386 us MUL_MAT_ID(ffn_moe_down-83): 268 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 393 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 410 us MUL_MAT_ID(ffn_moe_down-85): 224 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 356 us MUL_MAT_ID(ffn_moe_down-86): 198 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 467 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 437 us MUL_MAT_ID(ffn_moe_down-88): 213 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 378 us MUL_MAT_ID(ffn_moe_down-89): 205 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 430 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 414 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 43 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 354 us MUL_MAT_ID(ffn_moe_down-25): 397 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 376 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 392 us MUL_MAT_ID(ffn_moe_down-27): 213 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 418 us MUL_MAT_ID(ffn_moe_down-28): 226 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 468 us MUL_MAT_ID(ffn_moe_down-29): 183 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 382 us MUL_MAT_ID(ffn_moe_down-30): 217 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 473 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 439 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 405 us MUL_MAT_ID(ffn_moe_down-33): 214 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 433 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 437 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 416 us MUL_MAT_ID(ffn_moe_down-36): 180 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 391 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 394 us MUL_MAT_ID(ffn_moe_down-38): 422 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 395 us MUL_MAT_ID(ffn_moe_down-39): 695 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 391 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 400 us MUL_MAT_ID(ffn_moe_down-41): 197 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 443 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 455 us MUL_MAT_ID(ffn_moe_down-43): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 429 us MUL_MAT_ID(ffn_moe_down-44): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 359 us MUL_MAT_ID(ffn_moe_down-45): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 368 us MUL_MAT_ID(ffn_moe_down-46): 219 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 361 us MUL_MAT_ID(ffn_moe_down-47): 291 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 385 us MUL_MAT_ID(ffn_moe_down-48): 190 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 366 us MUL_MAT_ID(ffn_moe_down-49): 214 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 457 us MUL_MAT_ID(ffn_moe_down-50): 639 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 410 us MUL_MAT_ID(ffn_moe_down-51): 530 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 369 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 411 us MUL_MAT_ID(ffn_moe_down-53): 297 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 406 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 422 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 383 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 432 us MUL_MAT_ID(ffn_moe_down-57): 568 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 416 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 359 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 366 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 380 us MUL_MAT_ID(ffn_moe_down-61): 201 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 384 us MUL_MAT_ID(ffn_moe_down-62): 234 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 554 us MUL_MAT_ID(ffn_moe_down-63): 343 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 652 us MUL_MAT_ID(ffn_moe_down-64): 326 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 345 us MUL_MAT_ID(ffn_moe_down-65): 200 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 402 us MUL_MAT_ID(ffn_moe_down-66): 199 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 401 us MUL_MAT_ID(ffn_moe_down-67): 252 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 393 us MUL_MAT_ID(ffn_moe_down-68): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 414 us MUL_MAT_ID(ffn_moe_down-69): 512 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 390 us MUL_MAT_ID(ffn_moe_down-70): 326 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 412 us MUL_MAT_ID(ffn_moe_down-71): 208 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 397 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 395 us MUL_MAT_ID(ffn_moe_down-73): 213 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 408 us MUL_MAT_ID(ffn_moe_down-74): 203 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 404 us MUL_MAT_ID(ffn_moe_down-75): 226 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 392 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 400 us MUL_MAT_ID(ffn_moe_down-77): 205 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 405 us MUL_MAT_ID(ffn_moe_down-78): 491 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 393 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 386 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 346 us MUL_MAT_ID(ffn_moe_down-81): 210 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 380 us MUL_MAT_ID(ffn_moe_down-82): 179 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 383 us MUL_MAT_ID(ffn_moe_down-83): 210 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 370 us MUL_MAT_ID(ffn_moe_down-84): 223 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 370 us MUL_MAT_ID(ffn_moe_down-85): 252 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 382 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 417 us MUL_MAT_ID(ffn_moe_down-87): 188 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 422 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 343 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 395 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 423 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 25 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 389 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 377 us MUL_MAT_ID(ffn_moe_down-26): 184 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 432 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 390 us MUL_MAT_ID(ffn_moe_down-28): 190 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 360 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 397 us MUL_MAT_ID(ffn_moe_down-30): 405 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 423 us MUL_MAT_ID(ffn_moe_down-31): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 338 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 330 us MUL_MAT_ID(ffn_moe_down-33): 196 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 212 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 417 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 456 us MUL_MAT_ID(ffn_moe_down-36): 331 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 448 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 399 us MUL_MAT_ID(ffn_moe_down-38): 218 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 397 us MUL_MAT_ID(ffn_moe_down-39): 228 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 448 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 390 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 442 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 414 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 374 us MUL_MAT_ID(ffn_moe_down-44): 234 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 425 us MUL_MAT_ID(ffn_moe_down-45): 361 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 383 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 453 us MUL_MAT_ID(ffn_moe_down-47): 213 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 366 us MUL_MAT_ID(ffn_moe_down-48): 205 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 202 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 416 us MUL_MAT_ID(ffn_moe_down-50): 183 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 417 us MUL_MAT_ID(ffn_moe_down-51): 211 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 466 us MUL_MAT_ID(ffn_moe_down-52): 372 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 375 us MUL_MAT_ID(ffn_moe_down-53): 201 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 425 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 374 us MUL_MAT_ID(ffn_moe_down-55): 185 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 407 us MUL_MAT_ID(ffn_moe_down-56): 183 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 416 us MUL_MAT_ID(ffn_moe_down-57): 336 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 391 us MUL_MAT_ID(ffn_moe_down-58): 313 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 411 us MUL_MAT_ID(ffn_moe_down-59): 205 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 410 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 382 us MUL_MAT_ID(ffn_moe_down-61): 420 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 386 us MUL_MAT_ID(ffn_moe_down-62): 198 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 381 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 384 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 402 us MUL_MAT_ID(ffn_moe_down-65): 215 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 374 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 460 us MUL_MAT_ID(ffn_moe_down-67): 208 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 414 us MUL_MAT_ID(ffn_moe_down-68): 192 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 380 us MUL_MAT_ID(ffn_moe_down-69): 369 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 391 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 402 us MUL_MAT_ID(ffn_moe_down-71): 184 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 407 us MUL_MAT_ID(ffn_moe_down-72): 201 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 445 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 414 us MUL_MAT_ID(ffn_moe_down-74): 205 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 456 us MUL_MAT_ID(ffn_moe_down-75): 184 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 486 us MUL_MAT_ID(ffn_moe_down-76): 211 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 447 us MUL_MAT_ID(ffn_moe_down-77): 286 us ggml_barrier(...): 339 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 442 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 440 us MUL_MAT_ID(ffn_moe_down-79): 267 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 467 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 438 us MUL_MAT_ID(ffn_moe_down-81): 395 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 356 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 455 us MUL_MAT_ID(ffn_moe_down-83): 347 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 446 us MUL_MAT_ID(ffn_moe_down-84): 252 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 416 us MUL_MAT_ID(ffn_moe_down-85): 234 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 400 us MUL_MAT_ID(ffn_moe_down-86): 184 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 410 us MUL_MAT_ID(ffn_moe_down-87): 368 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 422 us MUL_MAT_ID(ffn_moe_down-88): 218 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 388 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 385 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 415 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 36 us GET_ROWS(inp_embd): 22 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 397 us MUL_MAT_ID(ffn_moe_down-25): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 396 us MUL_MAT_ID(ffn_moe_down-26): 408 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 398 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 366 us MUL_MAT_ID(ffn_moe_down-28): 200 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 408 us MUL_MAT_ID(ffn_moe_down-29): 593 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 358 us MUL_MAT_ID(ffn_moe_down-30): 264 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 386 us MUL_MAT_ID(ffn_moe_down-31): 193 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 405 us MUL_MAT_ID(ffn_moe_down-32): 180 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 408 us MUL_MAT_ID(ffn_moe_down-33): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 392 us MUL_MAT_ID(ffn_moe_down-34): 421 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 355 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 382 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 362 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 407 us MUL_MAT_ID(ffn_moe_down-39): 243 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 414 us MUL_MAT_ID(ffn_moe_down-40): 249 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 735 us MUL_MAT_ID(ffn_moe_down-41): 321 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 347 us MUL_MAT_ID(ffn_moe_down-42): 207 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 384 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 334 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 327 us MUL_MAT_ID(ffn_moe_down-45): 202 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 379 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 391 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 382 us MUL_MAT_ID(ffn_moe_down-48): 187 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 369 us MUL_MAT_ID(ffn_moe_down-49): 255 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 411 us MUL_MAT_ID(ffn_moe_down-50): 210 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 402 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 406 us MUL_MAT_ID(ffn_moe_down-52): 205 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 506 us MUL_MAT_ID(ffn_moe_down-53): 348 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 350 us MUL_MAT_ID(ffn_moe_down-54): 189 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 407 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 382 us MUL_MAT_ID(ffn_moe_down-56): 222 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 424 us MUL_MAT_ID(ffn_moe_down-57): 183 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 395 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 384 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 349 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 365 us MUL_MAT_ID(ffn_moe_down-61): 186 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 334 us MUL_MAT_ID(ffn_moe_down-62): 300 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 366 us MUL_MAT_ID(ffn_moe_down-63): 264 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 392 us MUL_MAT_ID(ffn_moe_down-64): 196 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 404 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 356 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 425 us MUL_MAT_ID(ffn_moe_down-67): 206 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 380 us MUL_MAT_ID(ffn_moe_down-68): 292 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 426 us MUL_MAT_ID(ffn_moe_down-69): 187 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 396 us MUL_MAT_ID(ffn_moe_down-70): 209 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 426 us MUL_MAT_ID(ffn_moe_down-71): 278 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 420 us MUL_MAT_ID(ffn_moe_down-73): 277 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 385 us MUL_MAT_ID(ffn_moe_down-74): 270 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 398 us MUL_MAT_ID(ffn_moe_down-75): 222 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 425 us MUL_MAT_ID(ffn_moe_down-76): 343 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 380 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 426 us MUL_MAT_ID(ffn_moe_down-78): 546 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 440 us MUL_MAT_ID(ffn_moe_down-79): 313 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 415 us MUL_MAT_ID(ffn_moe_down-80): 223 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 438 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 383 us MUL_MAT_ID(ffn_moe_down-82): 258 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 422 us MUL_MAT_ID(ffn_moe_down-83): 184 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 375 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 421 us MUL_MAT_ID(ffn_moe_down-85): 285 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 385 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 546 us MUL_MAT_ID(ffn_moe_down-87): 239 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 488 us MUL_MAT_ID(ffn_moe_down-88): 224 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 397 us MUL_MAT_ID(ffn_moe_down-89): 195 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 379 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 389 us MUL_MAT_ID(ffn_moe_down-91): 207 us ggml_barrier(...): 248 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 358 us MUL_MAT_ID(ffn_moe_down-25): 207 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 407 us MUL_MAT_ID(ffn_moe_down-26): 198 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 357 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 379 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 374 us MUL_MAT_ID(ffn_moe_down-29): 196 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 386 us MUL_MAT_ID(ffn_moe_down-30): 186 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 336 us MUL_MAT_ID(ffn_moe_down-31): 206 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 396 us MUL_MAT_ID(ffn_moe_down-32): 186 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 383 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 348 us MUL_MAT_ID(ffn_moe_down-34): 408 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 378 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 338 us MUL_MAT_ID(ffn_moe_down-36): 227 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 378 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 394 us MUL_MAT_ID(ffn_moe_down-38): 223 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 386 us MUL_MAT_ID(ffn_moe_down-39): 450 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 399 us MUL_MAT_ID(ffn_moe_down-40): 241 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 406 us MUL_MAT_ID(ffn_moe_down-41): 204 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 380 us MUL_MAT_ID(ffn_moe_down-42): 187 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 332 us MUL_MAT_ID(ffn_moe_down-43): 210 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 397 us MUL_MAT_ID(ffn_moe_down-44): 208 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 366 us MUL_MAT_ID(ffn_moe_down-45): 165 us ggml_barrier(...): 370 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 389 us MUL_MAT_ID(ffn_moe_down-46): 450 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 365 us MUL_MAT_ID(ffn_moe_down-47): 202 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 392 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 408 us MUL_MAT_ID(ffn_moe_down-49): 208 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 401 us MUL_MAT_ID(ffn_moe_down-50): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 385 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 384 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 390 us MUL_MAT_ID(ffn_moe_down-53): 198 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 430 us MUL_MAT_ID(ffn_moe_down-54): 214 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 486 us MUL_MAT_ID(ffn_moe_down-55): 271 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 736 us MUL_MAT_ID(ffn_moe_down-56): 333 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 378 us MUL_MAT_ID(ffn_moe_down-57): 206 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 417 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 394 us MUL_MAT_ID(ffn_moe_down-59): 226 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 423 us MUL_MAT_ID(ffn_moe_down-60): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 440 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 413 us MUL_MAT_ID(ffn_moe_down-62): 304 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 391 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 336 us MUL_MAT_ID(ffn_moe_down-64): 179 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 396 us MUL_MAT_ID(ffn_moe_down-65): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 346 us MUL_MAT_ID(ffn_moe_down-66): 197 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 394 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 673 us MUL_MAT_ID(ffn_moe_down-68): 323 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 376 us MUL_MAT_ID(ffn_moe_down-69): 434 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 391 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 451 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 354 us MUL_MAT_ID(ffn_moe_down-72): 254 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 428 us MUL_MAT_ID(ffn_moe_down-73): 211 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 411 us MUL_MAT_ID(ffn_moe_down-74): 220 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 360 us MUL_MAT_ID(ffn_moe_down-75): 200 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 386 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 358 us MUL_MAT_ID(ffn_moe_down-77): 300 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 418 us MUL_MAT_ID(ffn_moe_down-78): 212 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 426 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 367 us MUL_MAT_ID(ffn_moe_down-80): 189 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 413 us MUL_MAT_ID(ffn_moe_down-81): 583 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 203 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 456 us MUL_MAT_ID(ffn_moe_down-83): 185 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 387 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 396 us MUL_MAT_ID(ffn_moe_down-85): 237 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 398 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 444 us MUL_MAT_ID(ffn_moe_down-87): 208 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 448 us MUL_MAT_ID(ffn_moe_down-88): 208 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 387 us MUL_MAT_ID(ffn_moe_down-89): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 409 us MUL_MAT_ID(ffn_moe_down-90): 257 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 397 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 36 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 376 us MUL_MAT_ID(ffn_moe_down-25): 202 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 393 us MUL_MAT_ID(ffn_moe_down-26): 186 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 357 us MUL_MAT_ID(ffn_moe_down-27): 232 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 376 us MUL_MAT_ID(ffn_moe_down-28): 191 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 391 us MUL_MAT_ID(ffn_moe_down-29): 227 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 330 us MUL_MAT_ID(ffn_moe_down-30): 183 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 354 us MUL_MAT_ID(ffn_moe_down-31): 187 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 393 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 386 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 189 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 434 us MUL_MAT_ID(ffn_moe_down-35): 182 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 381 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 385 us MUL_MAT_ID(ffn_moe_down-37): 225 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 360 us MUL_MAT_ID(ffn_moe_down-38): 189 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 359 us MUL_MAT_ID(ffn_moe_down-39): 672 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 353 us MUL_MAT_ID(ffn_moe_down-40): 677 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 395 us MUL_MAT_ID(ffn_moe_down-41): 206 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 377 us MUL_MAT_ID(ffn_moe_down-42): 183 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 353 us MUL_MAT_ID(ffn_moe_down-43): 384 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 398 us MUL_MAT_ID(ffn_moe_down-44): 202 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 354 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 392 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 401 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 366 us MUL_MAT_ID(ffn_moe_down-48): 226 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 421 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 371 us MUL_MAT_ID(ffn_moe_down-50): 204 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 402 us MUL_MAT_ID(ffn_moe_down-51): 188 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 396 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 438 us MUL_MAT_ID(ffn_moe_down-53): 549 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 407 us MUL_MAT_ID(ffn_moe_down-54): 221 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 379 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 407 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 582 us MUL_MAT_ID(ffn_moe_down-57): 218 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 398 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 376 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 427 us MUL_MAT_ID(ffn_moe_down-60): 187 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 447 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 392 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 682 us MUL_MAT_ID(ffn_moe_down-63): 262 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 657 us MUL_MAT_ID(ffn_moe_down-64): 270 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 394 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 433 us MUL_MAT_ID(ffn_moe_down-66): 206 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 424 us MUL_MAT_ID(ffn_moe_down-67): 189 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 410 us MUL_MAT_ID(ffn_moe_down-68): 209 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 433 us MUL_MAT_ID(ffn_moe_down-69): 238 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 795 us MUL_MAT_ID(ffn_moe_down-70): 360 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 365 us MUL_MAT_ID(ffn_moe_down-71): 213 us ggml_barrier(...): 391 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 346 us MUL_MAT_ID(ffn_moe_down-72): 209 us ggml_barrier(...): 509 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 353 us MUL_MAT_ID(ffn_moe_down-73): 195 us ggml_barrier(...): 379 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 430 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 399 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 435 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 433 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 353 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 417 us MUL_MAT_ID(ffn_moe_down-79): 214 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 401 us MUL_MAT_ID(ffn_moe_down-80): 664 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 484 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 397 us MUL_MAT_ID(ffn_moe_down-82): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 400 us MUL_MAT_ID(ffn_moe_down-83): 212 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 363 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 408 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 393 us MUL_MAT_ID(ffn_moe_down-86): 237 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 524 us MUL_MAT_ID(ffn_moe_down-87): 206 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 401 us MUL_MAT_ID(ffn_moe_down-88): 552 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 385 us MUL_MAT_ID(ffn_moe_down-89): 205 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 419 us MUL_MAT_ID(ffn_moe_down-90): 223 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 416 us MUL_MAT_ID(ffn_moe_down-91): 259 us ggml_barrier(...): 10 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 378 us MUL_MAT_ID(ffn_moe_down-25): 198 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 422 us MUL_MAT_ID(ffn_moe_down-26): 200 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 381 us MUL_MAT_ID(ffn_moe_down-27): 209 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 452 us MUL_MAT_ID(ffn_moe_down-28): 190 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 369 us MUL_MAT_ID(ffn_moe_down-29): 396 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 441 us MUL_MAT_ID(ffn_moe_down-31): 222 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 432 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 376 us MUL_MAT_ID(ffn_moe_down-33): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 400 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 419 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 414 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 392 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 380 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 398 us MUL_MAT_ID(ffn_moe_down-39): 253 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 428 us MUL_MAT_ID(ffn_moe_down-40): 281 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 391 us MUL_MAT_ID(ffn_moe_down-41): 224 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 393 us MUL_MAT_ID(ffn_moe_down-42): 188 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 414 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 366 us MUL_MAT_ID(ffn_moe_down-44): 186 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 466 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 331 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 404 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 438 us MUL_MAT_ID(ffn_moe_down-48): 164 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 404 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 369 us MUL_MAT_ID(ffn_moe_down-50): 229 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 420 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 377 us MUL_MAT_ID(ffn_moe_down-52): 237 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 411 us MUL_MAT_ID(ffn_moe_down-53): 255 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 381 us MUL_MAT_ID(ffn_moe_down-54): 212 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 361 us MUL_MAT_ID(ffn_moe_down-55): 187 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 375 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 364 us MUL_MAT_ID(ffn_moe_down-57): 186 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 389 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 392 us MUL_MAT_ID(ffn_moe_down-59): 208 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 347 us MUL_MAT_ID(ffn_moe_down-60): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 376 us MUL_MAT_ID(ffn_moe_down-61): 554 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 353 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 393 us MUL_MAT_ID(ffn_moe_down-63): 200 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 376 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 368 us MUL_MAT_ID(ffn_moe_down-65): 187 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 346 us MUL_MAT_ID(ffn_moe_down-66): 211 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 415 us MUL_MAT_ID(ffn_moe_down-67): 191 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 356 us MUL_MAT_ID(ffn_moe_down-68): 341 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 375 us MUL_MAT_ID(ffn_moe_down-69): 413 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 430 us MUL_MAT_ID(ffn_moe_down-70): 264 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 444 us MUL_MAT_ID(ffn_moe_down-71): 212 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 363 us MUL_MAT_ID(ffn_moe_down-72): 219 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 202 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 394 us MUL_MAT_ID(ffn_moe_down-74): 247 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 379 us MUL_MAT_ID(ffn_moe_down-75): 214 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 402 us MUL_MAT_ID(ffn_moe_down-76): 204 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 419 us MUL_MAT_ID(ffn_moe_down-77): 308 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 409 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 372 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 390 us MUL_MAT_ID(ffn_moe_down-80): 227 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 417 us MUL_MAT_ID(ffn_moe_down-81): 247 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 443 us MUL_MAT_ID(ffn_moe_down-82): 204 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 670 us MUL_MAT_ID(ffn_moe_down-83): 345 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 419 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 417 us MUL_MAT_ID(ffn_moe_down-85): 252 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 404 us MUL_MAT_ID(ffn_moe_down-86): 222 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 422 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 356 us MUL_MAT_ID(ffn_moe_down-88): 207 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 430 us MUL_MAT_ID(ffn_moe_down-89): 426 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 414 us MUL_MAT_ID(ffn_moe_down-90): 247 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 438 us MUL_MAT_ID(ffn_moe_down-91): 254 us ggml_barrier(...): 6 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 406 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 409 us MUL_MAT_ID(ffn_moe_down-26): 263 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 379 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 474 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 463 us MUL_MAT_ID(ffn_moe_down-29): 209 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 335 us MUL_MAT_ID(ffn_moe_down-30): 204 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 445 us MUL_MAT_ID(ffn_moe_down-31): 321 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 465 us MUL_MAT_ID(ffn_moe_down-32): 199 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 404 us MUL_MAT_ID(ffn_moe_down-33): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 419 us MUL_MAT_ID(ffn_moe_down-34): 213 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 398 us MUL_MAT_ID(ffn_moe_down-35): 202 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 445 us MUL_MAT_ID(ffn_moe_down-36): 211 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 370 us MUL_MAT_ID(ffn_moe_down-37): 267 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 588 us MUL_MAT_ID(ffn_moe_down-38): 344 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 367 us MUL_MAT_ID(ffn_moe_down-39): 484 us ggml_barrier(...): 406 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 369 us MUL_MAT_ID(ffn_moe_down-40): 532 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 389 us MUL_MAT_ID(ffn_moe_down-41): 225 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 431 us MUL_MAT_ID(ffn_moe_down-42): 189 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 439 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 419 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 403 us MUL_MAT_ID(ffn_moe_down-45): 193 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 336 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 370 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 402 us MUL_MAT_ID(ffn_moe_down-48): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 388 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 359 us MUL_MAT_ID(ffn_moe_down-50): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 343 us MUL_MAT_ID(ffn_moe_down-51): 191 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 382 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 393 us MUL_MAT_ID(ffn_moe_down-53): 411 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 333 us MUL_MAT_ID(ffn_moe_down-54): 184 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 385 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 356 us MUL_MAT_ID(ffn_moe_down-56): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 402 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 402 us MUL_MAT_ID(ffn_moe_down-58): 218 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 385 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 362 us MUL_MAT_ID(ffn_moe_down-60): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 379 us MUL_MAT_ID(ffn_moe_down-61): 360 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 661 us MUL_MAT_ID(ffn_moe_down-62): 295 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 364 us MUL_MAT_ID(ffn_moe_down-63): 194 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 404 us MUL_MAT_ID(ffn_moe_down-64): 187 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 390 us MUL_MAT_ID(ffn_moe_down-65): 398 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 395 us MUL_MAT_ID(ffn_moe_down-66): 218 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 401 us MUL_MAT_ID(ffn_moe_down-67): 312 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 369 us MUL_MAT_ID(ffn_moe_down-68): 214 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 389 us MUL_MAT_ID(ffn_moe_down-69): 333 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 386 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 393 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 337 us MUL_MAT_ID(ffn_moe_down-72): 581 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 406 us MUL_MAT_ID(ffn_moe_down-73): 434 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 386 us MUL_MAT_ID(ffn_moe_down-74): 407 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 358 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 423 us MUL_MAT_ID(ffn_moe_down-76): 269 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 374 us MUL_MAT_ID(ffn_moe_down-77): 199 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 430 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 409 us MUL_MAT_ID(ffn_moe_down-79): 186 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 360 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 375 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 352 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 404 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 420 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 425 us MUL_MAT_ID(ffn_moe_down-85): 417 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 377 us MUL_MAT_ID(ffn_moe_down-86): 184 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 370 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 370 us MUL_MAT_ID(ffn_moe_down-88): 182 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 393 us MUL_MAT_ID(ffn_moe_down-89): 392 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 406 us MUL_MAT_ID(ffn_moe_down-90): 238 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 404 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 64 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 389 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 380 us MUL_MAT_ID(ffn_moe_down-26): 183 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 405 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 429 us MUL_MAT_ID(ffn_moe_down-28): 271 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 412 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 399 us MUL_MAT_ID(ffn_moe_down-30): 277 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 399 us MUL_MAT_ID(ffn_moe_down-31): 204 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 400 us MUL_MAT_ID(ffn_moe_down-32): 249 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 418 us MUL_MAT_ID(ffn_moe_down-33): 209 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 400 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 425 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 405 us MUL_MAT_ID(ffn_moe_down-36): 372 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 397 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 393 us MUL_MAT_ID(ffn_moe_down-38): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 402 us MUL_MAT_ID(ffn_moe_down-39): 464 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 355 us MUL_MAT_ID(ffn_moe_down-40): 253 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 398 us MUL_MAT_ID(ffn_moe_down-41): 395 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 377 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 404 us MUL_MAT_ID(ffn_moe_down-43): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 380 us MUL_MAT_ID(ffn_moe_down-44): 207 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 449 us MUL_MAT_ID(ffn_moe_down-45): 202 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 345 us MUL_MAT_ID(ffn_moe_down-46): 363 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 379 us MUL_MAT_ID(ffn_moe_down-47): 187 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 411 us MUL_MAT_ID(ffn_moe_down-48): 399 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 360 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 407 us MUL_MAT_ID(ffn_moe_down-50): 217 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 419 us MUL_MAT_ID(ffn_moe_down-51): 185 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 433 us MUL_MAT_ID(ffn_moe_down-52): 203 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 409 us MUL_MAT_ID(ffn_moe_down-53): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 390 us MUL_MAT_ID(ffn_moe_down-54): 204 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 390 us MUL_MAT_ID(ffn_moe_down-55): 300 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 389 us MUL_MAT_ID(ffn_moe_down-56): 333 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 386 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 421 us MUL_MAT_ID(ffn_moe_down-58): 202 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 411 us MUL_MAT_ID(ffn_moe_down-59): 204 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 379 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 367 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 226 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 438 us MUL_MAT_ID(ffn_moe_down-63): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 429 us MUL_MAT_ID(ffn_moe_down-64): 211 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 325 us MUL_MAT_ID(ffn_moe_down-65): 206 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 385 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 379 us MUL_MAT_ID(ffn_moe_down-67): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 395 us MUL_MAT_ID(ffn_moe_down-68): 325 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 400 us MUL_MAT_ID(ffn_moe_down-69): 209 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 419 us MUL_MAT_ID(ffn_moe_down-70): 402 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 401 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 329 us MUL_MAT_ID(ffn_moe_down-72): 454 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 399 us MUL_MAT_ID(ffn_moe_down-73): 196 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 521 us MUL_MAT_ID(ffn_moe_down-74): 192 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 444 us MUL_MAT_ID(ffn_moe_down-75): 291 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 410 us MUL_MAT_ID(ffn_moe_down-76): 337 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 384 us MUL_MAT_ID(ffn_moe_down-77): 200 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 384 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 396 us MUL_MAT_ID(ffn_moe_down-79): 293 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 367 us MUL_MAT_ID(ffn_moe_down-80): 210 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 374 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 399 us MUL_MAT_ID(ffn_moe_down-82): 527 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 418 us MUL_MAT_ID(ffn_moe_down-83): 208 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 373 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 401 us MUL_MAT_ID(ffn_moe_down-85): 305 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 367 us MUL_MAT_ID(ffn_moe_down-86): 213 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 396 us MUL_MAT_ID(ffn_moe_down-87): 200 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 380 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 386 us MUL_MAT_ID(ffn_moe_down-89): 211 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 437 us MUL_MAT_ID(ffn_moe_down-90): 256 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 366 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 79 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 390 us MUL_MAT_ID(ffn_moe_down-25): 203 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 354 us MUL_MAT_ID(ffn_moe_down-26): 199 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 406 us MUL_MAT_ID(ffn_moe_down-27): 565 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 398 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 389 us MUL_MAT_ID(ffn_moe_down-29): 176 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 353 us MUL_MAT_ID(ffn_moe_down-30): 214 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 428 us MUL_MAT_ID(ffn_moe_down-31): 300 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 372 us MUL_MAT_ID(ffn_moe_down-32): 266 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 470 us MUL_MAT_ID(ffn_moe_down-33): 208 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 380 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 551 us MUL_MAT_ID(ffn_moe_down-35): 322 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 386 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 412 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 423 us MUL_MAT_ID(ffn_moe_down-38): 418 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 364 us MUL_MAT_ID(ffn_moe_down-39): 242 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 448 us MUL_MAT_ID(ffn_moe_down-40): 258 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 382 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 439 us MUL_MAT_ID(ffn_moe_down-42): 203 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 416 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 362 us MUL_MAT_ID(ffn_moe_down-44): 234 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 426 us MUL_MAT_ID(ffn_moe_down-45): 210 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 399 us MUL_MAT_ID(ffn_moe_down-46): 391 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 401 us MUL_MAT_ID(ffn_moe_down-47): 370 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 439 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 443 us MUL_MAT_ID(ffn_moe_down-49): 394 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 410 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 363 us MUL_MAT_ID(ffn_moe_down-51): 369 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 417 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 423 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 422 us MUL_MAT_ID(ffn_moe_down-54): 203 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 429 us MUL_MAT_ID(ffn_moe_down-55): 207 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 409 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 404 us MUL_MAT_ID(ffn_moe_down-57): 203 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 384 us MUL_MAT_ID(ffn_moe_down-58): 219 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 407 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 448 us MUL_MAT_ID(ffn_moe_down-60): 365 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 387 us MUL_MAT_ID(ffn_moe_down-61): 181 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 395 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 402 us MUL_MAT_ID(ffn_moe_down-63): 267 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 440 us MUL_MAT_ID(ffn_moe_down-64): 539 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 419 us MUL_MAT_ID(ffn_moe_down-65): 393 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 413 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 437 us MUL_MAT_ID(ffn_moe_down-67): 226 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 445 us MUL_MAT_ID(ffn_moe_down-68): 283 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 439 us MUL_MAT_ID(ffn_moe_down-69): 324 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 556 us MUL_MAT_ID(ffn_moe_down-70): 194 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 402 us MUL_MAT_ID(ffn_moe_down-71): 310 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 497 us MUL_MAT_ID(ffn_moe_down-72): 284 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 405 us MUL_MAT_ID(ffn_moe_down-73): 201 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 424 us MUL_MAT_ID(ffn_moe_down-74): 408 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 392 us MUL_MAT_ID(ffn_moe_down-75): 317 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 407 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 410 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 380 us MUL_MAT_ID(ffn_moe_down-78): 204 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 381 us MUL_MAT_ID(ffn_moe_down-79): 296 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 389 us MUL_MAT_ID(ffn_moe_down-80): 206 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 394 us MUL_MAT_ID(ffn_moe_down-81): 206 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 410 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 510 us MUL_MAT_ID(ffn_moe_down-83): 203 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 379 us MUL_MAT_ID(ffn_moe_down-84): 336 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 422 us MUL_MAT_ID(ffn_moe_down-85): 465 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 354 us MUL_MAT_ID(ffn_moe_down-86): 200 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 386 us MUL_MAT_ID(ffn_moe_down-87): 201 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 440 us MUL_MAT_ID(ffn_moe_down-88): 219 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 398 us MUL_MAT_ID(ffn_moe_down-89): 174 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 413 us MUL_MAT_ID(ffn_moe_down-90): 248 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 386 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 27 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 376 us MUL_MAT_ID(ffn_moe_down-25): 226 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 374 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 375 us MUL_MAT_ID(ffn_moe_down-27): 223 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 392 us MUL_MAT_ID(ffn_moe_down-28): 186 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 374 us MUL_MAT_ID(ffn_moe_down-29): 187 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 438 us MUL_MAT_ID(ffn_moe_down-30): 406 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 390 us MUL_MAT_ID(ffn_moe_down-31): 183 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 449 us MUL_MAT_ID(ffn_moe_down-32): 183 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 387 us MUL_MAT_ID(ffn_moe_down-33): 247 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 449 us MUL_MAT_ID(ffn_moe_down-34): 217 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 658 us MUL_MAT_ID(ffn_moe_down-35): 259 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 661 us MUL_MAT_ID(ffn_moe_down-36): 180 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 421 us MUL_MAT_ID(ffn_moe_down-37): 432 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 390 us MUL_MAT_ID(ffn_moe_down-38): 197 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 401 us MUL_MAT_ID(ffn_moe_down-39): 261 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 419 us MUL_MAT_ID(ffn_moe_down-40): 253 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 439 us MUL_MAT_ID(ffn_moe_down-41): 425 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 392 us MUL_MAT_ID(ffn_moe_down-42): 175 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 408 us MUL_MAT_ID(ffn_moe_down-43): 278 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 425 us MUL_MAT_ID(ffn_moe_down-44): 183 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 573 us MUL_MAT_ID(ffn_moe_down-45): 315 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 357 us MUL_MAT_ID(ffn_moe_down-46): 184 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 387 us MUL_MAT_ID(ffn_moe_down-47): 197 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 377 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 429 us MUL_MAT_ID(ffn_moe_down-49): 187 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 357 us MUL_MAT_ID(ffn_moe_down-50): 201 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 402 us MUL_MAT_ID(ffn_moe_down-51): 326 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 398 us MUL_MAT_ID(ffn_moe_down-52): 199 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 510 us MUL_MAT_ID(ffn_moe_down-53): 225 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 509 us MUL_MAT_ID(ffn_moe_down-54): 551 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 358 us MUL_MAT_ID(ffn_moe_down-55): 249 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 402 us MUL_MAT_ID(ffn_moe_down-56): 189 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 380 us MUL_MAT_ID(ffn_moe_down-57): 259 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 423 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 353 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 374 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 358 us MUL_MAT_ID(ffn_moe_down-61): 202 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 390 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 371 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 375 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 406 us MUL_MAT_ID(ffn_moe_down-65): 188 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 347 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 359 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 392 us MUL_MAT_ID(ffn_moe_down-68): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 387 us MUL_MAT_ID(ffn_moe_down-69): 198 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 359 us MUL_MAT_ID(ffn_moe_down-70): 185 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 397 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 359 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 386 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 393 us MUL_MAT_ID(ffn_moe_down-74): 204 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 354 us MUL_MAT_ID(ffn_moe_down-75): 540 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 342 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 412 us MUL_MAT_ID(ffn_moe_down-77): 190 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 386 us MUL_MAT_ID(ffn_moe_down-78): 185 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 391 us MUL_MAT_ID(ffn_moe_down-79): 273 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 378 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 414 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 351 us MUL_MAT_ID(ffn_moe_down-82): 189 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 377 us MUL_MAT_ID(ffn_moe_down-83): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 360 us MUL_MAT_ID(ffn_moe_down-84): 243 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 429 us MUL_MAT_ID(ffn_moe_down-85): 257 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 390 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 427 us MUL_MAT_ID(ffn_moe_down-87): 405 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 395 us MUL_MAT_ID(ffn_moe_down-88): 186 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 371 us MUL_MAT_ID(ffn_moe_down-89): 376 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 375 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 419 us MUL_MAT_ID(ffn_moe_down-91): 288 us ggml_barrier(...): 179 us GET_ROWS(inp_embd): 5 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 380 us MUL_MAT_ID(ffn_moe_down-25): 205 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 424 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 445 us MUL_MAT_ID(ffn_moe_down-27): 186 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 347 us MUL_MAT_ID(ffn_moe_down-28): 423 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 405 us MUL_MAT_ID(ffn_moe_down-29): 209 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 430 us MUL_MAT_ID(ffn_moe_down-30): 198 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 388 us MUL_MAT_ID(ffn_moe_down-31): 583 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 385 us MUL_MAT_ID(ffn_moe_down-32): 219 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 432 us MUL_MAT_ID(ffn_moe_down-33): 580 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 403 us MUL_MAT_ID(ffn_moe_down-34): 386 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 456 us MUL_MAT_ID(ffn_moe_down-35): 204 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 407 us MUL_MAT_ID(ffn_moe_down-36): 247 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 410 us MUL_MAT_ID(ffn_moe_down-37): 266 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 415 us MUL_MAT_ID(ffn_moe_down-38): 204 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 465 us MUL_MAT_ID(ffn_moe_down-39): 266 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 422 us MUL_MAT_ID(ffn_moe_down-40): 235 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 435 us MUL_MAT_ID(ffn_moe_down-41): 216 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 415 us MUL_MAT_ID(ffn_moe_down-42): 227 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 368 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 375 us MUL_MAT_ID(ffn_moe_down-44): 216 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 416 us MUL_MAT_ID(ffn_moe_down-45): 187 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 361 us MUL_MAT_ID(ffn_moe_down-46): 251 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 431 us MUL_MAT_ID(ffn_moe_down-47): 216 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 659 us MUL_MAT_ID(ffn_moe_down-48): 268 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 344 us MUL_MAT_ID(ffn_moe_down-49): 300 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 421 us MUL_MAT_ID(ffn_moe_down-50): 373 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 391 us MUL_MAT_ID(ffn_moe_down-51): 180 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 395 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 354 us MUL_MAT_ID(ffn_moe_down-53): 187 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 403 us MUL_MAT_ID(ffn_moe_down-54): 279 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 405 us MUL_MAT_ID(ffn_moe_down-55): 189 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 216 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 391 us MUL_MAT_ID(ffn_moe_down-57): 183 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 379 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 459 us MUL_MAT_ID(ffn_moe_down-59): 326 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 368 us MUL_MAT_ID(ffn_moe_down-60): 185 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 386 us MUL_MAT_ID(ffn_moe_down-61): 166 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 424 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 397 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 342 us MUL_MAT_ID(ffn_moe_down-64): 208 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 364 us MUL_MAT_ID(ffn_moe_down-65): 184 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 373 us MUL_MAT_ID(ffn_moe_down-66): 210 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 370 us MUL_MAT_ID(ffn_moe_down-67): 204 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 435 us MUL_MAT_ID(ffn_moe_down-68): 223 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 409 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 345 us MUL_MAT_ID(ffn_moe_down-70): 185 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 386 us MUL_MAT_ID(ffn_moe_down-71): 277 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 359 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 395 us MUL_MAT_ID(ffn_moe_down-73): 197 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 347 us MUL_MAT_ID(ffn_moe_down-74): 184 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 431 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 384 us MUL_MAT_ID(ffn_moe_down-76): 212 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 401 us MUL_MAT_ID(ffn_moe_down-77): 417 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 374 us MUL_MAT_ID(ffn_moe_down-78): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 391 us MUL_MAT_ID(ffn_moe_down-79): 252 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 437 us MUL_MAT_ID(ffn_moe_down-80): 344 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 384 us MUL_MAT_ID(ffn_moe_down-81): 207 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 430 us MUL_MAT_ID(ffn_moe_down-82): 206 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 438 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 416 us MUL_MAT_ID(ffn_moe_down-84): 243 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 428 us MUL_MAT_ID(ffn_moe_down-85): 412 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 356 us MUL_MAT_ID(ffn_moe_down-86): 195 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 449 us MUL_MAT_ID(ffn_moe_down-87): 201 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 454 us MUL_MAT_ID(ffn_moe_down-88): 198 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 428 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 462 us MUL_MAT_ID(ffn_moe_down-90): 263 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 398 us MUL_MAT_ID(ffn_moe_down-91): 249 us ggml_barrier(...): 36 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 368 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 432 us MUL_MAT_ID(ffn_moe_down-26): 216 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 402 us MUL_MAT_ID(ffn_moe_down-27): 262 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 449 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 433 us MUL_MAT_ID(ffn_moe_down-29): 228 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 445 us MUL_MAT_ID(ffn_moe_down-30): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 415 us MUL_MAT_ID(ffn_moe_down-31): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 458 us MUL_MAT_ID(ffn_moe_down-32): 446 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 413 us MUL_MAT_ID(ffn_moe_down-33): 212 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 443 us MUL_MAT_ID(ffn_moe_down-34): 473 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 437 us MUL_MAT_ID(ffn_moe_down-35): 189 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 437 us MUL_MAT_ID(ffn_moe_down-36): 192 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 444 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 416 us MUL_MAT_ID(ffn_moe_down-38): 196 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 402 us MUL_MAT_ID(ffn_moe_down-39): 246 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 439 us MUL_MAT_ID(ffn_moe_down-40): 240 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 466 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 452 us MUL_MAT_ID(ffn_moe_down-42): 557 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 402 us MUL_MAT_ID(ffn_moe_down-43): 485 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 354 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 441 us MUL_MAT_ID(ffn_moe_down-45): 203 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 403 us MUL_MAT_ID(ffn_moe_down-46): 203 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 490 us MUL_MAT_ID(ffn_moe_down-47): 212 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 408 us MUL_MAT_ID(ffn_moe_down-48): 204 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 394 us MUL_MAT_ID(ffn_moe_down-49): 204 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 386 us MUL_MAT_ID(ffn_moe_down-50): 197 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 437 us MUL_MAT_ID(ffn_moe_down-51): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 427 us MUL_MAT_ID(ffn_moe_down-52): 202 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 435 us MUL_MAT_ID(ffn_moe_down-53): 205 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 442 us MUL_MAT_ID(ffn_moe_down-54): 185 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 423 us MUL_MAT_ID(ffn_moe_down-55): 432 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 383 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 436 us MUL_MAT_ID(ffn_moe_down-57): 238 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 382 us MUL_MAT_ID(ffn_moe_down-58): 210 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 440 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 382 us MUL_MAT_ID(ffn_moe_down-60): 265 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 409 us MUL_MAT_ID(ffn_moe_down-61): 230 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 744 us MUL_MAT_ID(ffn_moe_down-62): 321 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 378 us MUL_MAT_ID(ffn_moe_down-63): 202 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 396 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 400 us MUL_MAT_ID(ffn_moe_down-65): 188 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 451 us MUL_MAT_ID(ffn_moe_down-66): 203 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 406 us MUL_MAT_ID(ffn_moe_down-67): 424 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 396 us MUL_MAT_ID(ffn_moe_down-68): 427 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 388 us MUL_MAT_ID(ffn_moe_down-69): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 391 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 366 us MUL_MAT_ID(ffn_moe_down-71): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 406 us MUL_MAT_ID(ffn_moe_down-72): 206 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 439 us MUL_MAT_ID(ffn_moe_down-73): 204 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 612 us MUL_MAT_ID(ffn_moe_down-74): 240 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 360 us MUL_MAT_ID(ffn_moe_down-75): 215 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 437 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 420 us MUL_MAT_ID(ffn_moe_down-77): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 383 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 392 us MUL_MAT_ID(ffn_moe_down-79): 190 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 433 us MUL_MAT_ID(ffn_moe_down-80): 209 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 379 us MUL_MAT_ID(ffn_moe_down-81): 183 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 359 us MUL_MAT_ID(ffn_moe_down-82): 202 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 390 us MUL_MAT_ID(ffn_moe_down-83): 188 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 457 us MUL_MAT_ID(ffn_moe_down-84): 659 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 361 us MUL_MAT_ID(ffn_moe_down-85): 683 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 462 us MUL_MAT_ID(ffn_moe_down-86): 233 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 651 us MUL_MAT_ID(ffn_moe_down-87): 276 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 394 us MUL_MAT_ID(ffn_moe_down-88): 238 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 402 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 402 us MUL_MAT_ID(ffn_moe_down-90): 251 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 586 us MUL_MAT_ID(ffn_moe_down-91): 636 us ggml_barrier(...): 191 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 415 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 490 us MUL_MAT_ID(ffn_moe_down-26): 188 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 414 us MUL_MAT_ID(ffn_moe_down-27): 393 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 432 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 430 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 454 us MUL_MAT_ID(ffn_moe_down-30): 280 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 384 us MUL_MAT_ID(ffn_moe_down-31): 268 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 351 us MUL_MAT_ID(ffn_moe_down-32): 204 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 439 us MUL_MAT_ID(ffn_moe_down-33): 536 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 372 us MUL_MAT_ID(ffn_moe_down-34): 217 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 410 us MUL_MAT_ID(ffn_moe_down-35): 202 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 363 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 407 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 363 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 417 us MUL_MAT_ID(ffn_moe_down-39): 544 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 383 us MUL_MAT_ID(ffn_moe_down-40): 490 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 390 us MUL_MAT_ID(ffn_moe_down-41): 226 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 387 us MUL_MAT_ID(ffn_moe_down-42): 202 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 381 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 387 us MUL_MAT_ID(ffn_moe_down-44): 252 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 411 us MUL_MAT_ID(ffn_moe_down-45): 214 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 379 us MUL_MAT_ID(ffn_moe_down-46): 186 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 382 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 343 us MUL_MAT_ID(ffn_moe_down-48): 183 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 363 us MUL_MAT_ID(ffn_moe_down-49): 313 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 411 us MUL_MAT_ID(ffn_moe_down-50): 205 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 360 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 392 us MUL_MAT_ID(ffn_moe_down-52): 212 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 372 us MUL_MAT_ID(ffn_moe_down-53): 213 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 396 us MUL_MAT_ID(ffn_moe_down-54): 409 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 395 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 425 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 386 us MUL_MAT_ID(ffn_moe_down-57): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 407 us MUL_MAT_ID(ffn_moe_down-58): 187 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 377 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 407 us MUL_MAT_ID(ffn_moe_down-60): 197 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 373 us MUL_MAT_ID(ffn_moe_down-61): 210 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 360 us MUL_MAT_ID(ffn_moe_down-62): 397 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 344 us MUL_MAT_ID(ffn_moe_down-63): 187 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 379 us MUL_MAT_ID(ffn_moe_down-64): 426 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 395 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 375 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 342 us MUL_MAT_ID(ffn_moe_down-67): 283 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 371 us MUL_MAT_ID(ffn_moe_down-68): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 400 us MUL_MAT_ID(ffn_moe_down-69): 216 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 424 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 403 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 343 us MUL_MAT_ID(ffn_moe_down-72): 192 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 430 us MUL_MAT_ID(ffn_moe_down-73): 211 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 415 us MUL_MAT_ID(ffn_moe_down-74): 224 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 810 us MUL_MAT_ID(ffn_moe_down-75): 341 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 367 us MUL_MAT_ID(ffn_moe_down-76): 212 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 424 us MUL_MAT_ID(ffn_moe_down-77): 205 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 383 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 412 us MUL_MAT_ID(ffn_moe_down-79): 186 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 409 us MUL_MAT_ID(ffn_moe_down-80): 214 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 400 us MUL_MAT_ID(ffn_moe_down-81): 208 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 434 us MUL_MAT_ID(ffn_moe_down-82): 418 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 407 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 359 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 401 us MUL_MAT_ID(ffn_moe_down-85): 237 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 434 us MUL_MAT_ID(ffn_moe_down-86): 295 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 537 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 394 us MUL_MAT_ID(ffn_moe_down-88): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 343 us MUL_MAT_ID(ffn_moe_down-89): 385 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 374 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 373 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 55 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 384 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 406 us MUL_MAT_ID(ffn_moe_down-26): 209 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 500 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 398 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 354 us MUL_MAT_ID(ffn_moe_down-29): 238 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 413 us MUL_MAT_ID(ffn_moe_down-30): 225 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 424 us MUL_MAT_ID(ffn_moe_down-31): 325 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 377 us MUL_MAT_ID(ffn_moe_down-32): 420 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 357 us MUL_MAT_ID(ffn_moe_down-33): 205 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 405 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 403 us MUL_MAT_ID(ffn_moe_down-35): 231 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 399 us MUL_MAT_ID(ffn_moe_down-36): 225 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 441 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 421 us MUL_MAT_ID(ffn_moe_down-38): 265 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 419 us MUL_MAT_ID(ffn_moe_down-39): 223 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 398 us MUL_MAT_ID(ffn_moe_down-40): 240 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 411 us MUL_MAT_ID(ffn_moe_down-41): 336 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 433 us MUL_MAT_ID(ffn_moe_down-42): 231 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 470 us MUL_MAT_ID(ffn_moe_down-43): 435 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 389 us MUL_MAT_ID(ffn_moe_down-44): 181 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 423 us MUL_MAT_ID(ffn_moe_down-45): 429 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 378 us MUL_MAT_ID(ffn_moe_down-46): 225 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 436 us MUL_MAT_ID(ffn_moe_down-47): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 551 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 436 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 355 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 388 us MUL_MAT_ID(ffn_moe_down-51): 203 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 385 us MUL_MAT_ID(ffn_moe_down-52): 427 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 487 us MUL_MAT_ID(ffn_moe_down-53): 201 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 414 us MUL_MAT_ID(ffn_moe_down-54): 232 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 419 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 421 us MUL_MAT_ID(ffn_moe_down-56): 205 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 464 us MUL_MAT_ID(ffn_moe_down-57): 193 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 436 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 440 us MUL_MAT_ID(ffn_moe_down-59): 222 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 454 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 375 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 352 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 359 us MUL_MAT_ID(ffn_moe_down-63): 329 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 429 us MUL_MAT_ID(ffn_moe_down-64): 202 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 406 us MUL_MAT_ID(ffn_moe_down-65): 213 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 380 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 437 us MUL_MAT_ID(ffn_moe_down-67): 304 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 435 us MUL_MAT_ID(ffn_moe_down-68): 575 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 383 us MUL_MAT_ID(ffn_moe_down-69): 201 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 396 us MUL_MAT_ID(ffn_moe_down-70): 211 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 414 us MUL_MAT_ID(ffn_moe_down-71): 191 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 201 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 356 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 396 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 473 us MUL_MAT_ID(ffn_moe_down-75): 323 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 437 us MUL_MAT_ID(ffn_moe_down-76): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 376 us MUL_MAT_ID(ffn_moe_down-77): 203 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 390 us MUL_MAT_ID(ffn_moe_down-78): 206 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 414 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 383 us MUL_MAT_ID(ffn_moe_down-80): 289 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 410 us MUL_MAT_ID(ffn_moe_down-81): 224 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 382 us MUL_MAT_ID(ffn_moe_down-82): 207 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 368 us MUL_MAT_ID(ffn_moe_down-83): 210 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 376 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 402 us MUL_MAT_ID(ffn_moe_down-85): 203 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 435 us MUL_MAT_ID(ffn_moe_down-86): 243 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 431 us MUL_MAT_ID(ffn_moe_down-87): 227 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 642 us MUL_MAT_ID(ffn_moe_down-88): 338 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 330 us MUL_MAT_ID(ffn_moe_down-89): 216 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 417 us MUL_MAT_ID(ffn_moe_down-90): 237 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 409 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 39 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 392 us MUL_MAT_ID(ffn_moe_down-25): 391 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 355 us MUL_MAT_ID(ffn_moe_down-26): 567 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 492 us MUL_MAT_ID(ffn_moe_down-27): 213 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 385 us MUL_MAT_ID(ffn_moe_down-28): 183 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 393 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 358 us MUL_MAT_ID(ffn_moe_down-30): 203 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 409 us MUL_MAT_ID(ffn_moe_down-31): 187 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 388 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 431 us MUL_MAT_ID(ffn_moe_down-33): 448 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 507 us MUL_MAT_ID(ffn_moe_down-34): 289 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 421 us MUL_MAT_ID(ffn_moe_down-35): 430 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 497 us MUL_MAT_ID(ffn_moe_down-36): 205 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 425 us MUL_MAT_ID(ffn_moe_down-37): 330 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 349 us MUL_MAT_ID(ffn_moe_down-38): 219 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 495 us MUL_MAT_ID(ffn_moe_down-39): 259 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 394 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 511 us MUL_MAT_ID(ffn_moe_down-41): 216 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 389 us MUL_MAT_ID(ffn_moe_down-42): 206 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 471 us MUL_MAT_ID(ffn_moe_down-43): 271 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 418 us MUL_MAT_ID(ffn_moe_down-44): 241 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 393 us MUL_MAT_ID(ffn_moe_down-45): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 336 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 376 us MUL_MAT_ID(ffn_moe_down-47): 192 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 380 us MUL_MAT_ID(ffn_moe_down-48): 221 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 439 us MUL_MAT_ID(ffn_moe_down-49): 477 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 374 us MUL_MAT_ID(ffn_moe_down-50): 217 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 396 us MUL_MAT_ID(ffn_moe_down-51): 384 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 381 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 374 us MUL_MAT_ID(ffn_moe_down-53): 320 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 555 us MUL_MAT_ID(ffn_moe_down-54): 170 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 408 us MUL_MAT_ID(ffn_moe_down-55): 193 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 362 us MUL_MAT_ID(ffn_moe_down-56): 203 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 394 us MUL_MAT_ID(ffn_moe_down-57): 519 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 373 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 385 us MUL_MAT_ID(ffn_moe_down-59): 301 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 472 us MUL_MAT_ID(ffn_moe_down-60): 308 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 414 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 203 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 347 us MUL_MAT_ID(ffn_moe_down-63): 218 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 375 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 357 us MUL_MAT_ID(ffn_moe_down-65): 210 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 394 us MUL_MAT_ID(ffn_moe_down-66): 516 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 426 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 342 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 405 us MUL_MAT_ID(ffn_moe_down-68): 222 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 405 us MUL_MAT_ID(ffn_moe_down-69): 176 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 363 us MUL_MAT_ID(ffn_moe_down-70): 287 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 410 us MUL_MAT_ID(ffn_moe_down-71): 290 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 407 us MUL_MAT_ID(ffn_moe_down-72): 208 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 353 us MUL_MAT_ID(ffn_moe_down-73): 226 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 375 us MUL_MAT_ID(ffn_moe_down-74): 205 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 423 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 501 us MUL_MAT_ID(ffn_moe_down-76): 182 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 376 us MUL_MAT_ID(ffn_moe_down-77): 199 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 391 us MUL_MAT_ID(ffn_moe_down-78): 236 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 378 us MUL_MAT_ID(ffn_moe_down-79): 270 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 536 us MUL_MAT_ID(ffn_moe_down-80): 336 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 375 us MUL_MAT_ID(ffn_moe_down-81): 209 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 354 us MUL_MAT_ID(ffn_moe_down-82): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 357 us MUL_MAT_ID(ffn_moe_down-83): 188 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 371 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 518 us MUL_MAT_ID(ffn_moe_down-85): 242 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 387 us MUL_MAT_ID(ffn_moe_down-86): 200 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 386 us MUL_MAT_ID(ffn_moe_down-87): 209 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 430 us MUL_MAT_ID(ffn_moe_down-88): 187 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 483 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 431 us MUL_MAT_ID(ffn_moe_down-90): 270 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 444 us MUL_MAT_ID(ffn_moe_down-91): 283 us ggml_barrier(...): 253 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 364 us MUL_MAT_ID(ffn_moe_down-25): 198 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 416 us MUL_MAT_ID(ffn_moe_down-26): 204 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 392 us MUL_MAT_ID(ffn_moe_down-27): 542 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 447 us MUL_MAT_ID(ffn_moe_down-28): 257 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 508 us MUL_MAT_ID(ffn_moe_down-29): 168 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 459 us MUL_MAT_ID(ffn_moe_down-30): 182 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 353 us MUL_MAT_ID(ffn_moe_down-31): 182 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 345 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 404 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 406 us MUL_MAT_ID(ffn_moe_down-34): 185 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 340 us MUL_MAT_ID(ffn_moe_down-35): 185 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 422 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 339 us MUL_MAT_ID(ffn_moe_down-37): 534 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 392 us MUL_MAT_ID(ffn_moe_down-38): 202 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 404 us MUL_MAT_ID(ffn_moe_down-39): 223 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 376 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 386 us MUL_MAT_ID(ffn_moe_down-41): 552 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 401 us MUL_MAT_ID(ffn_moe_down-42): 184 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 389 us MUL_MAT_ID(ffn_moe_down-43): 207 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 387 us MUL_MAT_ID(ffn_moe_down-44): 206 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 449 us MUL_MAT_ID(ffn_moe_down-45): 235 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 370 us MUL_MAT_ID(ffn_moe_down-46): 348 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 392 us MUL_MAT_ID(ffn_moe_down-47): 211 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 392 us MUL_MAT_ID(ffn_moe_down-48): 187 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 426 us MUL_MAT_ID(ffn_moe_down-49): 210 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 393 us MUL_MAT_ID(ffn_moe_down-50): 223 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 420 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 420 us MUL_MAT_ID(ffn_moe_down-52): 427 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 431 us MUL_MAT_ID(ffn_moe_down-53): 436 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 357 us MUL_MAT_ID(ffn_moe_down-54): 216 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 442 us MUL_MAT_ID(ffn_moe_down-55): 204 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 423 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 463 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 457 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 343 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 417 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 369 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 358 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 358 us MUL_MAT_ID(ffn_moe_down-63): 183 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 362 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 398 us MUL_MAT_ID(ffn_moe_down-65): 200 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 402 us MUL_MAT_ID(ffn_moe_down-66): 187 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 387 us MUL_MAT_ID(ffn_moe_down-67): 209 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 440 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 382 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 388 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 390 us MUL_MAT_ID(ffn_moe_down-71): 203 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 345 us MUL_MAT_ID(ffn_moe_down-72): 247 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 456 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 375 us MUL_MAT_ID(ffn_moe_down-74): 221 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 407 us MUL_MAT_ID(ffn_moe_down-75): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 373 us MUL_MAT_ID(ffn_moe_down-76): 188 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 411 us MUL_MAT_ID(ffn_moe_down-77): 200 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 393 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 376 us MUL_MAT_ID(ffn_moe_down-79): 186 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 345 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 409 us MUL_MAT_ID(ffn_moe_down-81): 219 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 421 us MUL_MAT_ID(ffn_moe_down-82): 203 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 405 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 347 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 370 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 396 us MUL_MAT_ID(ffn_moe_down-86): 185 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 379 us MUL_MAT_ID(ffn_moe_down-87): 184 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 385 us MUL_MAT_ID(ffn_moe_down-88): 221 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 403 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 357 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 406 us MUL_MAT_ID(ffn_moe_down-91): 233 us ggml_barrier(...): 30 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 362 us MUL_MAT_ID(ffn_moe_down-25): 206 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 418 us MUL_MAT_ID(ffn_moe_down-26): 210 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 419 us MUL_MAT_ID(ffn_moe_down-27): 181 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 425 us MUL_MAT_ID(ffn_moe_down-28): 163 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 430 us MUL_MAT_ID(ffn_moe_down-29): 229 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 786 us MUL_MAT_ID(ffn_moe_down-30): 329 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 398 us MUL_MAT_ID(ffn_moe_down-31): 178 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 422 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 376 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 409 us MUL_MAT_ID(ffn_moe_down-34): 213 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 415 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 355 us MUL_MAT_ID(ffn_moe_down-37): 435 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 378 us MUL_MAT_ID(ffn_moe_down-38): 178 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 392 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 401 us MUL_MAT_ID(ffn_moe_down-40): 481 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 561 us MUL_MAT_ID(ffn_moe_down-41): 333 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 362 us MUL_MAT_ID(ffn_moe_down-42): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 417 us MUL_MAT_ID(ffn_moe_down-43): 192 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 356 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 430 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 392 us MUL_MAT_ID(ffn_moe_down-46): 430 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 395 us MUL_MAT_ID(ffn_moe_down-47): 203 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 416 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 344 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 406 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 411 us MUL_MAT_ID(ffn_moe_down-51): 202 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 374 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 351 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 390 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 428 us MUL_MAT_ID(ffn_moe_down-55): 444 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 430 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 429 us MUL_MAT_ID(ffn_moe_down-57): 208 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 401 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 393 us MUL_MAT_ID(ffn_moe_down-59): 202 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 360 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 366 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 400 us MUL_MAT_ID(ffn_moe_down-63): 484 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 419 us MUL_MAT_ID(ffn_moe_down-64): 201 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 356 us MUL_MAT_ID(ffn_moe_down-65): 223 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 383 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 350 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 410 us MUL_MAT_ID(ffn_moe_down-68): 187 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 409 us MUL_MAT_ID(ffn_moe_down-69): 231 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 423 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 457 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 331 us MUL_MAT_ID(ffn_moe_down-72): 249 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 433 us MUL_MAT_ID(ffn_moe_down-73): 206 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 416 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 400 us MUL_MAT_ID(ffn_moe_down-75): 199 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 384 us MUL_MAT_ID(ffn_moe_down-76): 265 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 421 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 418 us MUL_MAT_ID(ffn_moe_down-78): 503 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 372 us MUL_MAT_ID(ffn_moe_down-79): 210 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 346 us MUL_MAT_ID(ffn_moe_down-80): 184 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 394 us MUL_MAT_ID(ffn_moe_down-81): 184 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 404 us MUL_MAT_ID(ffn_moe_down-82): 187 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 410 us MUL_MAT_ID(ffn_moe_down-83): 203 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 384 us MUL_MAT_ID(ffn_moe_down-84): 324 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 376 us MUL_MAT_ID(ffn_moe_down-85): 234 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 369 us MUL_MAT_ID(ffn_moe_down-86): 204 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 425 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 349 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 355 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 417 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 398 us MUL_MAT_ID(ffn_moe_down-91): 402 us ggml_barrier(...): 29 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 387 us MUL_MAT_ID(ffn_moe_down-25): 207 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 420 us MUL_MAT_ID(ffn_moe_down-26): 185 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 459 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 427 us MUL_MAT_ID(ffn_moe_down-28): 225 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 439 us MUL_MAT_ID(ffn_moe_down-29): 191 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 386 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 391 us MUL_MAT_ID(ffn_moe_down-31): 185 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 436 us MUL_MAT_ID(ffn_moe_down-32): 190 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 429 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 373 us MUL_MAT_ID(ffn_moe_down-34): 205 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 450 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 540 us MUL_MAT_ID(ffn_moe_down-36): 175 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 403 us MUL_MAT_ID(ffn_moe_down-37): 680 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 369 us MUL_MAT_ID(ffn_moe_down-38): 179 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 420 us MUL_MAT_ID(ffn_moe_down-39): 264 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 374 us MUL_MAT_ID(ffn_moe_down-40): 239 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 452 us MUL_MAT_ID(ffn_moe_down-41): 187 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 359 us MUL_MAT_ID(ffn_moe_down-42): 433 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 418 us MUL_MAT_ID(ffn_moe_down-43): 218 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 416 us MUL_MAT_ID(ffn_moe_down-44): 217 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 472 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 464 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 498 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 355 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 426 us MUL_MAT_ID(ffn_moe_down-48): 542 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 411 us MUL_MAT_ID(ffn_moe_down-49): 197 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 399 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 429 us MUL_MAT_ID(ffn_moe_down-51): 355 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 417 us MUL_MAT_ID(ffn_moe_down-52): 181 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 411 us MUL_MAT_ID(ffn_moe_down-53): 378 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 409 us MUL_MAT_ID(ffn_moe_down-54): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 449 us MUL_MAT_ID(ffn_moe_down-55): 250 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 386 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 392 us MUL_MAT_ID(ffn_moe_down-57): 198 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 423 us MUL_MAT_ID(ffn_moe_down-58): 188 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 380 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 454 us MUL_MAT_ID(ffn_moe_down-60): 230 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 396 us MUL_MAT_ID(ffn_moe_down-61): 395 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 423 us MUL_MAT_ID(ffn_moe_down-62): 221 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 411 us MUL_MAT_ID(ffn_moe_down-63): 201 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 370 us MUL_MAT_ID(ffn_moe_down-64): 347 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 398 us MUL_MAT_ID(ffn_moe_down-65): 411 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 400 us MUL_MAT_ID(ffn_moe_down-66): 368 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 365 us MUL_MAT_ID(ffn_moe_down-67): 403 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 205 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 360 us MUL_MAT_ID(ffn_moe_down-69): 206 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 380 us MUL_MAT_ID(ffn_moe_down-70): 211 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 351 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 397 us MUL_MAT_ID(ffn_moe_down-72): 203 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 379 us MUL_MAT_ID(ffn_moe_down-74): 368 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 387 us MUL_MAT_ID(ffn_moe_down-75): 194 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 364 us MUL_MAT_ID(ffn_moe_down-76): 386 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 419 us MUL_MAT_ID(ffn_moe_down-77): 221 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 358 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 382 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 378 us MUL_MAT_ID(ffn_moe_down-80): 202 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 435 us MUL_MAT_ID(ffn_moe_down-81): 426 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 393 us MUL_MAT_ID(ffn_moe_down-82): 217 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 398 us MUL_MAT_ID(ffn_moe_down-83): 413 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 396 us MUL_MAT_ID(ffn_moe_down-84): 644 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 390 us MUL_MAT_ID(ffn_moe_down-85): 290 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 380 us MUL_MAT_ID(ffn_moe_down-86): 400 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 346 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 378 us MUL_MAT_ID(ffn_moe_down-88): 449 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 393 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 374 us MUL_MAT_ID(ffn_moe_down-90): 225 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 395 us MUL_MAT_ID(ffn_moe_down-91): 271 us ggml_barrier(...): 21 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 385 us MUL_MAT_ID(ffn_moe_down-25): 661 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 404 us MUL_MAT_ID(ffn_moe_down-26): 213 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 350 us MUL_MAT_ID(ffn_moe_down-27): 189 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 416 us MUL_MAT_ID(ffn_moe_down-28): 188 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 370 us MUL_MAT_ID(ffn_moe_down-29): 188 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 413 us MUL_MAT_ID(ffn_moe_down-30): 197 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 415 us MUL_MAT_ID(ffn_moe_down-31): 200 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 390 us MUL_MAT_ID(ffn_moe_down-32): 186 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 393 us MUL_MAT_ID(ffn_moe_down-33): 432 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 368 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 392 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 426 us MUL_MAT_ID(ffn_moe_down-36): 229 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 403 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 354 us MUL_MAT_ID(ffn_moe_down-38): 204 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 426 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 407 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 512 us MUL_MAT_ID(ffn_moe_down-41): 207 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 412 us MUL_MAT_ID(ffn_moe_down-42): 174 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 425 us MUL_MAT_ID(ffn_moe_down-43): 234 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 454 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 400 us MUL_MAT_ID(ffn_moe_down-45): 433 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 374 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 413 us MUL_MAT_ID(ffn_moe_down-47): 209 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 409 us MUL_MAT_ID(ffn_moe_down-48): 197 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 446 us MUL_MAT_ID(ffn_moe_down-49): 390 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 403 us MUL_MAT_ID(ffn_moe_down-50): 593 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 412 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 396 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 386 us MUL_MAT_ID(ffn_moe_down-53): 183 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 403 us MUL_MAT_ID(ffn_moe_down-54): 225 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 401 us MUL_MAT_ID(ffn_moe_down-55): 340 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 428 us MUL_MAT_ID(ffn_moe_down-56): 197 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 421 us MUL_MAT_ID(ffn_moe_down-57): 179 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 381 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 414 us MUL_MAT_ID(ffn_moe_down-59): 229 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 354 us MUL_MAT_ID(ffn_moe_down-60): 412 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 366 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 374 us MUL_MAT_ID(ffn_moe_down-62): 189 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 347 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 352 us MUL_MAT_ID(ffn_moe_down-64): 248 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 433 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 393 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 400 us MUL_MAT_ID(ffn_moe_down-67): 271 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 364 us MUL_MAT_ID(ffn_moe_down-68): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 412 us MUL_MAT_ID(ffn_moe_down-69): 188 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 384 us MUL_MAT_ID(ffn_moe_down-70): 543 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 472 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 395 us MUL_MAT_ID(ffn_moe_down-72): 201 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 405 us MUL_MAT_ID(ffn_moe_down-73): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 347 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 407 us MUL_MAT_ID(ffn_moe_down-75): 361 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 371 us MUL_MAT_ID(ffn_moe_down-76): 187 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 380 us MUL_MAT_ID(ffn_moe_down-77): 199 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 384 us MUL_MAT_ID(ffn_moe_down-78): 191 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 365 us MUL_MAT_ID(ffn_moe_down-79): 177 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 406 us MUL_MAT_ID(ffn_moe_down-80): 242 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 403 us MUL_MAT_ID(ffn_moe_down-81): 363 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 411 us MUL_MAT_ID(ffn_moe_down-82): 574 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 358 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 416 us MUL_MAT_ID(ffn_moe_down-84): 250 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 418 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 383 us MUL_MAT_ID(ffn_moe_down-86): 219 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 380 us MUL_MAT_ID(ffn_moe_down-87): 189 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 340 us MUL_MAT_ID(ffn_moe_down-88): 236 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 387 us MUL_MAT_ID(ffn_moe_down-89): 175 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 376 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 394 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 27 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 395 us MUL_MAT_ID(ffn_moe_down-25): 205 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 407 us MUL_MAT_ID(ffn_moe_down-26): 205 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 436 us MUL_MAT_ID(ffn_moe_down-27): 217 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 418 us MUL_MAT_ID(ffn_moe_down-28): 210 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 335 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 383 us MUL_MAT_ID(ffn_moe_down-30): 226 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 361 us MUL_MAT_ID(ffn_moe_down-31): 200 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 398 us MUL_MAT_ID(ffn_moe_down-32): 188 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 373 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 413 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 432 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 395 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 390 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 379 us MUL_MAT_ID(ffn_moe_down-38): 182 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 351 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 383 us MUL_MAT_ID(ffn_moe_down-40): 244 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 401 us MUL_MAT_ID(ffn_moe_down-41): 195 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 434 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 389 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 379 us MUL_MAT_ID(ffn_moe_down-44): 222 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 384 us MUL_MAT_ID(ffn_moe_down-45): 213 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 430 us MUL_MAT_ID(ffn_moe_down-46): 206 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 429 us MUL_MAT_ID(ffn_moe_down-47): 198 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 409 us MUL_MAT_ID(ffn_moe_down-48): 279 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 415 us MUL_MAT_ID(ffn_moe_down-49): 196 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 387 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 432 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 388 us MUL_MAT_ID(ffn_moe_down-52): 197 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 434 us MUL_MAT_ID(ffn_moe_down-53): 209 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 386 us MUL_MAT_ID(ffn_moe_down-54): 220 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 433 us MUL_MAT_ID(ffn_moe_down-55): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 395 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 378 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 389 us MUL_MAT_ID(ffn_moe_down-58): 210 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 430 us MUL_MAT_ID(ffn_moe_down-59): 438 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 384 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 389 us MUL_MAT_ID(ffn_moe_down-61): 202 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 373 us MUL_MAT_ID(ffn_moe_down-62): 203 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 409 us MUL_MAT_ID(ffn_moe_down-63): 205 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 407 us MUL_MAT_ID(ffn_moe_down-64): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 397 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 501 us MUL_MAT_ID(ffn_moe_down-66): 169 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 414 us MUL_MAT_ID(ffn_moe_down-67): 200 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 373 us MUL_MAT_ID(ffn_moe_down-68): 214 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 380 us MUL_MAT_ID(ffn_moe_down-69): 199 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 405 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 435 us MUL_MAT_ID(ffn_moe_down-71): 561 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 374 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 376 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 426 us MUL_MAT_ID(ffn_moe_down-74): 195 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 416 us MUL_MAT_ID(ffn_moe_down-75): 579 us ggml_barrier(...): 336 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 362 us MUL_MAT_ID(ffn_moe_down-76): 188 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 385 us MUL_MAT_ID(ffn_moe_down-77): 193 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 430 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 389 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 474 us MUL_MAT_ID(ffn_moe_down-80): 182 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 490 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 394 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 347 us MUL_MAT_ID(ffn_moe_down-83): 187 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 378 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 482 us MUL_MAT_ID(ffn_moe_down-85): 246 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 383 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 394 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 347 us MUL_MAT_ID(ffn_moe_down-88): 189 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 389 us MUL_MAT_ID(ffn_moe_down-89): 183 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 332 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 418 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 17 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 396 us MUL_MAT_ID(ffn_moe_down-25): 198 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 386 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 418 us MUL_MAT_ID(ffn_moe_down-27): 208 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 391 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 393 us MUL_MAT_ID(ffn_moe_down-29): 187 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 408 us MUL_MAT_ID(ffn_moe_down-30): 201 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 379 us MUL_MAT_ID(ffn_moe_down-31): 537 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 347 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 415 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 392 us MUL_MAT_ID(ffn_moe_down-34): 212 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 403 us MUL_MAT_ID(ffn_moe_down-35): 184 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 401 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 415 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 388 us MUL_MAT_ID(ffn_moe_down-38): 185 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 393 us MUL_MAT_ID(ffn_moe_down-39): 242 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 388 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 420 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 372 us MUL_MAT_ID(ffn_moe_down-42): 184 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 374 us MUL_MAT_ID(ffn_moe_down-43): 198 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 327 us MUL_MAT_ID(ffn_moe_down-44): 243 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 427 us MUL_MAT_ID(ffn_moe_down-45): 229 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 381 us MUL_MAT_ID(ffn_moe_down-46): 396 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 356 us MUL_MAT_ID(ffn_moe_down-47): 192 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 398 us MUL_MAT_ID(ffn_moe_down-48): 331 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 388 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 330 us MUL_MAT_ID(ffn_moe_down-50): 212 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 383 us MUL_MAT_ID(ffn_moe_down-51): 186 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 396 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 375 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 362 us MUL_MAT_ID(ffn_moe_down-54): 210 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 372 us MUL_MAT_ID(ffn_moe_down-55): 207 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 398 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 382 us MUL_MAT_ID(ffn_moe_down-57): 185 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 421 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 389 us MUL_MAT_ID(ffn_moe_down-59): 204 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 357 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 493 us MUL_MAT_ID(ffn_moe_down-61): 173 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 381 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 370 us MUL_MAT_ID(ffn_moe_down-63): 200 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 385 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 387 us MUL_MAT_ID(ffn_moe_down-65): 270 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 434 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 410 us MUL_MAT_ID(ffn_moe_down-67): 185 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 390 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 437 us MUL_MAT_ID(ffn_moe_down-69): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 395 us MUL_MAT_ID(ffn_moe_down-70): 207 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 381 us MUL_MAT_ID(ffn_moe_down-71): 201 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 440 us MUL_MAT_ID(ffn_moe_down-72): 330 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 385 us MUL_MAT_ID(ffn_moe_down-73): 235 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 495 us MUL_MAT_ID(ffn_moe_down-74): 261 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 388 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 360 us MUL_MAT_ID(ffn_moe_down-76): 185 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 432 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 414 us MUL_MAT_ID(ffn_moe_down-78): 198 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 401 us MUL_MAT_ID(ffn_moe_down-79): 190 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 411 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 396 us MUL_MAT_ID(ffn_moe_down-81): 185 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 410 us MUL_MAT_ID(ffn_moe_down-82): 536 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 389 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 434 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 402 us MUL_MAT_ID(ffn_moe_down-85): 228 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 397 us MUL_MAT_ID(ffn_moe_down-86): 421 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 403 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 395 us MUL_MAT_ID(ffn_moe_down-88): 211 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 456 us MUL_MAT_ID(ffn_moe_down-89): 226 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 416 us MUL_MAT_ID(ffn_moe_down-90): 273 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 617 us MUL_MAT_ID(ffn_moe_down-91): 280 us ggml_barrier(...): 72 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 400 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 364 us MUL_MAT_ID(ffn_moe_down-26): 207 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 387 us MUL_MAT_ID(ffn_moe_down-27): 556 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 431 us MUL_MAT_ID(ffn_moe_down-28): 193 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 403 us MUL_MAT_ID(ffn_moe_down-29): 202 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 419 us MUL_MAT_ID(ffn_moe_down-30): 186 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 422 us MUL_MAT_ID(ffn_moe_down-31): 214 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 411 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 449 us MUL_MAT_ID(ffn_moe_down-33): 309 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 404 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 356 us MUL_MAT_ID(ffn_moe_down-35): 214 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 387 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 404 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 385 us MUL_MAT_ID(ffn_moe_down-38): 219 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 428 us MUL_MAT_ID(ffn_moe_down-39): 380 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 363 us MUL_MAT_ID(ffn_moe_down-40): 449 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 391 us MUL_MAT_ID(ffn_moe_down-41): 216 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 398 us MUL_MAT_ID(ffn_moe_down-42): 280 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 425 us MUL_MAT_ID(ffn_moe_down-43): 257 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 415 us MUL_MAT_ID(ffn_moe_down-44): 187 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 413 us MUL_MAT_ID(ffn_moe_down-45): 215 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 360 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 388 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 322 us MUL_MAT_ID(ffn_moe_down-48): 377 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 386 us MUL_MAT_ID(ffn_moe_down-49): 206 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 401 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 384 us MUL_MAT_ID(ffn_moe_down-51): 209 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 422 us MUL_MAT_ID(ffn_moe_down-52): 205 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 403 us MUL_MAT_ID(ffn_moe_down-53): 201 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 369 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 387 us MUL_MAT_ID(ffn_moe_down-55): 214 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 345 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 386 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 411 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 396 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 388 us MUL_MAT_ID(ffn_moe_down-60): 189 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 386 us MUL_MAT_ID(ffn_moe_down-61): 182 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 394 us MUL_MAT_ID(ffn_moe_down-62): 332 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 350 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 393 us MUL_MAT_ID(ffn_moe_down-64): 185 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 424 us MUL_MAT_ID(ffn_moe_down-65): 215 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 410 us MUL_MAT_ID(ffn_moe_down-66): 207 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 414 us MUL_MAT_ID(ffn_moe_down-67): 231 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 408 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 394 us MUL_MAT_ID(ffn_moe_down-69): 242 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 411 us MUL_MAT_ID(ffn_moe_down-70): 209 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 347 us MUL_MAT_ID(ffn_moe_down-71): 226 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 391 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 382 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 366 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 373 us MUL_MAT_ID(ffn_moe_down-75): 184 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 401 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 376 us MUL_MAT_ID(ffn_moe_down-77): 215 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 353 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 422 us MUL_MAT_ID(ffn_moe_down-79): 215 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 393 us MUL_MAT_ID(ffn_moe_down-80): 563 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 346 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 346 us MUL_MAT_ID(ffn_moe_down-82): 376 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 414 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 393 us MUL_MAT_ID(ffn_moe_down-84): 226 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 410 us MUL_MAT_ID(ffn_moe_down-85): 222 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 372 us MUL_MAT_ID(ffn_moe_down-86): 206 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 376 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 416 us MUL_MAT_ID(ffn_moe_down-88): 188 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 374 us MUL_MAT_ID(ffn_moe_down-89): 187 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 419 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 415 us MUL_MAT_ID(ffn_moe_down-91): 412 us ggml_barrier(...): 17 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 410 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 438 us MUL_MAT_ID(ffn_moe_down-26): 200 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 359 us MUL_MAT_ID(ffn_moe_down-27): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 444 us MUL_MAT_ID(ffn_moe_down-28): 230 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 478 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 413 us MUL_MAT_ID(ffn_moe_down-30): 219 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 405 us MUL_MAT_ID(ffn_moe_down-31): 212 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 376 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 470 us MUL_MAT_ID(ffn_moe_down-33): 315 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 455 us MUL_MAT_ID(ffn_moe_down-34): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 413 us MUL_MAT_ID(ffn_moe_down-35): 197 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 442 us MUL_MAT_ID(ffn_moe_down-36): 247 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 363 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 406 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 384 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 399 us MUL_MAT_ID(ffn_moe_down-40): 227 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 390 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 437 us MUL_MAT_ID(ffn_moe_down-42): 183 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 417 us MUL_MAT_ID(ffn_moe_down-43): 187 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 406 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 367 us MUL_MAT_ID(ffn_moe_down-45): 277 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 409 us MUL_MAT_ID(ffn_moe_down-46): 307 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 426 us MUL_MAT_ID(ffn_moe_down-47): 200 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 373 us MUL_MAT_ID(ffn_moe_down-48): 297 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 376 us MUL_MAT_ID(ffn_moe_down-49): 396 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 375 us MUL_MAT_ID(ffn_moe_down-50): 186 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 344 us MUL_MAT_ID(ffn_moe_down-51): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 401 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 380 us MUL_MAT_ID(ffn_moe_down-53): 556 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 360 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 362 us MUL_MAT_ID(ffn_moe_down-55): 401 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 403 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 401 us MUL_MAT_ID(ffn_moe_down-57): 181 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 376 us MUL_MAT_ID(ffn_moe_down-58): 214 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 415 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 344 us MUL_MAT_ID(ffn_moe_down-60): 412 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 428 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 431 us MUL_MAT_ID(ffn_moe_down-62): 182 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 352 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 342 us MUL_MAT_ID(ffn_moe_down-64): 185 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 412 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 423 us MUL_MAT_ID(ffn_moe_down-66): 220 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 401 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 394 us MUL_MAT_ID(ffn_moe_down-68): 187 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 369 us MUL_MAT_ID(ffn_moe_down-69): 212 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 361 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 384 us MUL_MAT_ID(ffn_moe_down-71): 205 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 362 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 366 us MUL_MAT_ID(ffn_moe_down-73): 218 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 404 us MUL_MAT_ID(ffn_moe_down-74): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 380 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 400 us MUL_MAT_ID(ffn_moe_down-76): 243 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 197 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 372 us MUL_MAT_ID(ffn_moe_down-78): 240 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 385 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 408 us MUL_MAT_ID(ffn_moe_down-80): 229 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 417 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 408 us MUL_MAT_ID(ffn_moe_down-82): 213 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 360 us MUL_MAT_ID(ffn_moe_down-83): 438 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 395 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 374 us MUL_MAT_ID(ffn_moe_down-85): 260 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 379 us MUL_MAT_ID(ffn_moe_down-86): 206 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 408 us MUL_MAT_ID(ffn_moe_down-87): 201 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 414 us MUL_MAT_ID(ffn_moe_down-88): 198 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 401 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 373 us MUL_MAT_ID(ffn_moe_down-90): 378 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 418 us MUL_MAT_ID(ffn_moe_down-91): 250 us ggml_barrier(...): 92 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 420 us MUL_MAT_ID(ffn_moe_down-26): 198 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 414 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 400 us MUL_MAT_ID(ffn_moe_down-28): 195 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 387 us MUL_MAT_ID(ffn_moe_down-29): 215 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 430 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 379 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 368 us MUL_MAT_ID(ffn_moe_down-32): 190 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 453 us MUL_MAT_ID(ffn_moe_down-33): 232 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 460 us MUL_MAT_ID(ffn_moe_down-34): 465 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 537 us MUL_MAT_ID(ffn_moe_down-35): 217 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 426 us MUL_MAT_ID(ffn_moe_down-36): 343 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 472 us MUL_MAT_ID(ffn_moe_down-37): 368 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 350 us MUL_MAT_ID(ffn_moe_down-38): 189 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 401 us MUL_MAT_ID(ffn_moe_down-39): 227 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 418 us MUL_MAT_ID(ffn_moe_down-40): 249 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 410 us MUL_MAT_ID(ffn_moe_down-41): 187 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 464 us MUL_MAT_ID(ffn_moe_down-42): 194 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 449 us MUL_MAT_ID(ffn_moe_down-43): 203 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 397 us MUL_MAT_ID(ffn_moe_down-44): 219 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 444 us MUL_MAT_ID(ffn_moe_down-45): 279 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 490 us MUL_MAT_ID(ffn_moe_down-46): 202 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 401 us MUL_MAT_ID(ffn_moe_down-47): 288 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 402 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 382 us MUL_MAT_ID(ffn_moe_down-49): 206 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 350 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 377 us MUL_MAT_ID(ffn_moe_down-51): 383 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 401 us MUL_MAT_ID(ffn_moe_down-52): 188 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 363 us MUL_MAT_ID(ffn_moe_down-53): 195 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 372 us MUL_MAT_ID(ffn_moe_down-54): 209 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 382 us MUL_MAT_ID(ffn_moe_down-55): 221 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 386 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 399 us MUL_MAT_ID(ffn_moe_down-57): 357 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 390 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 449 us MUL_MAT_ID(ffn_moe_down-59): 188 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 388 us MUL_MAT_ID(ffn_moe_down-60): 203 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 386 us MUL_MAT_ID(ffn_moe_down-61): 278 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 387 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 407 us MUL_MAT_ID(ffn_moe_down-63): 185 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 398 us MUL_MAT_ID(ffn_moe_down-64): 183 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 415 us MUL_MAT_ID(ffn_moe_down-65): 227 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 361 us MUL_MAT_ID(ffn_moe_down-66): 209 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 341 us MUL_MAT_ID(ffn_moe_down-67): 223 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 397 us MUL_MAT_ID(ffn_moe_down-68): 201 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 435 us MUL_MAT_ID(ffn_moe_down-69): 204 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 400 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 352 us MUL_MAT_ID(ffn_moe_down-71): 215 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 364 us MUL_MAT_ID(ffn_moe_down-72): 179 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 436 us MUL_MAT_ID(ffn_moe_down-73): 270 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 405 us MUL_MAT_ID(ffn_moe_down-74): 194 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 395 us MUL_MAT_ID(ffn_moe_down-75): 317 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 408 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 382 us MUL_MAT_ID(ffn_moe_down-77): 186 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 362 us MUL_MAT_ID(ffn_moe_down-78): 207 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 401 us MUL_MAT_ID(ffn_moe_down-79): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 363 us MUL_MAT_ID(ffn_moe_down-80): 177 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 413 us MUL_MAT_ID(ffn_moe_down-81): 224 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 453 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 374 us MUL_MAT_ID(ffn_moe_down-83): 189 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 375 us MUL_MAT_ID(ffn_moe_down-84): 246 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 436 us MUL_MAT_ID(ffn_moe_down-85): 193 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 402 us MUL_MAT_ID(ffn_moe_down-86): 322 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 403 us MUL_MAT_ID(ffn_moe_down-87): 433 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 412 us MUL_MAT_ID(ffn_moe_down-88): 200 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 370 us MUL_MAT_ID(ffn_moe_down-89): 256 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 414 us MUL_MAT_ID(ffn_moe_down-90): 238 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 430 us MUL_MAT_ID(ffn_moe_down-91): 226 us ggml_barrier(...): 15 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 367 us MUL_MAT_ID(ffn_moe_down-25): 372 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 339 us MUL_MAT_ID(ffn_moe_down-26): 186 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 414 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 420 us MUL_MAT_ID(ffn_moe_down-28): 191 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 421 us MUL_MAT_ID(ffn_moe_down-29): 343 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 419 us MUL_MAT_ID(ffn_moe_down-30): 182 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 399 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 387 us MUL_MAT_ID(ffn_moe_down-32): 420 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 405 us MUL_MAT_ID(ffn_moe_down-33): 225 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 435 us MUL_MAT_ID(ffn_moe_down-34): 215 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 440 us MUL_MAT_ID(ffn_moe_down-35): 197 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 403 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 436 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 503 us MUL_MAT_ID(ffn_moe_down-38): 188 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 404 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 394 us MUL_MAT_ID(ffn_moe_down-40): 244 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 380 us MUL_MAT_ID(ffn_moe_down-41): 402 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 362 us MUL_MAT_ID(ffn_moe_down-42): 187 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 383 us MUL_MAT_ID(ffn_moe_down-43): 188 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 430 us MUL_MAT_ID(ffn_moe_down-44): 205 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 435 us MUL_MAT_ID(ffn_moe_down-45): 223 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 489 us MUL_MAT_ID(ffn_moe_down-46): 261 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 764 us MUL_MAT_ID(ffn_moe_down-47): 334 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 337 us MUL_MAT_ID(ffn_moe_down-48): 205 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 401 us MUL_MAT_ID(ffn_moe_down-49): 254 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 459 us MUL_MAT_ID(ffn_moe_down-50): 215 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 421 us MUL_MAT_ID(ffn_moe_down-51): 228 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 357 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 375 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 392 us MUL_MAT_ID(ffn_moe_down-54): 234 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 405 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 405 us MUL_MAT_ID(ffn_moe_down-56): 210 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 421 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 402 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 692 us MUL_MAT_ID(ffn_moe_down-59): 262 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 406 us MUL_MAT_ID(ffn_moe_down-60): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 413 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 378 us MUL_MAT_ID(ffn_moe_down-62): 203 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 411 us MUL_MAT_ID(ffn_moe_down-63): 414 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 515 us MUL_MAT_ID(ffn_moe_down-64): 185 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 398 us MUL_MAT_ID(ffn_moe_down-65): 209 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 416 us MUL_MAT_ID(ffn_moe_down-66): 401 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 357 us MUL_MAT_ID(ffn_moe_down-67): 219 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 454 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 410 us MUL_MAT_ID(ffn_moe_down-69): 485 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 389 us MUL_MAT_ID(ffn_moe_down-70): 194 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 477 us MUL_MAT_ID(ffn_moe_down-71): 173 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 425 us MUL_MAT_ID(ffn_moe_down-72): 295 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 393 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 343 us MUL_MAT_ID(ffn_moe_down-74): 190 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 395 us MUL_MAT_ID(ffn_moe_down-75): 189 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 336 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 466 us MUL_MAT_ID(ffn_moe_down-77): 213 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 416 us MUL_MAT_ID(ffn_moe_down-78): 212 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 439 us MUL_MAT_ID(ffn_moe_down-79): 184 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 389 us MUL_MAT_ID(ffn_moe_down-80): 185 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 389 us MUL_MAT_ID(ffn_moe_down-81): 184 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 328 us MUL_MAT_ID(ffn_moe_down-82): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 438 us MUL_MAT_ID(ffn_moe_down-83): 382 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 370 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 384 us MUL_MAT_ID(ffn_moe_down-85): 245 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 437 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 442 us MUL_MAT_ID(ffn_moe_down-87): 183 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 428 us MUL_MAT_ID(ffn_moe_down-88): 267 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 406 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 448 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 408 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 12 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 357 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 493 us MUL_MAT_ID(ffn_moe_down-26): 314 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 386 us MUL_MAT_ID(ffn_moe_down-27): 209 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 388 us MUL_MAT_ID(ffn_moe_down-28): 370 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 384 us MUL_MAT_ID(ffn_moe_down-29): 204 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 458 us MUL_MAT_ID(ffn_moe_down-30): 205 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 593 us MUL_MAT_ID(ffn_moe_down-31): 328 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 377 us MUL_MAT_ID(ffn_moe_down-32): 291 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 365 us MUL_MAT_ID(ffn_moe_down-33): 182 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 403 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 380 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 406 us MUL_MAT_ID(ffn_moe_down-36): 258 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 385 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 481 us MUL_MAT_ID(ffn_moe_down-38): 637 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 341 us MUL_MAT_ID(ffn_moe_down-39): 228 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 426 us MUL_MAT_ID(ffn_moe_down-40): 235 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 396 us MUL_MAT_ID(ffn_moe_down-41): 197 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 367 us MUL_MAT_ID(ffn_moe_down-42): 192 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 375 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 357 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 411 us MUL_MAT_ID(ffn_moe_down-45): 240 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 439 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 426 us MUL_MAT_ID(ffn_moe_down-47): 352 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 424 us MUL_MAT_ID(ffn_moe_down-48): 243 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 455 us MUL_MAT_ID(ffn_moe_down-49): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 383 us MUL_MAT_ID(ffn_moe_down-50): 354 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 388 us MUL_MAT_ID(ffn_moe_down-51): 210 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 462 us MUL_MAT_ID(ffn_moe_down-52): 235 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 482 us MUL_MAT_ID(ffn_moe_down-53): 252 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 450 us MUL_MAT_ID(ffn_moe_down-54): 286 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 425 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 452 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 460 us MUL_MAT_ID(ffn_moe_down-57): 236 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 597 us MUL_MAT_ID(ffn_moe_down-58): 311 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 482 us MUL_MAT_ID(ffn_moe_down-59): 205 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 436 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 387 us MUL_MAT_ID(ffn_moe_down-61): 184 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 405 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 397 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 343 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 397 us MUL_MAT_ID(ffn_moe_down-65): 200 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 551 us MUL_MAT_ID(ffn_moe_down-66): 226 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 420 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 401 us MUL_MAT_ID(ffn_moe_down-68): 589 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 602 us MUL_MAT_ID(ffn_moe_down-69): 332 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 352 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 508 us MUL_MAT_ID(ffn_moe_down-71): 411 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 365 us MUL_MAT_ID(ffn_moe_down-72): 206 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 368 us MUL_MAT_ID(ffn_moe_down-73): 424 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 375 us MUL_MAT_ID(ffn_moe_down-74): 216 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 412 us MUL_MAT_ID(ffn_moe_down-75): 525 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 398 us MUL_MAT_ID(ffn_moe_down-76): 201 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 392 us MUL_MAT_ID(ffn_moe_down-77): 216 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 384 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 398 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 427 us MUL_MAT_ID(ffn_moe_down-80): 384 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 414 us MUL_MAT_ID(ffn_moe_down-81): 234 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 416 us MUL_MAT_ID(ffn_moe_down-82): 580 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 410 us MUL_MAT_ID(ffn_moe_down-83): 399 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 385 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 366 us MUL_MAT_ID(ffn_moe_down-85): 484 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 364 us MUL_MAT_ID(ffn_moe_down-86): 276 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 421 us MUL_MAT_ID(ffn_moe_down-87): 218 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 368 us MUL_MAT_ID(ffn_moe_down-88): 162 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 430 us MUL_MAT_ID(ffn_moe_down-89): 365 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 398 us MUL_MAT_ID(ffn_moe_down-90): 226 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 479 us MUL_MAT_ID(ffn_moe_down-91): 227 us ggml_barrier(...): 139 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 422 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 431 us MUL_MAT_ID(ffn_moe_down-26): 185 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 451 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 385 us MUL_MAT_ID(ffn_moe_down-28): 209 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 413 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 397 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 382 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 417 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 447 us MUL_MAT_ID(ffn_moe_down-33): 236 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 381 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 467 us MUL_MAT_ID(ffn_moe_down-35): 185 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 435 us MUL_MAT_ID(ffn_moe_down-36): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 359 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 374 us MUL_MAT_ID(ffn_moe_down-38): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 396 us MUL_MAT_ID(ffn_moe_down-39): 239 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 512 us MUL_MAT_ID(ffn_moe_down-40): 307 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 403 us MUL_MAT_ID(ffn_moe_down-41): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 427 us MUL_MAT_ID(ffn_moe_down-42): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 370 us MUL_MAT_ID(ffn_moe_down-43): 589 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 395 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 349 us MUL_MAT_ID(ffn_moe_down-45): 220 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 405 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 438 us MUL_MAT_ID(ffn_moe_down-47): 184 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 392 us MUL_MAT_ID(ffn_moe_down-48): 517 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 382 us MUL_MAT_ID(ffn_moe_down-49): 290 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 392 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 413 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 390 us MUL_MAT_ID(ffn_moe_down-52): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 384 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 430 us MUL_MAT_ID(ffn_moe_down-54): 201 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 413 us MUL_MAT_ID(ffn_moe_down-55): 217 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 414 us MUL_MAT_ID(ffn_moe_down-56): 527 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 410 us MUL_MAT_ID(ffn_moe_down-57): 210 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 412 us MUL_MAT_ID(ffn_moe_down-58): 286 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 398 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 440 us MUL_MAT_ID(ffn_moe_down-60): 245 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 445 us MUL_MAT_ID(ffn_moe_down-61): 184 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 386 us MUL_MAT_ID(ffn_moe_down-62): 221 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 437 us MUL_MAT_ID(ffn_moe_down-63): 187 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 426 us MUL_MAT_ID(ffn_moe_down-64): 209 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 370 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 430 us MUL_MAT_ID(ffn_moe_down-66): 254 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 434 us MUL_MAT_ID(ffn_moe_down-67): 187 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 427 us MUL_MAT_ID(ffn_moe_down-68): 189 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 476 us MUL_MAT_ID(ffn_moe_down-69): 252 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 453 us MUL_MAT_ID(ffn_moe_down-70): 495 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 506 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 382 us MUL_MAT_ID(ffn_moe_down-72): 186 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 353 us MUL_MAT_ID(ffn_moe_down-73): 210 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 370 us MUL_MAT_ID(ffn_moe_down-74): 228 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 388 us MUL_MAT_ID(ffn_moe_down-75): 194 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 408 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 402 us MUL_MAT_ID(ffn_moe_down-77): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 389 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 358 us MUL_MAT_ID(ffn_moe_down-79): 189 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 344 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 397 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 389 us MUL_MAT_ID(ffn_moe_down-82): 464 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 507 us MUL_MAT_ID(ffn_moe_down-83): 262 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 418 us MUL_MAT_ID(ffn_moe_down-84): 250 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 417 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 393 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 386 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 410 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 379 us MUL_MAT_ID(ffn_moe_down-89): 186 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 410 us MUL_MAT_ID(ffn_moe_down-90): 390 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 384 us MUL_MAT_ID(ffn_moe_down-91): 233 us ggml_barrier(...): 37 us GET_ROWS(inp_embd): 5 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 545 us MUL_MAT_ID(ffn_moe_down-25): 327 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 384 us MUL_MAT_ID(ffn_moe_down-26): 212 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 392 us MUL_MAT_ID(ffn_moe_down-27): 211 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 420 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 423 us MUL_MAT_ID(ffn_moe_down-29): 203 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 350 us MUL_MAT_ID(ffn_moe_down-30): 184 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 462 us MUL_MAT_ID(ffn_moe_down-31): 498 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 399 us MUL_MAT_ID(ffn_moe_down-32): 228 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 374 us MUL_MAT_ID(ffn_moe_down-33): 180 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 394 us MUL_MAT_ID(ffn_moe_down-34): 293 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 418 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 368 us MUL_MAT_ID(ffn_moe_down-36): 214 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 383 us MUL_MAT_ID(ffn_moe_down-37): 360 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 379 us MUL_MAT_ID(ffn_moe_down-38): 384 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 394 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 346 us MUL_MAT_ID(ffn_moe_down-40): 344 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 408 us MUL_MAT_ID(ffn_moe_down-41): 249 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 441 us MUL_MAT_ID(ffn_moe_down-42): 185 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 377 us MUL_MAT_ID(ffn_moe_down-43): 203 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 355 us MUL_MAT_ID(ffn_moe_down-44): 186 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 404 us MUL_MAT_ID(ffn_moe_down-45): 320 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 384 us MUL_MAT_ID(ffn_moe_down-46): 203 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 446 us MUL_MAT_ID(ffn_moe_down-47): 211 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 396 us MUL_MAT_ID(ffn_moe_down-48): 205 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 367 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 360 us MUL_MAT_ID(ffn_moe_down-50): 206 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 390 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 387 us MUL_MAT_ID(ffn_moe_down-52): 214 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 395 us MUL_MAT_ID(ffn_moe_down-53): 219 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 424 us MUL_MAT_ID(ffn_moe_down-54): 295 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 385 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 396 us MUL_MAT_ID(ffn_moe_down-56): 265 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 448 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 411 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 453 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 369 us MUL_MAT_ID(ffn_moe_down-60): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 396 us MUL_MAT_ID(ffn_moe_down-61): 209 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 389 us MUL_MAT_ID(ffn_moe_down-62): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 351 us MUL_MAT_ID(ffn_moe_down-63): 205 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 396 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 362 us MUL_MAT_ID(ffn_moe_down-65): 215 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 426 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 376 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 403 us MUL_MAT_ID(ffn_moe_down-68): 361 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 419 us MUL_MAT_ID(ffn_moe_down-69): 265 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 412 us MUL_MAT_ID(ffn_moe_down-70): 548 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 415 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 377 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 205 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 443 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 420 us MUL_MAT_ID(ffn_moe_down-75): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 380 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 420 us MUL_MAT_ID(ffn_moe_down-77): 206 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 439 us MUL_MAT_ID(ffn_moe_down-78): 182 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 408 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 402 us MUL_MAT_ID(ffn_moe_down-80): 218 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 441 us MUL_MAT_ID(ffn_moe_down-82): 215 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 355 us MUL_MAT_ID(ffn_moe_down-83): 184 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 427 us MUL_MAT_ID(ffn_moe_down-84): 247 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 417 us MUL_MAT_ID(ffn_moe_down-85): 280 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 410 us MUL_MAT_ID(ffn_moe_down-86): 207 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 624 us MUL_MAT_ID(ffn_moe_down-87): 309 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 383 us MUL_MAT_ID(ffn_moe_down-88): 230 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 443 us MUL_MAT_ID(ffn_moe_down-89): 205 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 422 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 447 us MUL_MAT_ID(ffn_moe_down-91): 250 us ggml_barrier(...): 10 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 373 us MUL_MAT_ID(ffn_moe_down-25): 211 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 421 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 356 us MUL_MAT_ID(ffn_moe_down-27): 213 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 412 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 400 us MUL_MAT_ID(ffn_moe_down-29): 182 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 265 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 452 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 440 us MUL_MAT_ID(ffn_moe_down-32): 213 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 418 us MUL_MAT_ID(ffn_moe_down-33): 196 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 445 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 409 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 417 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 368 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 425 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 387 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 401 us MUL_MAT_ID(ffn_moe_down-40): 238 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 457 us MUL_MAT_ID(ffn_moe_down-41): 251 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 416 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 397 us MUL_MAT_ID(ffn_moe_down-43): 208 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 408 us MUL_MAT_ID(ffn_moe_down-44): 185 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 394 us MUL_MAT_ID(ffn_moe_down-45): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 392 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 474 us MUL_MAT_ID(ffn_moe_down-47): 300 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 445 us MUL_MAT_ID(ffn_moe_down-48): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 373 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 403 us MUL_MAT_ID(ffn_moe_down-50): 216 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 430 us MUL_MAT_ID(ffn_moe_down-51): 566 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 403 us MUL_MAT_ID(ffn_moe_down-52): 202 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 449 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 438 us MUL_MAT_ID(ffn_moe_down-54): 224 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 501 us MUL_MAT_ID(ffn_moe_down-55): 309 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 420 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 422 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 355 us MUL_MAT_ID(ffn_moe_down-58): 207 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 422 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 394 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 401 us MUL_MAT_ID(ffn_moe_down-61): 180 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 389 us MUL_MAT_ID(ffn_moe_down-62): 192 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 510 us MUL_MAT_ID(ffn_moe_down-63): 501 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 400 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 378 us MUL_MAT_ID(ffn_moe_down-65): 323 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 425 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 331 us MUL_MAT_ID(ffn_moe_down-67): 335 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 396 us MUL_MAT_ID(ffn_moe_down-68): 209 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 453 us MUL_MAT_ID(ffn_moe_down-69): 185 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 373 us MUL_MAT_ID(ffn_moe_down-70): 497 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 398 us MUL_MAT_ID(ffn_moe_down-71): 242 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 440 us MUL_MAT_ID(ffn_moe_down-72): 214 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 409 us MUL_MAT_ID(ffn_moe_down-73): 222 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 408 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 365 us MUL_MAT_ID(ffn_moe_down-75): 181 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 526 us MUL_MAT_ID(ffn_moe_down-76): 184 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 415 us MUL_MAT_ID(ffn_moe_down-77): 209 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 404 us MUL_MAT_ID(ffn_moe_down-78): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 397 us MUL_MAT_ID(ffn_moe_down-79): 186 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 352 us MUL_MAT_ID(ffn_moe_down-80): 252 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 459 us MUL_MAT_ID(ffn_moe_down-81): 173 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 415 us MUL_MAT_ID(ffn_moe_down-82): 579 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 417 us MUL_MAT_ID(ffn_moe_down-83): 187 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 361 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 412 us MUL_MAT_ID(ffn_moe_down-85): 226 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 391 us MUL_MAT_ID(ffn_moe_down-86): 184 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 421 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 381 us MUL_MAT_ID(ffn_moe_down-88): 191 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 522 us MUL_MAT_ID(ffn_moe_down-89): 244 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 635 us MUL_MAT_ID(ffn_moe_down-90): 300 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 649 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 35 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 363 us MUL_MAT_ID(ffn_moe_down-25): 572 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 417 us MUL_MAT_ID(ffn_moe_down-26): 224 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 584 us MUL_MAT_ID(ffn_moe_down-27): 287 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 355 us MUL_MAT_ID(ffn_moe_down-28): 183 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 378 us MUL_MAT_ID(ffn_moe_down-29): 187 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 388 us MUL_MAT_ID(ffn_moe_down-30): 234 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 384 us MUL_MAT_ID(ffn_moe_down-31): 217 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 454 us MUL_MAT_ID(ffn_moe_down-32): 252 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 335 us MUL_MAT_ID(ffn_moe_down-33): 183 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 391 us MUL_MAT_ID(ffn_moe_down-34): 210 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 440 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 414 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 362 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 430 us MUL_MAT_ID(ffn_moe_down-38): 209 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 462 us MUL_MAT_ID(ffn_moe_down-39): 223 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 412 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 456 us MUL_MAT_ID(ffn_moe_down-41): 183 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 422 us MUL_MAT_ID(ffn_moe_down-42): 204 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 437 us MUL_MAT_ID(ffn_moe_down-43): 201 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 454 us MUL_MAT_ID(ffn_moe_down-44): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 335 us MUL_MAT_ID(ffn_moe_down-45): 394 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 418 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 390 us MUL_MAT_ID(ffn_moe_down-47): 180 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 431 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 351 us MUL_MAT_ID(ffn_moe_down-49): 183 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 328 us MUL_MAT_ID(ffn_moe_down-50): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 431 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 418 us MUL_MAT_ID(ffn_moe_down-52): 439 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 396 us MUL_MAT_ID(ffn_moe_down-53): 200 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 441 us MUL_MAT_ID(ffn_moe_down-54): 189 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 379 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 444 us MUL_MAT_ID(ffn_moe_down-56): 175 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 472 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 456 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 452 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 383 us MUL_MAT_ID(ffn_moe_down-60): 228 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 388 us MUL_MAT_ID(ffn_moe_down-61): 180 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 408 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 397 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 431 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 437 us MUL_MAT_ID(ffn_moe_down-65): 202 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 387 us MUL_MAT_ID(ffn_moe_down-66): 187 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 405 us MUL_MAT_ID(ffn_moe_down-67): 208 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 420 us MUL_MAT_ID(ffn_moe_down-68): 188 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 440 us MUL_MAT_ID(ffn_moe_down-69): 181 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 395 us MUL_MAT_ID(ffn_moe_down-70): 188 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 381 us MUL_MAT_ID(ffn_moe_down-71): 628 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 412 us MUL_MAT_ID(ffn_moe_down-72): 246 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 451 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 408 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 412 us MUL_MAT_ID(ffn_moe_down-75): 363 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 432 us MUL_MAT_ID(ffn_moe_down-76): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 385 us MUL_MAT_ID(ffn_moe_down-77): 279 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 442 us MUL_MAT_ID(ffn_moe_down-78): 606 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 399 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 429 us MUL_MAT_ID(ffn_moe_down-80): 183 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 415 us MUL_MAT_ID(ffn_moe_down-81): 200 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 442 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 412 us MUL_MAT_ID(ffn_moe_down-83): 219 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 417 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 413 us MUL_MAT_ID(ffn_moe_down-85): 270 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 427 us MUL_MAT_ID(ffn_moe_down-86): 225 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 479 us MUL_MAT_ID(ffn_moe_down-87): 467 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 382 us MUL_MAT_ID(ffn_moe_down-88): 202 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 404 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 389 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 416 us MUL_MAT_ID(ffn_moe_down-91): 247 us ggml_barrier(...): 20 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 385 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 426 us MUL_MAT_ID(ffn_moe_down-26): 216 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 412 us MUL_MAT_ID(ffn_moe_down-27): 209 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 643 us MUL_MAT_ID(ffn_moe_down-28): 561 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 362 us MUL_MAT_ID(ffn_moe_down-29): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 373 us MUL_MAT_ID(ffn_moe_down-30): 230 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 400 us MUL_MAT_ID(ffn_moe_down-31): 280 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 420 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 368 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 403 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 394 us MUL_MAT_ID(ffn_moe_down-35): 215 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 401 us MUL_MAT_ID(ffn_moe_down-36): 201 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 458 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 411 us MUL_MAT_ID(ffn_moe_down-38): 191 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 403 us MUL_MAT_ID(ffn_moe_down-39): 298 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 444 us MUL_MAT_ID(ffn_moe_down-40): 855 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 396 us MUL_MAT_ID(ffn_moe_down-41): 191 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 452 us MUL_MAT_ID(ffn_moe_down-42): 185 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 429 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 384 us MUL_MAT_ID(ffn_moe_down-44): 203 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 412 us MUL_MAT_ID(ffn_moe_down-45): 262 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 370 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 386 us MUL_MAT_ID(ffn_moe_down-47): 279 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 424 us MUL_MAT_ID(ffn_moe_down-48): 391 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 386 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 375 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 395 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 400 us MUL_MAT_ID(ffn_moe_down-52): 210 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 417 us MUL_MAT_ID(ffn_moe_down-53): 198 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 385 us MUL_MAT_ID(ffn_moe_down-54): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 333 us MUL_MAT_ID(ffn_moe_down-55): 184 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 414 us MUL_MAT_ID(ffn_moe_down-56): 438 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 409 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 408 us MUL_MAT_ID(ffn_moe_down-58): 187 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 364 us MUL_MAT_ID(ffn_moe_down-59): 227 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 413 us MUL_MAT_ID(ffn_moe_down-60): 195 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 447 us MUL_MAT_ID(ffn_moe_down-61): 310 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 400 us MUL_MAT_ID(ffn_moe_down-62): 188 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 337 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 343 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 390 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 365 us MUL_MAT_ID(ffn_moe_down-66): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 444 us MUL_MAT_ID(ffn_moe_down-67): 211 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 452 us MUL_MAT_ID(ffn_moe_down-68): 189 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 445 us MUL_MAT_ID(ffn_moe_down-69): 418 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 355 us MUL_MAT_ID(ffn_moe_down-70): 213 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 388 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 354 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 381 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 398 us MUL_MAT_ID(ffn_moe_down-75): 216 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 389 us MUL_MAT_ID(ffn_moe_down-76): 185 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 424 us MUL_MAT_ID(ffn_moe_down-77): 231 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 402 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 376 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 373 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 389 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 367 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 378 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 394 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 430 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 397 us MUL_MAT_ID(ffn_moe_down-86): 321 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 414 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 343 us MUL_MAT_ID(ffn_moe_down-88): 243 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 379 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 397 us MUL_MAT_ID(ffn_moe_down-90): 241 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 399 us MUL_MAT_ID(ffn_moe_down-91): 287 us ggml_barrier(...): 40 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 394 us MUL_MAT_ID(ffn_moe_down-25): 210 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 403 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 447 us MUL_MAT_ID(ffn_moe_down-27): 190 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 361 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 444 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 436 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 453 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 413 us MUL_MAT_ID(ffn_moe_down-32): 279 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 429 us MUL_MAT_ID(ffn_moe_down-33): 407 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 422 us MUL_MAT_ID(ffn_moe_down-34): 264 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 412 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 624 us MUL_MAT_ID(ffn_moe_down-36): 548 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 348 us MUL_MAT_ID(ffn_moe_down-37): 305 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 386 us MUL_MAT_ID(ffn_moe_down-38): 209 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 414 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 358 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 402 us MUL_MAT_ID(ffn_moe_down-41): 208 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 449 us MUL_MAT_ID(ffn_moe_down-42): 274 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 412 us MUL_MAT_ID(ffn_moe_down-43): 469 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 413 us MUL_MAT_ID(ffn_moe_down-44): 209 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 629 us MUL_MAT_ID(ffn_moe_down-45): 281 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 350 us MUL_MAT_ID(ffn_moe_down-46): 220 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 400 us MUL_MAT_ID(ffn_moe_down-47): 225 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 393 us MUL_MAT_ID(ffn_moe_down-48): 185 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 388 us MUL_MAT_ID(ffn_moe_down-49): 213 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 391 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 407 us MUL_MAT_ID(ffn_moe_down-51): 262 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 397 us MUL_MAT_ID(ffn_moe_down-52): 187 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 383 us MUL_MAT_ID(ffn_moe_down-53): 181 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 454 us MUL_MAT_ID(ffn_moe_down-54): 182 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 441 us MUL_MAT_ID(ffn_moe_down-55): 233 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 441 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 401 us MUL_MAT_ID(ffn_moe_down-57): 213 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 453 us MUL_MAT_ID(ffn_moe_down-58): 363 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 377 us MUL_MAT_ID(ffn_moe_down-59): 419 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 444 us MUL_MAT_ID(ffn_moe_down-60): 182 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 384 us MUL_MAT_ID(ffn_moe_down-61): 263 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 469 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 446 us MUL_MAT_ID(ffn_moe_down-63): 375 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 362 us MUL_MAT_ID(ffn_moe_down-64): 430 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 377 us MUL_MAT_ID(ffn_moe_down-65): 448 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 422 us MUL_MAT_ID(ffn_moe_down-66): 304 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 457 us MUL_MAT_ID(ffn_moe_down-67): 208 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 429 us MUL_MAT_ID(ffn_moe_down-68): 218 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 477 us MUL_MAT_ID(ffn_moe_down-69): 269 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 418 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 427 us MUL_MAT_ID(ffn_moe_down-71): 330 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 504 us MUL_MAT_ID(ffn_moe_down-72): 220 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 395 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 412 us MUL_MAT_ID(ffn_moe_down-74): 321 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 363 us MUL_MAT_ID(ffn_moe_down-75): 547 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 367 us MUL_MAT_ID(ffn_moe_down-76): 203 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 392 us MUL_MAT_ID(ffn_moe_down-77): 212 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 430 us MUL_MAT_ID(ffn_moe_down-78): 218 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 418 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 410 us MUL_MAT_ID(ffn_moe_down-80): 204 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 417 us MUL_MAT_ID(ffn_moe_down-81): 187 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 377 us MUL_MAT_ID(ffn_moe_down-82): 200 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 371 us MUL_MAT_ID(ffn_moe_down-83): 408 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 360 us MUL_MAT_ID(ffn_moe_down-84): 335 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 413 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 415 us MUL_MAT_ID(ffn_moe_down-86): 229 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 408 us MUL_MAT_ID(ffn_moe_down-87): 187 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 380 us MUL_MAT_ID(ffn_moe_down-88): 220 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 403 us MUL_MAT_ID(ffn_moe_down-89): 375 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 404 us MUL_MAT_ID(ffn_moe_down-90): 288 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 391 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 20 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 380 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 429 us MUL_MAT_ID(ffn_moe_down-26): 471 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 506 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 412 us MUL_MAT_ID(ffn_moe_down-28): 228 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 460 us MUL_MAT_ID(ffn_moe_down-29): 185 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 459 us MUL_MAT_ID(ffn_moe_down-30): 185 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 456 us MUL_MAT_ID(ffn_moe_down-31): 306 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 407 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 421 us MUL_MAT_ID(ffn_moe_down-33): 215 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 434 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 439 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 411 us MUL_MAT_ID(ffn_moe_down-36): 208 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 370 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 391 us MUL_MAT_ID(ffn_moe_down-38): 258 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 407 us MUL_MAT_ID(ffn_moe_down-39): 246 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 363 us MUL_MAT_ID(ffn_moe_down-40): 347 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 361 us MUL_MAT_ID(ffn_moe_down-41): 193 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 377 us MUL_MAT_ID(ffn_moe_down-42): 188 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 404 us MUL_MAT_ID(ffn_moe_down-43): 409 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 398 us MUL_MAT_ID(ffn_moe_down-44): 351 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 402 us MUL_MAT_ID(ffn_moe_down-45): 202 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 402 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 400 us MUL_MAT_ID(ffn_moe_down-47): 376 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 357 us MUL_MAT_ID(ffn_moe_down-48): 196 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 376 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 361 us MUL_MAT_ID(ffn_moe_down-50): 207 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 368 us MUL_MAT_ID(ffn_moe_down-51): 189 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 523 us MUL_MAT_ID(ffn_moe_down-52): 591 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 483 us MUL_MAT_ID(ffn_moe_down-53): 247 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 454 us MUL_MAT_ID(ffn_moe_down-54): 183 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 451 us MUL_MAT_ID(ffn_moe_down-55): 363 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 434 us MUL_MAT_ID(ffn_moe_down-56): 264 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 603 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 508 us MUL_MAT_ID(ffn_moe_down-58): 206 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 414 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 475 us MUL_MAT_ID(ffn_moe_down-60): 183 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 419 us MUL_MAT_ID(ffn_moe_down-61): 206 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 455 us MUL_MAT_ID(ffn_moe_down-62): 499 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 432 us MUL_MAT_ID(ffn_moe_down-63): 166 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 395 us MUL_MAT_ID(ffn_moe_down-64): 205 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 409 us MUL_MAT_ID(ffn_moe_down-65): 238 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 395 us MUL_MAT_ID(ffn_moe_down-66): 202 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 629 us MUL_MAT_ID(ffn_moe_down-67): 316 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 393 us MUL_MAT_ID(ffn_moe_down-68): 299 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 459 us MUL_MAT_ID(ffn_moe_down-69): 308 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 514 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 453 us MUL_MAT_ID(ffn_moe_down-71): 212 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 413 us MUL_MAT_ID(ffn_moe_down-72): 210 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 431 us MUL_MAT_ID(ffn_moe_down-73): 292 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 444 us MUL_MAT_ID(ffn_moe_down-74): 219 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 466 us MUL_MAT_ID(ffn_moe_down-75): 325 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 386 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 449 us MUL_MAT_ID(ffn_moe_down-77): 199 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 360 us MUL_MAT_ID(ffn_moe_down-78): 185 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 389 us MUL_MAT_ID(ffn_moe_down-79): 208 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 365 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 376 us MUL_MAT_ID(ffn_moe_down-81): 416 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 377 us MUL_MAT_ID(ffn_moe_down-82): 187 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 501 us MUL_MAT_ID(ffn_moe_down-83): 201 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 380 us MUL_MAT_ID(ffn_moe_down-84): 246 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 429 us MUL_MAT_ID(ffn_moe_down-85): 377 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 407 us MUL_MAT_ID(ffn_moe_down-86): 223 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 404 us MUL_MAT_ID(ffn_moe_down-87): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 386 us MUL_MAT_ID(ffn_moe_down-88): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 374 us MUL_MAT_ID(ffn_moe_down-89): 202 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 389 us MUL_MAT_ID(ffn_moe_down-90): 249 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 386 us MUL_MAT_ID(ffn_moe_down-91): 247 us ggml_barrier(...): 48 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 390 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 360 us MUL_MAT_ID(ffn_moe_down-26): 180 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 375 us MUL_MAT_ID(ffn_moe_down-27): 186 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 369 us MUL_MAT_ID(ffn_moe_down-28): 316 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 378 us MUL_MAT_ID(ffn_moe_down-29): 182 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 415 us MUL_MAT_ID(ffn_moe_down-30): 566 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 415 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 438 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 359 us MUL_MAT_ID(ffn_moe_down-33): 190 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 458 us MUL_MAT_ID(ffn_moe_down-34): 185 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 438 us MUL_MAT_ID(ffn_moe_down-35): 184 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 412 us MUL_MAT_ID(ffn_moe_down-36): 220 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 441 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 455 us MUL_MAT_ID(ffn_moe_down-38): 214 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 380 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 421 us MUL_MAT_ID(ffn_moe_down-40): 226 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 336 us MUL_MAT_ID(ffn_moe_down-41): 188 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 391 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 340 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 418 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 413 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 435 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 391 us MUL_MAT_ID(ffn_moe_down-47): 207 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 376 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 420 us MUL_MAT_ID(ffn_moe_down-49): 221 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 463 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 416 us MUL_MAT_ID(ffn_moe_down-51): 217 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 374 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 396 us MUL_MAT_ID(ffn_moe_down-53): 593 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 409 us MUL_MAT_ID(ffn_moe_down-54): 219 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 390 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 455 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 428 us MUL_MAT_ID(ffn_moe_down-57): 217 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 417 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 456 us MUL_MAT_ID(ffn_moe_down-59): 187 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 363 us MUL_MAT_ID(ffn_moe_down-60): 206 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 448 us MUL_MAT_ID(ffn_moe_down-61): 220 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 390 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 358 us MUL_MAT_ID(ffn_moe_down-63): 406 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 397 us MUL_MAT_ID(ffn_moe_down-64): 200 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 396 us MUL_MAT_ID(ffn_moe_down-65): 221 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 414 us MUL_MAT_ID(ffn_moe_down-66): 223 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 437 us MUL_MAT_ID(ffn_moe_down-67): 202 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 396 us MUL_MAT_ID(ffn_moe_down-68): 185 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 506 us MUL_MAT_ID(ffn_moe_down-69): 238 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 399 us MUL_MAT_ID(ffn_moe_down-70): 207 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 637 us MUL_MAT_ID(ffn_moe_down-71): 258 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 419 us MUL_MAT_ID(ffn_moe_down-72): 221 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 380 us MUL_MAT_ID(ffn_moe_down-73): 217 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 344 us MUL_MAT_ID(ffn_moe_down-74): 187 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 371 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 576 us MUL_MAT_ID(ffn_moe_down-76): 413 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 413 us MUL_MAT_ID(ffn_moe_down-77): 191 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 377 us MUL_MAT_ID(ffn_moe_down-78): 187 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 409 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 412 us MUL_MAT_ID(ffn_moe_down-80): 202 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 402 us MUL_MAT_ID(ffn_moe_down-81): 196 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 421 us MUL_MAT_ID(ffn_moe_down-82): 313 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 444 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 387 us MUL_MAT_ID(ffn_moe_down-84): 243 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 402 us MUL_MAT_ID(ffn_moe_down-85): 242 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 340 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 449 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 364 us MUL_MAT_ID(ffn_moe_down-88): 530 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 373 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 439 us MUL_MAT_ID(ffn_moe_down-90): 240 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 422 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 33 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 392 us MUL_MAT_ID(ffn_moe_down-25): 188 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 412 us MUL_MAT_ID(ffn_moe_down-26): 186 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 365 us MUL_MAT_ID(ffn_moe_down-27): 225 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 364 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 408 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 367 us MUL_MAT_ID(ffn_moe_down-30): 202 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 392 us MUL_MAT_ID(ffn_moe_down-31): 190 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 392 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 364 us MUL_MAT_ID(ffn_moe_down-33): 211 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 409 us MUL_MAT_ID(ffn_moe_down-34): 367 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 417 us MUL_MAT_ID(ffn_moe_down-35): 215 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 424 us MUL_MAT_ID(ffn_moe_down-36): 192 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 380 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 212 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 417 us MUL_MAT_ID(ffn_moe_down-39): 286 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 413 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 355 us MUL_MAT_ID(ffn_moe_down-41): 372 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 350 us MUL_MAT_ID(ffn_moe_down-42): 179 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 398 us MUL_MAT_ID(ffn_moe_down-43): 393 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 395 us MUL_MAT_ID(ffn_moe_down-44): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 422 us MUL_MAT_ID(ffn_moe_down-45): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 464 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 403 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 387 us MUL_MAT_ID(ffn_moe_down-48): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 349 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 363 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 443 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 398 us MUL_MAT_ID(ffn_moe_down-52): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 365 us MUL_MAT_ID(ffn_moe_down-53): 223 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 392 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 442 us MUL_MAT_ID(ffn_moe_down-55): 436 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 378 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 402 us MUL_MAT_ID(ffn_moe_down-57): 547 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 357 us MUL_MAT_ID(ffn_moe_down-58): 214 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 403 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 393 us MUL_MAT_ID(ffn_moe_down-60): 218 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 567 us MUL_MAT_ID(ffn_moe_down-61): 340 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 601 us MUL_MAT_ID(ffn_moe_down-62): 323 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 392 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 402 us MUL_MAT_ID(ffn_moe_down-64): 180 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 446 us MUL_MAT_ID(ffn_moe_down-65): 208 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 439 us MUL_MAT_ID(ffn_moe_down-66): 187 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 398 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 395 us MUL_MAT_ID(ffn_moe_down-68): 201 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 372 us MUL_MAT_ID(ffn_moe_down-69): 215 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 424 us MUL_MAT_ID(ffn_moe_down-70): 191 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 416 us MUL_MAT_ID(ffn_moe_down-71): 219 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 417 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 346 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 372 us MUL_MAT_ID(ffn_moe_down-73): 566 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 406 us MUL_MAT_ID(ffn_moe_down-74): 194 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 381 us MUL_MAT_ID(ffn_moe_down-75): 452 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 382 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 377 us MUL_MAT_ID(ffn_moe_down-77): 187 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 378 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 405 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 413 us MUL_MAT_ID(ffn_moe_down-80): 200 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 377 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 405 us MUL_MAT_ID(ffn_moe_down-82): 207 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 418 us MUL_MAT_ID(ffn_moe_down-83): 233 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 463 us MUL_MAT_ID(ffn_moe_down-84): 204 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 411 us MUL_MAT_ID(ffn_moe_down-85): 716 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 414 us MUL_MAT_ID(ffn_moe_down-86): 214 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 410 us MUL_MAT_ID(ffn_moe_down-87): 238 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 416 us MUL_MAT_ID(ffn_moe_down-88): 188 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 346 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 403 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 611 us MUL_MAT_ID(ffn_moe_down-91): 363 us ggml_barrier(...): 5 us | glm4moe 355B.A32B Q3_K - Medium | 155.60 GiB | 369.08 B | CUDA | 94 | 8192 | 8192 | q8_0 | q8_0 | graph | 100.00 | tg128 | 11.11 ± 0.00 | Device 0: 11.8203 MiB Device 1: 12.2188 MiB Device 2: 12.6172 MiB Device 3: 12.2188 MiB GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 604 us MUL_MAT_ID(ffn_moe_down-25): 366 us ggml_barrier(...): 460 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 589 us MUL_MAT_ID(ffn_moe_down-26): 249 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 510 us MUL_MAT_ID(ffn_moe_down-27): 393 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 548 us MUL_MAT_ID(ffn_moe_down-28): 250 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 541 us MUL_MAT_ID(ffn_moe_down-29): 292 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 488 us MUL_MAT_ID(ffn_moe_down-30): 366 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 527 us MUL_MAT_ID(ffn_moe_down-31): 229 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 593 us MUL_MAT_ID(ffn_moe_down-32): 304 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 487 us MUL_MAT_ID(ffn_moe_down-33): 243 us ggml_barrier(...): 379 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 560 us MUL_MAT_ID(ffn_moe_down-34): 256 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 568 us MUL_MAT_ID(ffn_moe_down-35): 330 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 570 us MUL_MAT_ID(ffn_moe_down-36): 338 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 497 us MUL_MAT_ID(ffn_moe_down-37): 287 us ggml_barrier(...): 348 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 498 us MUL_MAT_ID(ffn_moe_down-38): 309 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 535 us MUL_MAT_ID(ffn_moe_down-39): 307 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 548 us MUL_MAT_ID(ffn_moe_down-40): 260 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 531 us MUL_MAT_ID(ffn_moe_down-41): 368 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 534 us MUL_MAT_ID(ffn_moe_down-42): 326 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 511 us MUL_MAT_ID(ffn_moe_down-43): 232 us ggml_barrier(...): 399 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 497 us MUL_MAT_ID(ffn_moe_down-44): 287 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 562 us MUL_MAT_ID(ffn_moe_down-45): 238 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 526 us MUL_MAT_ID(ffn_moe_down-46): 288 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 495 us MUL_MAT_ID(ffn_moe_down-47): 293 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 456 us MUL_MAT_ID(ffn_moe_down-48): 263 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 557 us MUL_MAT_ID(ffn_moe_down-49): 284 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 532 us MUL_MAT_ID(ffn_moe_down-50): 248 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 575 us MUL_MAT_ID(ffn_moe_down-51): 351 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 595 us MUL_MAT_ID(ffn_moe_down-52): 361 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 621 us MUL_MAT_ID(ffn_moe_down-53): 218 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 519 us MUL_MAT_ID(ffn_moe_down-54): 236 us ggml_barrier(...): 348 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 578 us MUL_MAT_ID(ffn_moe_down-55): 287 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 582 us MUL_MAT_ID(ffn_moe_down-56): 295 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 560 us MUL_MAT_ID(ffn_moe_down-57): 312 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 621 us MUL_MAT_ID(ffn_moe_down-58): 268 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 618 us MUL_MAT_ID(ffn_moe_down-59): 331 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 522 us MUL_MAT_ID(ffn_moe_down-60): 414 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 501 us MUL_MAT_ID(ffn_moe_down-61): 311 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 475 us MUL_MAT_ID(ffn_moe_down-62): 328 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 580 us MUL_MAT_ID(ffn_moe_down-63): 362 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 525 us MUL_MAT_ID(ffn_moe_down-64): 347 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 510 us MUL_MAT_ID(ffn_moe_down-65): 376 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 521 us MUL_MAT_ID(ffn_moe_down-66): 356 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 568 us MUL_MAT_ID(ffn_moe_down-67): 335 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 593 us MUL_MAT_ID(ffn_moe_down-68): 299 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 538 us MUL_MAT_ID(ffn_moe_down-69): 265 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 528 us MUL_MAT_ID(ffn_moe_down-70): 305 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 490 us MUL_MAT_ID(ffn_moe_down-71): 345 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 564 us MUL_MAT_ID(ffn_moe_down-72): 311 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 584 us MUL_MAT_ID(ffn_moe_down-73): 251 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 487 us MUL_MAT_ID(ffn_moe_down-74): 395 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 520 us MUL_MAT_ID(ffn_moe_down-75): 301 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 482 us MUL_MAT_ID(ffn_moe_down-76): 351 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 536 us MUL_MAT_ID(ffn_moe_down-77): 239 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 541 us MUL_MAT_ID(ffn_moe_down-78): 251 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 573 us MUL_MAT_ID(ffn_moe_down-79): 243 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 591 us MUL_MAT_ID(ffn_moe_down-80): 241 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 488 us MUL_MAT_ID(ffn_moe_down-81): 262 us ggml_barrier(...): 359 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 517 us MUL_MAT_ID(ffn_moe_down-82): 217 us ggml_barrier(...): 348 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 516 us MUL_MAT_ID(ffn_moe_down-83): 237 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 512 us MUL_MAT_ID(ffn_moe_down-84): 248 us ggml_barrier(...): 356 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 506 us MUL_MAT_ID(ffn_moe_down-85): 252 us ggml_barrier(...): 375 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 588 us MUL_MAT_ID(ffn_moe_down-86): 256 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 500 us MUL_MAT_ID(ffn_moe_down-87): 265 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 607 us MUL_MAT_ID(ffn_moe_down-88): 290 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 541 us MUL_MAT_ID(ffn_moe_down-89): 281 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 502 us MUL_MAT_ID(ffn_moe_down-90): 328 us ggml_barrier(...): 360 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 459 us MUL_MAT_ID(ffn_moe_down-91): 277 us ggml_barrier(...): 386 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 530 us MUL_MAT_ID(ffn_moe_down-25): 359 us ggml_barrier(...): 478 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 540 us MUL_MAT_ID(ffn_moe_down-26): 316 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 493 us MUL_MAT_ID(ffn_moe_down-27): 379 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 461 us MUL_MAT_ID(ffn_moe_down-28): 301 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 648 us MUL_MAT_ID(ffn_moe_down-29): 226 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 522 us MUL_MAT_ID(ffn_moe_down-30): 277 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 496 us MUL_MAT_ID(ffn_moe_down-31): 345 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 566 us MUL_MAT_ID(ffn_moe_down-32): 259 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 538 us MUL_MAT_ID(ffn_moe_down-33): 293 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 539 us MUL_MAT_ID(ffn_moe_down-34): 294 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 514 us MUL_MAT_ID(ffn_moe_down-35): 258 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 524 us MUL_MAT_ID(ffn_moe_down-36): 313 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 603 us MUL_MAT_ID(ffn_moe_down-37): 331 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 561 us MUL_MAT_ID(ffn_moe_down-38): 332 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 611 us MUL_MAT_ID(ffn_moe_down-39): 381 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 498 us MUL_MAT_ID(ffn_moe_down-40): 310 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 518 us MUL_MAT_ID(ffn_moe_down-41): 472 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 579 us MUL_MAT_ID(ffn_moe_down-42): 257 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 529 us MUL_MAT_ID(ffn_moe_down-43): 356 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 544 us MUL_MAT_ID(ffn_moe_down-44): 208 us ggml_barrier(...): 346 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 491 us MUL_MAT_ID(ffn_moe_down-45): 384 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 546 us MUL_MAT_ID(ffn_moe_down-46): 288 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 508 us MUL_MAT_ID(ffn_moe_down-47): 253 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 523 us MUL_MAT_ID(ffn_moe_down-48): 271 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 493 us MUL_MAT_ID(ffn_moe_down-49): 228 us ggml_barrier(...): 347 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 541 us MUL_MAT_ID(ffn_moe_down-50): 256 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 502 us MUL_MAT_ID(ffn_moe_down-51): 265 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 495 us MUL_MAT_ID(ffn_moe_down-52): 353 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 526 us MUL_MAT_ID(ffn_moe_down-53): 329 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 478 us MUL_MAT_ID(ffn_moe_down-54): 279 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 581 us MUL_MAT_ID(ffn_moe_down-55): 207 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 544 us MUL_MAT_ID(ffn_moe_down-56): 423 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 527 us MUL_MAT_ID(ffn_moe_down-57): 315 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 478 us MUL_MAT_ID(ffn_moe_down-58): 330 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 485 us MUL_MAT_ID(ffn_moe_down-59): 346 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 459 us MUL_MAT_ID(ffn_moe_down-60): 283 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 558 us MUL_MAT_ID(ffn_moe_down-61): 260 us ggml_barrier(...): 322 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 597 us MUL_MAT_ID(ffn_moe_down-62): 280 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 657 us MUL_MAT_ID(ffn_moe_down-63): 264 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 540 us MUL_MAT_ID(ffn_moe_down-64): 330 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 596 us MUL_MAT_ID(ffn_moe_down-65): 249 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 455 us MUL_MAT_ID(ffn_moe_down-66): 260 us ggml_barrier(...): 363 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 474 us MUL_MAT_ID(ffn_moe_down-67): 275 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 583 us MUL_MAT_ID(ffn_moe_down-68): 292 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 541 us MUL_MAT_ID(ffn_moe_down-69): 322 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 527 us MUL_MAT_ID(ffn_moe_down-70): 398 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 466 us MUL_MAT_ID(ffn_moe_down-71): 353 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 491 us MUL_MAT_ID(ffn_moe_down-72): 274 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 506 us MUL_MAT_ID(ffn_moe_down-73): 302 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 505 us MUL_MAT_ID(ffn_moe_down-74): 263 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 543 us MUL_MAT_ID(ffn_moe_down-75): 304 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 589 us MUL_MAT_ID(ffn_moe_down-76): 235 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 517 us MUL_MAT_ID(ffn_moe_down-77): 276 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 485 us MUL_MAT_ID(ffn_moe_down-78): 247 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 492 us MUL_MAT_ID(ffn_moe_down-79): 334 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 509 us MUL_MAT_ID(ffn_moe_down-80): 235 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 483 us MUL_MAT_ID(ffn_moe_down-81): 268 us ggml_barrier(...): 357 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 542 us MUL_MAT_ID(ffn_moe_down-82): 209 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 525 us MUL_MAT_ID(ffn_moe_down-83): 361 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 460 us MUL_MAT_ID(ffn_moe_down-84): 369 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 523 us MUL_MAT_ID(ffn_moe_down-85): 279 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 588 us MUL_MAT_ID(ffn_moe_down-86): 306 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 536 us MUL_MAT_ID(ffn_moe_down-87): 235 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 475 us MUL_MAT_ID(ffn_moe_down-88): 270 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 585 us MUL_MAT_ID(ffn_moe_down-89): 346 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 525 us MUL_MAT_ID(ffn_moe_down-90): 290 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 571 us MUL_MAT_ID(ffn_moe_down-91): 333 us ggml_barrier(...): 195 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 551 us MUL_MAT_ID(ffn_moe_down-25): 330 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 470 us MUL_MAT_ID(ffn_moe_down-26): 433 us ggml_barrier(...): 339 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 506 us MUL_MAT_ID(ffn_moe_down-27): 370 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 467 us MUL_MAT_ID(ffn_moe_down-28): 267 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 501 us MUL_MAT_ID(ffn_moe_down-29): 298 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 495 us MUL_MAT_ID(ffn_moe_down-30): 273 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 447 us MUL_MAT_ID(ffn_moe_down-31): 315 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 441 us MUL_MAT_ID(ffn_moe_down-32): 211 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 428 us MUL_MAT_ID(ffn_moe_down-33): 366 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 446 us MUL_MAT_ID(ffn_moe_down-34): 210 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 564 us MUL_MAT_ID(ffn_moe_down-35): 268 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 472 us MUL_MAT_ID(ffn_moe_down-36): 277 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 523 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 475 us MUL_MAT_ID(ffn_moe_down-38): 425 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 471 us MUL_MAT_ID(ffn_moe_down-39): 466 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 497 us MUL_MAT_ID(ffn_moe_down-40): 299 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 538 us MUL_MAT_ID(ffn_moe_down-41): 222 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 479 us MUL_MAT_ID(ffn_moe_down-42): 264 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 532 us MUL_MAT_ID(ffn_moe_down-43): 230 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 528 us MUL_MAT_ID(ffn_moe_down-44): 290 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 456 us MUL_MAT_ID(ffn_moe_down-45): 298 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 530 us MUL_MAT_ID(ffn_moe_down-46): 302 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 505 us MUL_MAT_ID(ffn_moe_down-47): 222 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 623 us MUL_MAT_ID(ffn_moe_down-48): 289 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 453 us MUL_MAT_ID(ffn_moe_down-49): 232 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 429 us MUL_MAT_ID(ffn_moe_down-50): 266 us ggml_barrier(...): 344 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 485 us MUL_MAT_ID(ffn_moe_down-51): 277 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 478 us MUL_MAT_ID(ffn_moe_down-52): 295 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 479 us MUL_MAT_ID(ffn_moe_down-53): 277 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 481 us MUL_MAT_ID(ffn_moe_down-54): 228 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 563 us MUL_MAT_ID(ffn_moe_down-55): 361 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 574 us MUL_MAT_ID(ffn_moe_down-56): 242 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 504 us MUL_MAT_ID(ffn_moe_down-57): 236 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 408 us MUL_MAT_ID(ffn_moe_down-58): 236 us ggml_barrier(...): 395 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 485 us MUL_MAT_ID(ffn_moe_down-59): 354 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 485 us MUL_MAT_ID(ffn_moe_down-60): 226 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 445 us MUL_MAT_ID(ffn_moe_down-61): 290 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 533 us MUL_MAT_ID(ffn_moe_down-62): 293 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 472 us MUL_MAT_ID(ffn_moe_down-63): 206 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 539 us MUL_MAT_ID(ffn_moe_down-64): 281 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 481 us MUL_MAT_ID(ffn_moe_down-65): 360 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 546 us MUL_MAT_ID(ffn_moe_down-66): 224 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 434 us MUL_MAT_ID(ffn_moe_down-67): 298 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 559 us MUL_MAT_ID(ffn_moe_down-68): 299 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 547 us MUL_MAT_ID(ffn_moe_down-69): 224 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 546 us MUL_MAT_ID(ffn_moe_down-70): 231 us ggml_barrier(...): 371 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 504 us MUL_MAT_ID(ffn_moe_down-71): 286 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 480 us MUL_MAT_ID(ffn_moe_down-72): 215 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 549 us MUL_MAT_ID(ffn_moe_down-73): 278 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 499 us MUL_MAT_ID(ffn_moe_down-74): 260 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 521 us MUL_MAT_ID(ffn_moe_down-75): 271 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 526 us MUL_MAT_ID(ffn_moe_down-76): 343 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 474 us MUL_MAT_ID(ffn_moe_down-77): 277 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 457 us MUL_MAT_ID(ffn_moe_down-78): 510 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 468 us MUL_MAT_ID(ffn_moe_down-79): 308 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 479 us MUL_MAT_ID(ffn_moe_down-80): 323 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 496 us MUL_MAT_ID(ffn_moe_down-81): 343 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 503 us MUL_MAT_ID(ffn_moe_down-82): 222 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 593 us MUL_MAT_ID(ffn_moe_down-83): 289 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 580 us MUL_MAT_ID(ffn_moe_down-84): 326 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 461 us MUL_MAT_ID(ffn_moe_down-85): 326 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 443 us MUL_MAT_ID(ffn_moe_down-86): 679 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 521 us MUL_MAT_ID(ffn_moe_down-87): 224 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 473 us MUL_MAT_ID(ffn_moe_down-88): 255 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 430 us MUL_MAT_ID(ffn_moe_down-89): 249 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 547 us MUL_MAT_ID(ffn_moe_down-90): 322 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 526 us MUL_MAT_ID(ffn_moe_down-91): 291 us ggml_barrier(...): 293 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 492 us MUL_MAT_ID(ffn_moe_down-25): 250 us ggml_barrier(...): 545 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 428 us MUL_MAT_ID(ffn_moe_down-26): 325 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 443 us MUL_MAT_ID(ffn_moe_down-27): 293 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 450 us MUL_MAT_ID(ffn_moe_down-28): 326 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 458 us MUL_MAT_ID(ffn_moe_down-29): 292 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 430 us MUL_MAT_ID(ffn_moe_down-30): 412 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 489 us MUL_MAT_ID(ffn_moe_down-31): 212 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 490 us MUL_MAT_ID(ffn_moe_down-32): 277 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 504 us MUL_MAT_ID(ffn_moe_down-33): 248 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 422 us MUL_MAT_ID(ffn_moe_down-34): 293 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 418 us MUL_MAT_ID(ffn_moe_down-35): 231 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 408 us MUL_MAT_ID(ffn_moe_down-36): 250 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 381 us MUL_MAT_ID(ffn_moe_down-37): 263 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 467 us MUL_MAT_ID(ffn_moe_down-38): 305 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 464 us MUL_MAT_ID(ffn_moe_down-39): 303 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 449 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 408 us MUL_MAT_ID(ffn_moe_down-41): 288 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 510 us MUL_MAT_ID(ffn_moe_down-42): 325 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 528 us MUL_MAT_ID(ffn_moe_down-43): 296 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 516 us MUL_MAT_ID(ffn_moe_down-44): 238 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 441 us MUL_MAT_ID(ffn_moe_down-45): 270 us ggml_barrier(...): 341 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 531 us MUL_MAT_ID(ffn_moe_down-46): 236 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 460 us MUL_MAT_ID(ffn_moe_down-47): 250 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 424 us MUL_MAT_ID(ffn_moe_down-48): 230 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 532 us MUL_MAT_ID(ffn_moe_down-49): 291 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 462 us MUL_MAT_ID(ffn_moe_down-50): 246 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 511 us MUL_MAT_ID(ffn_moe_down-51): 446 us ggml_barrier(...): 338 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 479 us MUL_MAT_ID(ffn_moe_down-52): 244 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 490 us MUL_MAT_ID(ffn_moe_down-53): 213 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 503 us MUL_MAT_ID(ffn_moe_down-54): 321 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 568 us MUL_MAT_ID(ffn_moe_down-55): 241 us ggml_barrier(...): 348 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 498 us MUL_MAT_ID(ffn_moe_down-56): 210 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 522 us MUL_MAT_ID(ffn_moe_down-57): 252 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 463 us MUL_MAT_ID(ffn_moe_down-58): 265 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 431 us MUL_MAT_ID(ffn_moe_down-59): 213 us ggml_barrier(...): 345 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 472 us MUL_MAT_ID(ffn_moe_down-60): 251 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 468 us MUL_MAT_ID(ffn_moe_down-61): 388 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 442 us MUL_MAT_ID(ffn_moe_down-62): 238 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 409 us MUL_MAT_ID(ffn_moe_down-63): 310 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 546 us MUL_MAT_ID(ffn_moe_down-64): 288 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 402 us MUL_MAT_ID(ffn_moe_down-65): 225 us ggml_barrier(...): 334 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 452 us MUL_MAT_ID(ffn_moe_down-66): 255 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 491 us MUL_MAT_ID(ffn_moe_down-67): 297 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 462 us MUL_MAT_ID(ffn_moe_down-68): 316 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 502 us MUL_MAT_ID(ffn_moe_down-69): 295 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 549 us MUL_MAT_ID(ffn_moe_down-70): 251 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 591 us MUL_MAT_ID(ffn_moe_down-71): 232 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 414 us MUL_MAT_ID(ffn_moe_down-72): 271 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 451 us MUL_MAT_ID(ffn_moe_down-73): 303 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 515 us MUL_MAT_ID(ffn_moe_down-74): 463 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 531 us MUL_MAT_ID(ffn_moe_down-75): 212 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 615 us MUL_MAT_ID(ffn_moe_down-76): 334 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 521 us MUL_MAT_ID(ffn_moe_down-77): 255 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 534 us MUL_MAT_ID(ffn_moe_down-78): 267 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 574 us MUL_MAT_ID(ffn_moe_down-79): 247 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 485 us MUL_MAT_ID(ffn_moe_down-80): 212 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 547 us MUL_MAT_ID(ffn_moe_down-81): 208 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 507 us MUL_MAT_ID(ffn_moe_down-82): 241 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 474 us MUL_MAT_ID(ffn_moe_down-83): 333 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 560 us MUL_MAT_ID(ffn_moe_down-84): 283 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 493 us MUL_MAT_ID(ffn_moe_down-85): 385 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 517 us MUL_MAT_ID(ffn_moe_down-86): 336 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 440 us MUL_MAT_ID(ffn_moe_down-87): 311 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 430 us MUL_MAT_ID(ffn_moe_down-88): 335 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 496 us MUL_MAT_ID(ffn_moe_down-89): 303 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 455 us MUL_MAT_ID(ffn_moe_down-90): 334 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 445 us MUL_MAT_ID(ffn_moe_down-91): 307 us ggml_barrier(...): 349 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 432 us MUL_MAT_ID(ffn_moe_down-25): 300 us ggml_barrier(...): 523 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 497 us MUL_MAT_ID(ffn_moe_down-26): 263 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 439 us MUL_MAT_ID(ffn_moe_down-27): 292 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 434 us MUL_MAT_ID(ffn_moe_down-28): 250 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 456 us MUL_MAT_ID(ffn_moe_down-29): 252 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 523 us MUL_MAT_ID(ffn_moe_down-30): 284 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 495 us MUL_MAT_ID(ffn_moe_down-31): 284 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 466 us MUL_MAT_ID(ffn_moe_down-32): 223 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 404 us MUL_MAT_ID(ffn_moe_down-33): 245 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 468 us MUL_MAT_ID(ffn_moe_down-34): 222 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 481 us MUL_MAT_ID(ffn_moe_down-35): 224 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 422 us MUL_MAT_ID(ffn_moe_down-36): 185 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 459 us MUL_MAT_ID(ffn_moe_down-37): 306 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 437 us MUL_MAT_ID(ffn_moe_down-38): 237 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 501 us MUL_MAT_ID(ffn_moe_down-39): 324 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 555 us MUL_MAT_ID(ffn_moe_down-40): 251 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 581 us MUL_MAT_ID(ffn_moe_down-41): 268 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 458 us MUL_MAT_ID(ffn_moe_down-42): 232 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 484 us MUL_MAT_ID(ffn_moe_down-43): 278 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 370 us MUL_MAT_ID(ffn_moe_down-44): 246 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 531 us MUL_MAT_ID(ffn_moe_down-45): 211 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 520 us MUL_MAT_ID(ffn_moe_down-46): 449 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 570 us MUL_MAT_ID(ffn_moe_down-47): 252 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 502 us MUL_MAT_ID(ffn_moe_down-48): 294 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 556 us MUL_MAT_ID(ffn_moe_down-49): 274 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 440 us MUL_MAT_ID(ffn_moe_down-50): 219 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 557 us MUL_MAT_ID(ffn_moe_down-51): 396 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 449 us MUL_MAT_ID(ffn_moe_down-52): 296 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 458 us MUL_MAT_ID(ffn_moe_down-53): 299 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 532 us MUL_MAT_ID(ffn_moe_down-54): 218 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 461 us MUL_MAT_ID(ffn_moe_down-55): 288 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 487 us MUL_MAT_ID(ffn_moe_down-56): 261 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 488 us MUL_MAT_ID(ffn_moe_down-57): 331 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 489 us MUL_MAT_ID(ffn_moe_down-58): 374 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 486 us MUL_MAT_ID(ffn_moe_down-59): 399 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 502 us MUL_MAT_ID(ffn_moe_down-60): 342 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 532 us MUL_MAT_ID(ffn_moe_down-61): 334 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 497 us MUL_MAT_ID(ffn_moe_down-62): 303 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 568 us MUL_MAT_ID(ffn_moe_down-63): 225 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 550 us MUL_MAT_ID(ffn_moe_down-64): 362 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 506 us MUL_MAT_ID(ffn_moe_down-65): 245 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 493 us MUL_MAT_ID(ffn_moe_down-66): 242 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 412 us MUL_MAT_ID(ffn_moe_down-67): 245 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 477 us MUL_MAT_ID(ffn_moe_down-68): 267 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 483 us MUL_MAT_ID(ffn_moe_down-69): 631 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 541 us MUL_MAT_ID(ffn_moe_down-70): 316 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 537 us MUL_MAT_ID(ffn_moe_down-71): 306 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 464 us MUL_MAT_ID(ffn_moe_down-72): 291 us ggml_barrier(...): 356 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 532 us MUL_MAT_ID(ffn_moe_down-73): 313 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 608 us MUL_MAT_ID(ffn_moe_down-74): 251 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 495 us MUL_MAT_ID(ffn_moe_down-75): 301 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 483 us MUL_MAT_ID(ffn_moe_down-76): 314 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 514 us MUL_MAT_ID(ffn_moe_down-77): 265 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 502 us MUL_MAT_ID(ffn_moe_down-78): 236 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 446 us MUL_MAT_ID(ffn_moe_down-79): 274 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 569 us MUL_MAT_ID(ffn_moe_down-80): 255 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 446 us MUL_MAT_ID(ffn_moe_down-81): 241 us ggml_barrier(...): 347 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 446 us MUL_MAT_ID(ffn_moe_down-82): 292 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 537 us MUL_MAT_ID(ffn_moe_down-83): 236 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 556 us MUL_MAT_ID(ffn_moe_down-84): 558 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 552 us MUL_MAT_ID(ffn_moe_down-85): 237 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 516 us MUL_MAT_ID(ffn_moe_down-86): 244 us ggml_barrier(...): 443 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 446 us MUL_MAT_ID(ffn_moe_down-87): 233 us ggml_barrier(...): 361 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 463 us MUL_MAT_ID(ffn_moe_down-88): 215 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 555 us MUL_MAT_ID(ffn_moe_down-89): 320 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 490 us MUL_MAT_ID(ffn_moe_down-90): 328 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 566 us MUL_MAT_ID(ffn_moe_down-91): 312 us ggml_barrier(...): 275 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 774 us MUL_MAT_ID(ffn_moe_down-25): 209 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 460 us MUL_MAT_ID(ffn_moe_down-26): 268 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 392 us MUL_MAT_ID(ffn_moe_down-27): 236 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 468 us MUL_MAT_ID(ffn_moe_down-28): 229 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 371 us MUL_MAT_ID(ffn_moe_down-29): 207 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 413 us MUL_MAT_ID(ffn_moe_down-30): 219 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 510 us MUL_MAT_ID(ffn_moe_down-31): 284 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 414 us MUL_MAT_ID(ffn_moe_down-32): 217 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 444 us MUL_MAT_ID(ffn_moe_down-33): 293 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 499 us MUL_MAT_ID(ffn_moe_down-34): 253 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 406 us MUL_MAT_ID(ffn_moe_down-35): 205 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 437 us MUL_MAT_ID(ffn_moe_down-36): 233 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 407 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 522 us MUL_MAT_ID(ffn_moe_down-38): 258 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 455 us MUL_MAT_ID(ffn_moe_down-39): 204 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 424 us MUL_MAT_ID(ffn_moe_down-40): 258 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 510 us MUL_MAT_ID(ffn_moe_down-41): 263 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 415 us MUL_MAT_ID(ffn_moe_down-42): 255 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 416 us MUL_MAT_ID(ffn_moe_down-43): 246 us ggml_barrier(...): 322 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 525 us MUL_MAT_ID(ffn_moe_down-44): 259 us ggml_barrier(...): 406 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 383 us MUL_MAT_ID(ffn_moe_down-45): 299 us ggml_barrier(...): 356 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 511 us MUL_MAT_ID(ffn_moe_down-46): 238 us ggml_barrier(...): 339 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 472 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 442 us MUL_MAT_ID(ffn_moe_down-48): 402 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 446 us MUL_MAT_ID(ffn_moe_down-49): 241 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 359 us MUL_MAT_ID(ffn_moe_down-50): 211 us ggml_barrier(...): 348 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 452 us MUL_MAT_ID(ffn_moe_down-51): 400 us ggml_barrier(...): 384 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 488 us MUL_MAT_ID(ffn_moe_down-52): 279 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 440 us MUL_MAT_ID(ffn_moe_down-53): 331 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 513 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 449 us MUL_MAT_ID(ffn_moe_down-55): 257 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 460 us MUL_MAT_ID(ffn_moe_down-56): 334 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 513 us MUL_MAT_ID(ffn_moe_down-57): 289 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 496 us MUL_MAT_ID(ffn_moe_down-58): 235 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 450 us MUL_MAT_ID(ffn_moe_down-59): 228 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 532 us MUL_MAT_ID(ffn_moe_down-60): 311 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 494 us MUL_MAT_ID(ffn_moe_down-61): 309 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 487 us MUL_MAT_ID(ffn_moe_down-62): 294 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 507 us MUL_MAT_ID(ffn_moe_down-63): 283 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 535 us MUL_MAT_ID(ffn_moe_down-64): 352 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 546 us MUL_MAT_ID(ffn_moe_down-65): 248 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 464 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 547 us MUL_MAT_ID(ffn_moe_down-67): 323 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 489 us MUL_MAT_ID(ffn_moe_down-68): 196 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 525 us MUL_MAT_ID(ffn_moe_down-69): 336 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 556 us MUL_MAT_ID(ffn_moe_down-70): 267 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 498 us MUL_MAT_ID(ffn_moe_down-71): 214 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 531 us MUL_MAT_ID(ffn_moe_down-72): 208 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 485 us MUL_MAT_ID(ffn_moe_down-73): 291 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 433 us MUL_MAT_ID(ffn_moe_down-74): 257 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 454 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 441 us MUL_MAT_ID(ffn_moe_down-76): 268 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 431 us MUL_MAT_ID(ffn_moe_down-77): 222 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 468 us MUL_MAT_ID(ffn_moe_down-78): 258 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 506 us MUL_MAT_ID(ffn_moe_down-79): 338 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 471 us MUL_MAT_ID(ffn_moe_down-80): 296 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 527 us MUL_MAT_ID(ffn_moe_down-81): 285 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 435 us MUL_MAT_ID(ffn_moe_down-82): 305 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 539 us MUL_MAT_ID(ffn_moe_down-83): 344 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 447 us MUL_MAT_ID(ffn_moe_down-84): 555 us ggml_barrier(...): 356 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 447 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 401 us MUL_MAT_ID(ffn_moe_down-86): 452 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 546 us MUL_MAT_ID(ffn_moe_down-87): 315 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 487 us MUL_MAT_ID(ffn_moe_down-88): 355 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 442 us MUL_MAT_ID(ffn_moe_down-89): 232 us ggml_barrier(...): 339 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 478 us MUL_MAT_ID(ffn_moe_down-90): 296 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 474 us MUL_MAT_ID(ffn_moe_down-91): 241 us ggml_barrier(...): 238 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 426 us MUL_MAT_ID(ffn_moe_down-25): 341 us ggml_barrier(...): 542 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 378 us MUL_MAT_ID(ffn_moe_down-26): 244 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 471 us MUL_MAT_ID(ffn_moe_down-27): 198 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 479 us MUL_MAT_ID(ffn_moe_down-28): 224 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 395 us MUL_MAT_ID(ffn_moe_down-29): 210 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 419 us MUL_MAT_ID(ffn_moe_down-30): 198 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 397 us MUL_MAT_ID(ffn_moe_down-31): 264 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 477 us MUL_MAT_ID(ffn_moe_down-32): 295 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 468 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 432 us MUL_MAT_ID(ffn_moe_down-34): 205 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 462 us MUL_MAT_ID(ffn_moe_down-35): 312 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 485 us MUL_MAT_ID(ffn_moe_down-36): 219 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 432 us MUL_MAT_ID(ffn_moe_down-37): 297 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 427 us MUL_MAT_ID(ffn_moe_down-38): 341 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 394 us MUL_MAT_ID(ffn_moe_down-39): 246 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 548 us MUL_MAT_ID(ffn_moe_down-40): 354 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 406 us MUL_MAT_ID(ffn_moe_down-41): 297 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 410 us MUL_MAT_ID(ffn_moe_down-42): 238 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 432 us MUL_MAT_ID(ffn_moe_down-43): 241 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 476 us MUL_MAT_ID(ffn_moe_down-44): 207 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 446 us MUL_MAT_ID(ffn_moe_down-45): 592 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 421 us MUL_MAT_ID(ffn_moe_down-46): 263 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 490 us MUL_MAT_ID(ffn_moe_down-47): 290 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 429 us MUL_MAT_ID(ffn_moe_down-48): 248 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 517 us MUL_MAT_ID(ffn_moe_down-49): 262 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 510 us MUL_MAT_ID(ffn_moe_down-50): 240 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 539 us MUL_MAT_ID(ffn_moe_down-51): 239 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 446 us MUL_MAT_ID(ffn_moe_down-52): 303 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 438 us MUL_MAT_ID(ffn_moe_down-53): 452 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 445 us MUL_MAT_ID(ffn_moe_down-54): 204 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 438 us MUL_MAT_ID(ffn_moe_down-55): 231 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 448 us MUL_MAT_ID(ffn_moe_down-56): 406 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 437 us MUL_MAT_ID(ffn_moe_down-57): 227 us ggml_barrier(...): 356 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 438 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 466 us MUL_MAT_ID(ffn_moe_down-59): 247 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 532 us MUL_MAT_ID(ffn_moe_down-60): 268 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 469 us MUL_MAT_ID(ffn_moe_down-61): 205 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 441 us MUL_MAT_ID(ffn_moe_down-62): 285 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 497 us MUL_MAT_ID(ffn_moe_down-63): 266 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 437 us MUL_MAT_ID(ffn_moe_down-64): 263 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 417 us MUL_MAT_ID(ffn_moe_down-65): 220 us ggml_barrier(...): 348 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 486 us MUL_MAT_ID(ffn_moe_down-66): 285 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 408 us MUL_MAT_ID(ffn_moe_down-67): 220 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 423 us MUL_MAT_ID(ffn_moe_down-68): 252 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 503 us MUL_MAT_ID(ffn_moe_down-69): 233 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 510 us MUL_MAT_ID(ffn_moe_down-70): 284 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 528 us MUL_MAT_ID(ffn_moe_down-71): 251 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 498 us MUL_MAT_ID(ffn_moe_down-72): 296 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 564 us MUL_MAT_ID(ffn_moe_down-73): 214 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 445 us MUL_MAT_ID(ffn_moe_down-74): 326 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 480 us MUL_MAT_ID(ffn_moe_down-75): 316 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 542 us MUL_MAT_ID(ffn_moe_down-76): 437 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 515 us MUL_MAT_ID(ffn_moe_down-77): 219 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 444 us MUL_MAT_ID(ffn_moe_down-78): 295 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 401 us MUL_MAT_ID(ffn_moe_down-79): 439 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 453 us MUL_MAT_ID(ffn_moe_down-80): 209 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 507 us MUL_MAT_ID(ffn_moe_down-81): 235 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 521 us MUL_MAT_ID(ffn_moe_down-82): 420 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 587 us MUL_MAT_ID(ffn_moe_down-83): 256 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 396 us MUL_MAT_ID(ffn_moe_down-84): 245 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 589 us MUL_MAT_ID(ffn_moe_down-85): 590 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 532 us MUL_MAT_ID(ffn_moe_down-86): 226 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 518 us MUL_MAT_ID(ffn_moe_down-87): 255 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 416 us MUL_MAT_ID(ffn_moe_down-88): 432 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 518 us MUL_MAT_ID(ffn_moe_down-89): 347 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 554 us MUL_MAT_ID(ffn_moe_down-90): 690 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 589 us MUL_MAT_ID(ffn_moe_down-91): 249 us ggml_barrier(...): 222 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 404 us MUL_MAT_ID(ffn_moe_down-25): 261 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 434 us MUL_MAT_ID(ffn_moe_down-26): 203 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 510 us MUL_MAT_ID(ffn_moe_down-27): 198 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 407 us MUL_MAT_ID(ffn_moe_down-28): 195 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 480 us MUL_MAT_ID(ffn_moe_down-29): 238 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 478 us MUL_MAT_ID(ffn_moe_down-30): 205 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 396 us MUL_MAT_ID(ffn_moe_down-31): 214 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 493 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 444 us MUL_MAT_ID(ffn_moe_down-33): 295 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 437 us MUL_MAT_ID(ffn_moe_down-34): 243 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 401 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 467 us MUL_MAT_ID(ffn_moe_down-36): 280 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 416 us MUL_MAT_ID(ffn_moe_down-37): 299 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 443 us MUL_MAT_ID(ffn_moe_down-38): 260 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 432 us MUL_MAT_ID(ffn_moe_down-39): 225 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 553 us MUL_MAT_ID(ffn_moe_down-40): 644 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 428 us MUL_MAT_ID(ffn_moe_down-41): 210 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 558 us MUL_MAT_ID(ffn_moe_down-42): 221 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 540 us MUL_MAT_ID(ffn_moe_down-43): 240 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 502 us MUL_MAT_ID(ffn_moe_down-44): 214 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 430 us MUL_MAT_ID(ffn_moe_down-45): 216 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 549 us MUL_MAT_ID(ffn_moe_down-46): 225 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 443 us MUL_MAT_ID(ffn_moe_down-47): 287 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 377 us MUL_MAT_ID(ffn_moe_down-48): 265 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 457 us MUL_MAT_ID(ffn_moe_down-49): 230 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 442 us MUL_MAT_ID(ffn_moe_down-50): 314 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 425 us MUL_MAT_ID(ffn_moe_down-51): 278 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 420 us MUL_MAT_ID(ffn_moe_down-52): 207 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 427 us MUL_MAT_ID(ffn_moe_down-53): 199 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 441 us MUL_MAT_ID(ffn_moe_down-54): 244 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 500 us MUL_MAT_ID(ffn_moe_down-55): 296 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 459 us MUL_MAT_ID(ffn_moe_down-56): 283 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 462 us MUL_MAT_ID(ffn_moe_down-57): 212 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 414 us MUL_MAT_ID(ffn_moe_down-58): 276 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 447 us MUL_MAT_ID(ffn_moe_down-59): 487 us ggml_barrier(...): 418 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 418 us MUL_MAT_ID(ffn_moe_down-60): 228 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 437 us MUL_MAT_ID(ffn_moe_down-61): 232 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 416 us MUL_MAT_ID(ffn_moe_down-62): 230 us ggml_barrier(...): 377 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 470 us MUL_MAT_ID(ffn_moe_down-63): 281 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 475 us MUL_MAT_ID(ffn_moe_down-64): 391 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 451 us MUL_MAT_ID(ffn_moe_down-65): 208 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 458 us MUL_MAT_ID(ffn_moe_down-66): 277 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 501 us MUL_MAT_ID(ffn_moe_down-67): 308 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 388 us MUL_MAT_ID(ffn_moe_down-68): 276 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 435 us MUL_MAT_ID(ffn_moe_down-69): 198 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 455 us MUL_MAT_ID(ffn_moe_down-70): 313 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 431 us MUL_MAT_ID(ffn_moe_down-71): 255 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 465 us MUL_MAT_ID(ffn_moe_down-72): 578 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 522 us MUL_MAT_ID(ffn_moe_down-73): 223 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 491 us MUL_MAT_ID(ffn_moe_down-74): 306 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 542 us MUL_MAT_ID(ffn_moe_down-75): 271 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 502 us MUL_MAT_ID(ffn_moe_down-76): 238 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 497 us MUL_MAT_ID(ffn_moe_down-77): 311 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 457 us MUL_MAT_ID(ffn_moe_down-78): 278 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 556 us MUL_MAT_ID(ffn_moe_down-79): 266 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 464 us MUL_MAT_ID(ffn_moe_down-80): 269 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 482 us MUL_MAT_ID(ffn_moe_down-81): 378 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 612 us MUL_MAT_ID(ffn_moe_down-82): 291 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 472 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 514 us MUL_MAT_ID(ffn_moe_down-84): 454 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 486 us MUL_MAT_ID(ffn_moe_down-85): 308 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 419 us MUL_MAT_ID(ffn_moe_down-86): 221 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 509 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 464 us MUL_MAT_ID(ffn_moe_down-88): 185 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 506 us MUL_MAT_ID(ffn_moe_down-89): 267 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 436 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 464 us MUL_MAT_ID(ffn_moe_down-91): 242 us ggml_barrier(...): 282 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 521 us MUL_MAT_ID(ffn_moe_down-25): 308 us ggml_barrier(...): 451 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 446 us MUL_MAT_ID(ffn_moe_down-26): 233 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 365 us MUL_MAT_ID(ffn_moe_down-27): 246 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 467 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 421 us MUL_MAT_ID(ffn_moe_down-29): 256 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 432 us MUL_MAT_ID(ffn_moe_down-30): 299 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 378 us MUL_MAT_ID(ffn_moe_down-31): 363 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 386 us MUL_MAT_ID(ffn_moe_down-32): 208 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 468 us MUL_MAT_ID(ffn_moe_down-33): 197 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 412 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 421 us MUL_MAT_ID(ffn_moe_down-35): 239 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 481 us MUL_MAT_ID(ffn_moe_down-36): 188 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 502 us MUL_MAT_ID(ffn_moe_down-37): 304 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 472 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 448 us MUL_MAT_ID(ffn_moe_down-39): 286 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 457 us MUL_MAT_ID(ffn_moe_down-40): 237 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 464 us MUL_MAT_ID(ffn_moe_down-41): 168 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 386 us MUL_MAT_ID(ffn_moe_down-42): 222 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 414 us MUL_MAT_ID(ffn_moe_down-43): 226 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 391 us MUL_MAT_ID(ffn_moe_down-44): 232 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 375 us MUL_MAT_ID(ffn_moe_down-45): 249 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 390 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 506 us MUL_MAT_ID(ffn_moe_down-47): 392 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 388 us MUL_MAT_ID(ffn_moe_down-48): 377 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 425 us MUL_MAT_ID(ffn_moe_down-49): 209 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 457 us MUL_MAT_ID(ffn_moe_down-50): 275 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 431 us MUL_MAT_ID(ffn_moe_down-51): 250 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 418 us MUL_MAT_ID(ffn_moe_down-52): 215 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 478 us MUL_MAT_ID(ffn_moe_down-53): 401 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 421 us MUL_MAT_ID(ffn_moe_down-54): 240 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 472 us MUL_MAT_ID(ffn_moe_down-55): 202 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 456 us MUL_MAT_ID(ffn_moe_down-56): 214 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 505 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 452 us MUL_MAT_ID(ffn_moe_down-58): 198 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 439 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 518 us MUL_MAT_ID(ffn_moe_down-60): 214 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 463 us MUL_MAT_ID(ffn_moe_down-61): 236 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 398 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 347 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 527 us MUL_MAT_ID(ffn_moe_down-63): 278 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 367 us MUL_MAT_ID(ffn_moe_down-64): 213 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 490 us MUL_MAT_ID(ffn_moe_down-65): 210 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 433 us MUL_MAT_ID(ffn_moe_down-66): 217 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 505 us MUL_MAT_ID(ffn_moe_down-67): 575 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 453 us MUL_MAT_ID(ffn_moe_down-68): 254 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 505 us MUL_MAT_ID(ffn_moe_down-69): 211 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 436 us MUL_MAT_ID(ffn_moe_down-70): 267 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 579 us MUL_MAT_ID(ffn_moe_down-71): 222 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 464 us MUL_MAT_ID(ffn_moe_down-72): 237 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 524 us MUL_MAT_ID(ffn_moe_down-73): 208 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 453 us MUL_MAT_ID(ffn_moe_down-74): 302 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 417 us MUL_MAT_ID(ffn_moe_down-75): 196 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 422 us MUL_MAT_ID(ffn_moe_down-76): 198 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 424 us MUL_MAT_ID(ffn_moe_down-77): 208 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 424 us MUL_MAT_ID(ffn_moe_down-78): 250 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 447 us MUL_MAT_ID(ffn_moe_down-79): 249 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 391 us MUL_MAT_ID(ffn_moe_down-80): 217 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 476 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 392 us MUL_MAT_ID(ffn_moe_down-82): 204 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 468 us MUL_MAT_ID(ffn_moe_down-83): 285 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 466 us MUL_MAT_ID(ffn_moe_down-84): 347 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 467 us MUL_MAT_ID(ffn_moe_down-85): 334 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 396 us MUL_MAT_ID(ffn_moe_down-86): 185 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 396 us MUL_MAT_ID(ffn_moe_down-87): 177 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 495 us MUL_MAT_ID(ffn_moe_down-88): 181 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 403 us MUL_MAT_ID(ffn_moe_down-89): 220 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 375 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 382 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 34 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 584 us MUL_MAT_ID(ffn_moe_down-25): 203 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 396 us MUL_MAT_ID(ffn_moe_down-26): 289 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 459 us MUL_MAT_ID(ffn_moe_down-27): 274 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 417 us MUL_MAT_ID(ffn_moe_down-28): 340 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 481 us MUL_MAT_ID(ffn_moe_down-29): 311 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 445 us MUL_MAT_ID(ffn_moe_down-30): 298 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 455 us MUL_MAT_ID(ffn_moe_down-31): 322 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 384 us MUL_MAT_ID(ffn_moe_down-32): 201 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 388 us MUL_MAT_ID(ffn_moe_down-33): 196 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 438 us MUL_MAT_ID(ffn_moe_down-34): 253 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 454 us MUL_MAT_ID(ffn_moe_down-35): 374 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 441 us MUL_MAT_ID(ffn_moe_down-36): 264 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 472 us MUL_MAT_ID(ffn_moe_down-37): 214 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 386 us MUL_MAT_ID(ffn_moe_down-38): 240 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 464 us MUL_MAT_ID(ffn_moe_down-39): 247 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 389 us MUL_MAT_ID(ffn_moe_down-40): 328 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 439 us MUL_MAT_ID(ffn_moe_down-41): 199 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 377 us MUL_MAT_ID(ffn_moe_down-42): 245 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 469 us MUL_MAT_ID(ffn_moe_down-43): 267 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 395 us MUL_MAT_ID(ffn_moe_down-44): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 370 us MUL_MAT_ID(ffn_moe_down-45): 234 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 354 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 501 us MUL_MAT_ID(ffn_moe_down-47): 241 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 392 us MUL_MAT_ID(ffn_moe_down-48): 199 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 389 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 418 us MUL_MAT_ID(ffn_moe_down-50): 385 us ggml_barrier(...): 336 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 438 us MUL_MAT_ID(ffn_moe_down-51): 247 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 410 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 441 us MUL_MAT_ID(ffn_moe_down-53): 172 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 517 us MUL_MAT_ID(ffn_moe_down-54): 321 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 436 us MUL_MAT_ID(ffn_moe_down-55): 308 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 432 us MUL_MAT_ID(ffn_moe_down-56): 288 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 437 us MUL_MAT_ID(ffn_moe_down-57): 258 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 449 us MUL_MAT_ID(ffn_moe_down-58): 302 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 402 us MUL_MAT_ID(ffn_moe_down-59): 284 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 418 us MUL_MAT_ID(ffn_moe_down-60): 227 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 438 us MUL_MAT_ID(ffn_moe_down-61): 184 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 505 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 484 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 474 us MUL_MAT_ID(ffn_moe_down-64): 200 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 439 us MUL_MAT_ID(ffn_moe_down-65): 213 us ggml_barrier(...): 379 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 430 us MUL_MAT_ID(ffn_moe_down-66): 271 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 444 us MUL_MAT_ID(ffn_moe_down-67): 185 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 379 us MUL_MAT_ID(ffn_moe_down-68): 210 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 403 us MUL_MAT_ID(ffn_moe_down-69): 195 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 423 us MUL_MAT_ID(ffn_moe_down-70): 189 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 480 us MUL_MAT_ID(ffn_moe_down-71): 301 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 406 us MUL_MAT_ID(ffn_moe_down-72): 226 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 416 us MUL_MAT_ID(ffn_moe_down-73): 233 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 401 us MUL_MAT_ID(ffn_moe_down-74): 240 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 417 us MUL_MAT_ID(ffn_moe_down-75): 282 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 436 us MUL_MAT_ID(ffn_moe_down-76): 185 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 470 us MUL_MAT_ID(ffn_moe_down-77): 287 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 397 us MUL_MAT_ID(ffn_moe_down-78): 272 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 383 us MUL_MAT_ID(ffn_moe_down-79): 260 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 361 us MUL_MAT_ID(ffn_moe_down-80): 196 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 507 us MUL_MAT_ID(ffn_moe_down-81): 231 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 423 us MUL_MAT_ID(ffn_moe_down-82): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 436 us MUL_MAT_ID(ffn_moe_down-83): 213 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 377 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 476 us MUL_MAT_ID(ffn_moe_down-85): 244 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 396 us MUL_MAT_ID(ffn_moe_down-86): 177 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 399 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 397 us MUL_MAT_ID(ffn_moe_down-88): 238 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 384 us MUL_MAT_ID(ffn_moe_down-89): 398 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 527 us MUL_MAT_ID(ffn_moe_down-90): 254 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 364 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 48 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 503 us MUL_MAT_ID(ffn_moe_down-25): 283 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 450 us MUL_MAT_ID(ffn_moe_down-26): 633 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 433 us MUL_MAT_ID(ffn_moe_down-27): 284 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 455 us MUL_MAT_ID(ffn_moe_down-28): 178 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 369 us MUL_MAT_ID(ffn_moe_down-29): 178 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 397 us MUL_MAT_ID(ffn_moe_down-30): 197 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 503 us MUL_MAT_ID(ffn_moe_down-31): 203 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 395 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 406 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 425 us MUL_MAT_ID(ffn_moe_down-34): 185 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 400 us MUL_MAT_ID(ffn_moe_down-35): 257 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 447 us MUL_MAT_ID(ffn_moe_down-36): 251 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 457 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 505 us MUL_MAT_ID(ffn_moe_down-38): 311 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 449 us MUL_MAT_ID(ffn_moe_down-39): 251 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 477 us MUL_MAT_ID(ffn_moe_down-40): 256 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 396 us MUL_MAT_ID(ffn_moe_down-41): 681 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 371 us MUL_MAT_ID(ffn_moe_down-42): 203 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 417 us MUL_MAT_ID(ffn_moe_down-43): 181 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 472 us MUL_MAT_ID(ffn_moe_down-44): 269 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 459 us MUL_MAT_ID(ffn_moe_down-45): 465 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 526 us MUL_MAT_ID(ffn_moe_down-46): 216 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 465 us MUL_MAT_ID(ffn_moe_down-47): 283 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 365 us MUL_MAT_ID(ffn_moe_down-48): 194 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 461 us MUL_MAT_ID(ffn_moe_down-49): 278 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 380 us MUL_MAT_ID(ffn_moe_down-50): 238 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 404 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 391 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 539 us MUL_MAT_ID(ffn_moe_down-52): 209 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 359 us MUL_MAT_ID(ffn_moe_down-53): 250 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 463 us MUL_MAT_ID(ffn_moe_down-54): 173 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 421 us MUL_MAT_ID(ffn_moe_down-55): 321 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 449 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 424 us MUL_MAT_ID(ffn_moe_down-57): 300 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 501 us MUL_MAT_ID(ffn_moe_down-58): 218 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 426 us MUL_MAT_ID(ffn_moe_down-59): 176 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 408 us MUL_MAT_ID(ffn_moe_down-60): 176 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 422 us MUL_MAT_ID(ffn_moe_down-61): 435 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 557 us MUL_MAT_ID(ffn_moe_down-62): 210 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 462 us MUL_MAT_ID(ffn_moe_down-63): 210 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 441 us MUL_MAT_ID(ffn_moe_down-64): 205 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 439 us MUL_MAT_ID(ffn_moe_down-65): 330 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 410 us MUL_MAT_ID(ffn_moe_down-66): 274 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 459 us MUL_MAT_ID(ffn_moe_down-67): 206 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 471 us MUL_MAT_ID(ffn_moe_down-68): 307 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 477 us MUL_MAT_ID(ffn_moe_down-69): 310 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 412 us MUL_MAT_ID(ffn_moe_down-70): 272 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 450 us MUL_MAT_ID(ffn_moe_down-71): 278 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 424 us MUL_MAT_ID(ffn_moe_down-72): 257 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 424 us MUL_MAT_ID(ffn_moe_down-73): 264 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 462 us MUL_MAT_ID(ffn_moe_down-74): 314 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 497 us MUL_MAT_ID(ffn_moe_down-75): 316 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 452 us MUL_MAT_ID(ffn_moe_down-76): 206 us ggml_barrier(...): 337 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 460 us MUL_MAT_ID(ffn_moe_down-77): 225 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 509 us MUL_MAT_ID(ffn_moe_down-78): 243 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 408 us MUL_MAT_ID(ffn_moe_down-79): 207 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 436 us MUL_MAT_ID(ffn_moe_down-80): 190 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 431 us MUL_MAT_ID(ffn_moe_down-81): 296 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 400 us MUL_MAT_ID(ffn_moe_down-82): 251 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 465 us MUL_MAT_ID(ffn_moe_down-83): 191 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 400 us MUL_MAT_ID(ffn_moe_down-84): 275 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 486 us MUL_MAT_ID(ffn_moe_down-85): 212 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 490 us MUL_MAT_ID(ffn_moe_down-86): 262 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 524 us MUL_MAT_ID(ffn_moe_down-87): 189 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 388 us MUL_MAT_ID(ffn_moe_down-88): 596 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 412 us MUL_MAT_ID(ffn_moe_down-89): 314 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 531 us MUL_MAT_ID(ffn_moe_down-90): 336 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 456 us MUL_MAT_ID(ffn_moe_down-91): 219 us ggml_barrier(...): 325 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 418 us MUL_MAT_ID(ffn_moe_down-25): 278 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 441 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 339 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 478 us MUL_MAT_ID(ffn_moe_down-27): 228 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 487 us MUL_MAT_ID(ffn_moe_down-28): 285 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 383 us MUL_MAT_ID(ffn_moe_down-29): 224 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 412 us MUL_MAT_ID(ffn_moe_down-30): 219 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 452 us MUL_MAT_ID(ffn_moe_down-31): 319 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 411 us MUL_MAT_ID(ffn_moe_down-32): 262 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 368 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 397 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 414 us MUL_MAT_ID(ffn_moe_down-35): 207 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 478 us MUL_MAT_ID(ffn_moe_down-36): 293 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 454 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 471 us MUL_MAT_ID(ffn_moe_down-38): 275 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 415 us MUL_MAT_ID(ffn_moe_down-39): 246 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 485 us MUL_MAT_ID(ffn_moe_down-40): 286 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 497 us MUL_MAT_ID(ffn_moe_down-41): 229 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 463 us MUL_MAT_ID(ffn_moe_down-42): 194 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 522 us MUL_MAT_ID(ffn_moe_down-43): 217 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 423 us MUL_MAT_ID(ffn_moe_down-44): 259 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 442 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 411 us MUL_MAT_ID(ffn_moe_down-46): 243 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 373 us MUL_MAT_ID(ffn_moe_down-47): 243 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 429 us MUL_MAT_ID(ffn_moe_down-48): 214 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 379 us MUL_MAT_ID(ffn_moe_down-49): 407 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 456 us MUL_MAT_ID(ffn_moe_down-50): 187 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 449 us MUL_MAT_ID(ffn_moe_down-51): 641 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 372 us MUL_MAT_ID(ffn_moe_down-52): 256 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 364 us MUL_MAT_ID(ffn_moe_down-53): 191 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 488 us MUL_MAT_ID(ffn_moe_down-54): 197 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 505 us MUL_MAT_ID(ffn_moe_down-55): 215 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 456 us MUL_MAT_ID(ffn_moe_down-56): 354 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 568 us MUL_MAT_ID(ffn_moe_down-57): 206 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 431 us MUL_MAT_ID(ffn_moe_down-58): 219 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 459 us MUL_MAT_ID(ffn_moe_down-59): 236 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 414 us MUL_MAT_ID(ffn_moe_down-60): 219 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 447 us MUL_MAT_ID(ffn_moe_down-61): 202 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 448 us MUL_MAT_ID(ffn_moe_down-62): 270 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 405 us MUL_MAT_ID(ffn_moe_down-63): 233 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 552 us MUL_MAT_ID(ffn_moe_down-64): 206 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 395 us MUL_MAT_ID(ffn_moe_down-65): 219 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 419 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 481 us MUL_MAT_ID(ffn_moe_down-67): 224 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 442 us MUL_MAT_ID(ffn_moe_down-68): 219 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 528 us MUL_MAT_ID(ffn_moe_down-69): 239 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 449 us MUL_MAT_ID(ffn_moe_down-70): 492 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 497 us MUL_MAT_ID(ffn_moe_down-71): 214 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 381 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 495 us MUL_MAT_ID(ffn_moe_down-73): 220 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 405 us MUL_MAT_ID(ffn_moe_down-74): 312 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 384 us MUL_MAT_ID(ffn_moe_down-75): 317 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 386 us MUL_MAT_ID(ffn_moe_down-76): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 482 us MUL_MAT_ID(ffn_moe_down-77): 506 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 428 us MUL_MAT_ID(ffn_moe_down-78): 240 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 405 us MUL_MAT_ID(ffn_moe_down-79): 205 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 396 us MUL_MAT_ID(ffn_moe_down-80): 189 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 487 us MUL_MAT_ID(ffn_moe_down-81): 286 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 488 us MUL_MAT_ID(ffn_moe_down-82): 214 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 449 us MUL_MAT_ID(ffn_moe_down-83): 261 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 477 us MUL_MAT_ID(ffn_moe_down-84): 324 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 449 us MUL_MAT_ID(ffn_moe_down-85): 217 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 368 us MUL_MAT_ID(ffn_moe_down-86): 505 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 429 us MUL_MAT_ID(ffn_moe_down-87): 308 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 390 us MUL_MAT_ID(ffn_moe_down-88): 193 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 428 us MUL_MAT_ID(ffn_moe_down-89): 213 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 393 us MUL_MAT_ID(ffn_moe_down-90): 695 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 378 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 29 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 402 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 479 us MUL_MAT_ID(ffn_moe_down-26): 415 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 364 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 381 us MUL_MAT_ID(ffn_moe_down-28): 185 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 431 us MUL_MAT_ID(ffn_moe_down-29): 178 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 451 us MUL_MAT_ID(ffn_moe_down-30): 171 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 444 us MUL_MAT_ID(ffn_moe_down-31): 243 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 404 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 397 us MUL_MAT_ID(ffn_moe_down-33): 186 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 387 us MUL_MAT_ID(ffn_moe_down-34): 212 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 371 us MUL_MAT_ID(ffn_moe_down-35): 250 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 372 us MUL_MAT_ID(ffn_moe_down-36): 186 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 474 us MUL_MAT_ID(ffn_moe_down-37): 315 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 480 us MUL_MAT_ID(ffn_moe_down-38): 223 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 450 us MUL_MAT_ID(ffn_moe_down-39): 303 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 366 us MUL_MAT_ID(ffn_moe_down-40): 238 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 351 us MUL_MAT_ID(ffn_moe_down-41): 404 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 385 us MUL_MAT_ID(ffn_moe_down-42): 259 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 485 us MUL_MAT_ID(ffn_moe_down-43): 213 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 415 us MUL_MAT_ID(ffn_moe_down-44): 259 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 495 us MUL_MAT_ID(ffn_moe_down-45): 219 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 361 us MUL_MAT_ID(ffn_moe_down-46): 188 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 426 us MUL_MAT_ID(ffn_moe_down-47): 301 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 393 us MUL_MAT_ID(ffn_moe_down-48): 390 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 409 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 367 us MUL_MAT_ID(ffn_moe_down-50): 234 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 356 us MUL_MAT_ID(ffn_moe_down-51): 194 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 386 us MUL_MAT_ID(ffn_moe_down-52): 299 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 398 us MUL_MAT_ID(ffn_moe_down-53): 491 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 436 us MUL_MAT_ID(ffn_moe_down-54): 559 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 391 us MUL_MAT_ID(ffn_moe_down-55): 241 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 426 us MUL_MAT_ID(ffn_moe_down-56): 362 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 413 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 391 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 455 us MUL_MAT_ID(ffn_moe_down-59): 210 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 401 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 369 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 487 us MUL_MAT_ID(ffn_moe_down-62): 174 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 450 us MUL_MAT_ID(ffn_moe_down-63): 324 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 389 us MUL_MAT_ID(ffn_moe_down-64): 204 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 401 us MUL_MAT_ID(ffn_moe_down-65): 216 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 397 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 383 us MUL_MAT_ID(ffn_moe_down-67): 357 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 469 us MUL_MAT_ID(ffn_moe_down-68): 254 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 387 us MUL_MAT_ID(ffn_moe_down-69): 214 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 413 us MUL_MAT_ID(ffn_moe_down-70): 430 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 385 us MUL_MAT_ID(ffn_moe_down-71): 241 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 395 us MUL_MAT_ID(ffn_moe_down-72): 220 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 336 us MUL_MAT_ID(ffn_moe_down-73): 279 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 464 us MUL_MAT_ID(ffn_moe_down-74): 178 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 459 us MUL_MAT_ID(ffn_moe_down-75): 230 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 461 us MUL_MAT_ID(ffn_moe_down-76): 277 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 364 us MUL_MAT_ID(ffn_moe_down-77): 336 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 424 us MUL_MAT_ID(ffn_moe_down-78): 309 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 465 us MUL_MAT_ID(ffn_moe_down-79): 206 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 368 us MUL_MAT_ID(ffn_moe_down-80): 191 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 397 us MUL_MAT_ID(ffn_moe_down-81): 186 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 394 us MUL_MAT_ID(ffn_moe_down-82): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 478 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 377 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 406 us MUL_MAT_ID(ffn_moe_down-85): 237 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 427 us MUL_MAT_ID(ffn_moe_down-86): 407 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 388 us MUL_MAT_ID(ffn_moe_down-87): 217 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 403 us MUL_MAT_ID(ffn_moe_down-88): 216 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 387 us MUL_MAT_ID(ffn_moe_down-89): 185 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 402 us MUL_MAT_ID(ffn_moe_down-90): 222 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 406 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 27 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 461 us MUL_MAT_ID(ffn_moe_down-25): 293 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 437 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 459 us MUL_MAT_ID(ffn_moe_down-27): 215 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 416 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 405 us MUL_MAT_ID(ffn_moe_down-29): 241 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 393 us MUL_MAT_ID(ffn_moe_down-30): 203 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 437 us MUL_MAT_ID(ffn_moe_down-31): 248 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 393 us MUL_MAT_ID(ffn_moe_down-32): 206 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 427 us MUL_MAT_ID(ffn_moe_down-33): 213 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 379 us MUL_MAT_ID(ffn_moe_down-34): 205 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 503 us MUL_MAT_ID(ffn_moe_down-35): 224 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 351 us MUL_MAT_ID(ffn_moe_down-36): 227 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 498 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 481 us MUL_MAT_ID(ffn_moe_down-38): 191 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 400 us MUL_MAT_ID(ffn_moe_down-39): 267 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 399 us MUL_MAT_ID(ffn_moe_down-40): 633 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 368 us MUL_MAT_ID(ffn_moe_down-41): 269 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 421 us MUL_MAT_ID(ffn_moe_down-42): 171 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 438 us MUL_MAT_ID(ffn_moe_down-43): 242 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 421 us MUL_MAT_ID(ffn_moe_down-44): 236 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 478 us MUL_MAT_ID(ffn_moe_down-45): 320 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 403 us MUL_MAT_ID(ffn_moe_down-46): 202 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 354 us MUL_MAT_ID(ffn_moe_down-47): 228 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 387 us MUL_MAT_ID(ffn_moe_down-48): 252 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 437 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 371 us MUL_MAT_ID(ffn_moe_down-50): 363 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 399 us MUL_MAT_ID(ffn_moe_down-51): 301 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 410 us MUL_MAT_ID(ffn_moe_down-52): 223 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 399 us MUL_MAT_ID(ffn_moe_down-53): 208 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 423 us MUL_MAT_ID(ffn_moe_down-54): 311 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 405 us MUL_MAT_ID(ffn_moe_down-55): 235 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 435 us MUL_MAT_ID(ffn_moe_down-56): 283 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 472 us MUL_MAT_ID(ffn_moe_down-57): 208 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 384 us MUL_MAT_ID(ffn_moe_down-58): 378 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 447 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 396 us MUL_MAT_ID(ffn_moe_down-60): 338 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 427 us MUL_MAT_ID(ffn_moe_down-61): 224 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 452 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 469 us MUL_MAT_ID(ffn_moe_down-63): 214 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 502 us MUL_MAT_ID(ffn_moe_down-64): 210 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 450 us MUL_MAT_ID(ffn_moe_down-65): 206 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 459 us MUL_MAT_ID(ffn_moe_down-66): 363 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 442 us MUL_MAT_ID(ffn_moe_down-67): 218 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 410 us MUL_MAT_ID(ffn_moe_down-68): 301 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 506 us MUL_MAT_ID(ffn_moe_down-69): 234 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 476 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 433 us MUL_MAT_ID(ffn_moe_down-71): 299 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 437 us MUL_MAT_ID(ffn_moe_down-72): 269 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 481 us MUL_MAT_ID(ffn_moe_down-73): 239 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 507 us MUL_MAT_ID(ffn_moe_down-74): 207 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 494 us MUL_MAT_ID(ffn_moe_down-75): 243 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 410 us MUL_MAT_ID(ffn_moe_down-76): 286 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 391 us MUL_MAT_ID(ffn_moe_down-77): 236 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 473 us MUL_MAT_ID(ffn_moe_down-78): 213 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 423 us MUL_MAT_ID(ffn_moe_down-79): 315 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 411 us MUL_MAT_ID(ffn_moe_down-80): 229 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 417 us MUL_MAT_ID(ffn_moe_down-81): 186 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 459 us MUL_MAT_ID(ffn_moe_down-82): 265 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 442 us MUL_MAT_ID(ffn_moe_down-83): 209 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 470 us MUL_MAT_ID(ffn_moe_down-84): 266 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 488 us MUL_MAT_ID(ffn_moe_down-85): 263 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 383 us MUL_MAT_ID(ffn_moe_down-86): 244 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 426 us MUL_MAT_ID(ffn_moe_down-87): 198 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 441 us MUL_MAT_ID(ffn_moe_down-88): 248 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 442 us MUL_MAT_ID(ffn_moe_down-89): 270 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 403 us MUL_MAT_ID(ffn_moe_down-90): 244 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 468 us MUL_MAT_ID(ffn_moe_down-91): 299 us ggml_barrier(...): 260 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 382 us MUL_MAT_ID(ffn_moe_down-25): 297 us ggml_barrier(...): 506 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 486 us MUL_MAT_ID(ffn_moe_down-26): 216 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 424 us MUL_MAT_ID(ffn_moe_down-27): 219 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 427 us MUL_MAT_ID(ffn_moe_down-28): 200 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 378 us MUL_MAT_ID(ffn_moe_down-29): 190 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 463 us MUL_MAT_ID(ffn_moe_down-30): 265 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 466 us MUL_MAT_ID(ffn_moe_down-31): 220 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 394 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 422 us MUL_MAT_ID(ffn_moe_down-33): 244 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 427 us MUL_MAT_ID(ffn_moe_down-34): 248 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 401 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 401 us MUL_MAT_ID(ffn_moe_down-36): 377 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 448 us MUL_MAT_ID(ffn_moe_down-37): 294 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 449 us MUL_MAT_ID(ffn_moe_down-38): 176 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 434 us MUL_MAT_ID(ffn_moe_down-39): 662 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 377 us MUL_MAT_ID(ffn_moe_down-40): 248 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 436 us MUL_MAT_ID(ffn_moe_down-41): 267 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 437 us MUL_MAT_ID(ffn_moe_down-42): 216 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 397 us MUL_MAT_ID(ffn_moe_down-43): 350 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 480 us MUL_MAT_ID(ffn_moe_down-44): 193 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 419 us MUL_MAT_ID(ffn_moe_down-45): 189 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 395 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 425 us MUL_MAT_ID(ffn_moe_down-47): 226 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 422 us MUL_MAT_ID(ffn_moe_down-48): 247 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 409 us MUL_MAT_ID(ffn_moe_down-49): 292 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 490 us MUL_MAT_ID(ffn_moe_down-50): 219 us ggml_barrier(...): 374 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 432 us MUL_MAT_ID(ffn_moe_down-51): 336 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 415 us MUL_MAT_ID(ffn_moe_down-52): 245 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 409 us MUL_MAT_ID(ffn_moe_down-53): 251 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 449 us MUL_MAT_ID(ffn_moe_down-54): 284 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 495 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 509 us MUL_MAT_ID(ffn_moe_down-56): 246 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 459 us MUL_MAT_ID(ffn_moe_down-57): 179 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 488 us MUL_MAT_ID(ffn_moe_down-58): 235 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 406 us MUL_MAT_ID(ffn_moe_down-59): 246 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 439 us MUL_MAT_ID(ffn_moe_down-60): 241 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 446 us MUL_MAT_ID(ffn_moe_down-61): 359 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 355 us MUL_MAT_ID(ffn_moe_down-62): 262 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 468 us MUL_MAT_ID(ffn_moe_down-63): 205 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 401 us MUL_MAT_ID(ffn_moe_down-64): 220 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 440 us MUL_MAT_ID(ffn_moe_down-65): 221 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 410 us MUL_MAT_ID(ffn_moe_down-66): 229 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 433 us MUL_MAT_ID(ffn_moe_down-67): 252 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 442 us MUL_MAT_ID(ffn_moe_down-68): 323 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 418 us MUL_MAT_ID(ffn_moe_down-69): 206 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 413 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 497 us MUL_MAT_ID(ffn_moe_down-71): 304 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 468 us MUL_MAT_ID(ffn_moe_down-72): 278 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 438 us MUL_MAT_ID(ffn_moe_down-73): 225 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 429 us MUL_MAT_ID(ffn_moe_down-74): 181 us ggml_barrier(...): 337 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 529 us MUL_MAT_ID(ffn_moe_down-75): 213 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 386 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 410 us MUL_MAT_ID(ffn_moe_down-77): 327 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 418 us MUL_MAT_ID(ffn_moe_down-78): 381 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 432 us MUL_MAT_ID(ffn_moe_down-79): 431 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 520 us MUL_MAT_ID(ffn_moe_down-80): 277 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 527 us MUL_MAT_ID(ffn_moe_down-81): 231 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 443 us MUL_MAT_ID(ffn_moe_down-82): 300 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 523 us MUL_MAT_ID(ffn_moe_down-83): 218 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 474 us MUL_MAT_ID(ffn_moe_down-84): 240 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 396 us MUL_MAT_ID(ffn_moe_down-85): 535 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 465 us MUL_MAT_ID(ffn_moe_down-86): 251 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 440 us MUL_MAT_ID(ffn_moe_down-87): 343 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 420 us MUL_MAT_ID(ffn_moe_down-88): 553 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 429 us MUL_MAT_ID(ffn_moe_down-89): 219 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 399 us MUL_MAT_ID(ffn_moe_down-90): 255 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 481 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 174 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 405 us MUL_MAT_ID(ffn_moe_down-25): 182 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 377 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 370 us MUL_MAT_ID(ffn_moe_down-27): 210 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 472 us MUL_MAT_ID(ffn_moe_down-28): 183 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 364 us MUL_MAT_ID(ffn_moe_down-29): 210 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 376 us MUL_MAT_ID(ffn_moe_down-30): 189 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 371 us MUL_MAT_ID(ffn_moe_down-31): 201 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 439 us MUL_MAT_ID(ffn_moe_down-32): 271 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 404 us MUL_MAT_ID(ffn_moe_down-33): 217 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 432 us MUL_MAT_ID(ffn_moe_down-34): 184 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 438 us MUL_MAT_ID(ffn_moe_down-35): 277 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 392 us MUL_MAT_ID(ffn_moe_down-36): 247 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 412 us MUL_MAT_ID(ffn_moe_down-37): 244 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 379 us MUL_MAT_ID(ffn_moe_down-38): 208 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 358 us MUL_MAT_ID(ffn_moe_down-39): 231 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 367 us MUL_MAT_ID(ffn_moe_down-40): 197 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 405 us MUL_MAT_ID(ffn_moe_down-41): 387 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 404 us MUL_MAT_ID(ffn_moe_down-42): 188 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 362 us MUL_MAT_ID(ffn_moe_down-43): 179 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 374 us MUL_MAT_ID(ffn_moe_down-44): 248 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 406 us MUL_MAT_ID(ffn_moe_down-45): 275 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 518 us MUL_MAT_ID(ffn_moe_down-46): 210 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 364 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 386 us MUL_MAT_ID(ffn_moe_down-48): 240 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 450 us MUL_MAT_ID(ffn_moe_down-49): 372 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 408 us MUL_MAT_ID(ffn_moe_down-50): 246 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 381 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 401 us MUL_MAT_ID(ffn_moe_down-52): 260 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 357 us MUL_MAT_ID(ffn_moe_down-53): 204 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 464 us MUL_MAT_ID(ffn_moe_down-54): 408 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 401 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 377 us MUL_MAT_ID(ffn_moe_down-56): 283 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 393 us MUL_MAT_ID(ffn_moe_down-57): 178 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 410 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 400 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 354 us MUL_MAT_ID(ffn_moe_down-60): 180 us ggml_barrier(...): 504 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 462 us MUL_MAT_ID(ffn_moe_down-61): 194 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 398 us MUL_MAT_ID(ffn_moe_down-62): 414 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 395 us MUL_MAT_ID(ffn_moe_down-63): 245 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 391 us MUL_MAT_ID(ffn_moe_down-64): 524 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 392 us MUL_MAT_ID(ffn_moe_down-65): 290 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 373 us MUL_MAT_ID(ffn_moe_down-66): 413 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 401 us MUL_MAT_ID(ffn_moe_down-67): 221 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 367 us MUL_MAT_ID(ffn_moe_down-68): 387 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 398 us MUL_MAT_ID(ffn_moe_down-69): 278 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 388 us MUL_MAT_ID(ffn_moe_down-70): 218 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 383 us MUL_MAT_ID(ffn_moe_down-71): 206 us ggml_barrier(...): 397 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 509 us MUL_MAT_ID(ffn_moe_down-72): 214 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 414 us MUL_MAT_ID(ffn_moe_down-73): 200 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 439 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 392 us MUL_MAT_ID(ffn_moe_down-75): 369 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 395 us MUL_MAT_ID(ffn_moe_down-76): 196 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 397 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 454 us MUL_MAT_ID(ffn_moe_down-78): 258 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 407 us MUL_MAT_ID(ffn_moe_down-79): 187 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 441 us MUL_MAT_ID(ffn_moe_down-80): 249 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 437 us MUL_MAT_ID(ffn_moe_down-81): 271 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 368 us MUL_MAT_ID(ffn_moe_down-82): 203 us ggml_barrier(...): 414 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 399 us MUL_MAT_ID(ffn_moe_down-83): 211 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 430 us MUL_MAT_ID(ffn_moe_down-84): 314 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 435 us MUL_MAT_ID(ffn_moe_down-85): 346 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 407 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 399 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 504 us MUL_MAT_ID(ffn_moe_down-88): 475 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 462 us MUL_MAT_ID(ffn_moe_down-89): 223 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 435 us MUL_MAT_ID(ffn_moe_down-90): 267 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 423 us MUL_MAT_ID(ffn_moe_down-91): 283 us ggml_barrier(...): 228 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 511 us MUL_MAT_ID(ffn_moe_down-25): 388 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 453 us MUL_MAT_ID(ffn_moe_down-26): 263 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 410 us MUL_MAT_ID(ffn_moe_down-27): 237 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 501 us MUL_MAT_ID(ffn_moe_down-28): 218 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 407 us MUL_MAT_ID(ffn_moe_down-29): 264 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 509 us MUL_MAT_ID(ffn_moe_down-30): 325 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 363 us MUL_MAT_ID(ffn_moe_down-31): 216 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 383 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 507 us MUL_MAT_ID(ffn_moe_down-33): 219 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 387 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 526 us MUL_MAT_ID(ffn_moe_down-35): 225 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 410 us MUL_MAT_ID(ffn_moe_down-36): 222 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 379 us MUL_MAT_ID(ffn_moe_down-37): 222 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 392 us MUL_MAT_ID(ffn_moe_down-38): 179 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 386 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 422 us MUL_MAT_ID(ffn_moe_down-40): 374 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 436 us MUL_MAT_ID(ffn_moe_down-41): 253 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 487 us MUL_MAT_ID(ffn_moe_down-42): 193 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 387 us MUL_MAT_ID(ffn_moe_down-43): 192 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 376 us MUL_MAT_ID(ffn_moe_down-44): 220 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 407 us MUL_MAT_ID(ffn_moe_down-45): 268 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 381 us MUL_MAT_ID(ffn_moe_down-46): 342 us ggml_barrier(...): 337 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 382 us MUL_MAT_ID(ffn_moe_down-47): 513 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 377 us MUL_MAT_ID(ffn_moe_down-48): 230 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 514 us MUL_MAT_ID(ffn_moe_down-49): 241 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 487 us MUL_MAT_ID(ffn_moe_down-50): 213 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 478 us MUL_MAT_ID(ffn_moe_down-51): 226 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 399 us MUL_MAT_ID(ffn_moe_down-52): 202 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 383 us MUL_MAT_ID(ffn_moe_down-53): 218 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 431 us MUL_MAT_ID(ffn_moe_down-54): 282 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 364 us MUL_MAT_ID(ffn_moe_down-55): 189 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 409 us MUL_MAT_ID(ffn_moe_down-56): 472 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 438 us MUL_MAT_ID(ffn_moe_down-57): 324 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 422 us MUL_MAT_ID(ffn_moe_down-58): 294 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 398 us MUL_MAT_ID(ffn_moe_down-59): 219 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 514 us MUL_MAT_ID(ffn_moe_down-60): 209 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 408 us MUL_MAT_ID(ffn_moe_down-61): 179 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 469 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 346 us MUL_MAT_ID(ffn_moe_down-63): 208 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 404 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 386 us MUL_MAT_ID(ffn_moe_down-65): 198 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 440 us MUL_MAT_ID(ffn_moe_down-66): 213 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 462 us MUL_MAT_ID(ffn_moe_down-67): 186 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 382 us MUL_MAT_ID(ffn_moe_down-68): 547 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 460 us MUL_MAT_ID(ffn_moe_down-69): 268 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 453 us MUL_MAT_ID(ffn_moe_down-70): 279 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 418 us MUL_MAT_ID(ffn_moe_down-71): 251 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 400 us MUL_MAT_ID(ffn_moe_down-72): 239 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 422 us MUL_MAT_ID(ffn_moe_down-73): 227 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 495 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 396 us MUL_MAT_ID(ffn_moe_down-75): 173 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 381 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 370 us MUL_MAT_ID(ffn_moe_down-77): 232 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 387 us MUL_MAT_ID(ffn_moe_down-78): 269 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 487 us MUL_MAT_ID(ffn_moe_down-79): 222 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 351 us MUL_MAT_ID(ffn_moe_down-80): 204 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 371 us MUL_MAT_ID(ffn_moe_down-81): 217 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 351 us MUL_MAT_ID(ffn_moe_down-82): 211 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 403 us MUL_MAT_ID(ffn_moe_down-83): 256 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 440 us MUL_MAT_ID(ffn_moe_down-84): 187 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 392 us MUL_MAT_ID(ffn_moe_down-85): 242 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 380 us MUL_MAT_ID(ffn_moe_down-86): 295 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 375 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 485 us MUL_MAT_ID(ffn_moe_down-88): 276 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 421 us MUL_MAT_ID(ffn_moe_down-89): 166 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 345 us MUL_MAT_ID(ffn_moe_down-90): 214 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 482 us MUL_MAT_ID(ffn_moe_down-91): 250 us ggml_barrier(...): 168 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 571 us MUL_MAT_ID(ffn_moe_down-25): 400 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 436 us MUL_MAT_ID(ffn_moe_down-26): 229 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 481 us MUL_MAT_ID(ffn_moe_down-27): 310 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 455 us MUL_MAT_ID(ffn_moe_down-28): 320 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 438 us MUL_MAT_ID(ffn_moe_down-29): 209 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 480 us MUL_MAT_ID(ffn_moe_down-30): 269 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 441 us MUL_MAT_ID(ffn_moe_down-31): 171 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 387 us MUL_MAT_ID(ffn_moe_down-32): 199 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 453 us MUL_MAT_ID(ffn_moe_down-33): 265 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 403 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 377 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 393 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 392 us MUL_MAT_ID(ffn_moe_down-37): 367 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 360 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 406 us MUL_MAT_ID(ffn_moe_down-39): 731 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 372 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 417 us MUL_MAT_ID(ffn_moe_down-41): 216 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 407 us MUL_MAT_ID(ffn_moe_down-42): 202 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 394 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 377 us MUL_MAT_ID(ffn_moe_down-44): 223 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 387 us MUL_MAT_ID(ffn_moe_down-45): 266 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 460 us MUL_MAT_ID(ffn_moe_down-46): 223 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 500 us MUL_MAT_ID(ffn_moe_down-47): 269 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 441 us MUL_MAT_ID(ffn_moe_down-48): 182 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 379 us MUL_MAT_ID(ffn_moe_down-49): 181 us ggml_barrier(...): 357 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 397 us MUL_MAT_ID(ffn_moe_down-50): 208 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 380 us MUL_MAT_ID(ffn_moe_down-51): 212 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 445 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 490 us MUL_MAT_ID(ffn_moe_down-53): 177 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 474 us MUL_MAT_ID(ffn_moe_down-54): 197 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 413 us MUL_MAT_ID(ffn_moe_down-55): 219 us ggml_barrier(...): 341 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 400 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 377 us MUL_MAT_ID(ffn_moe_down-57): 366 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 406 us MUL_MAT_ID(ffn_moe_down-58): 204 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 468 us MUL_MAT_ID(ffn_moe_down-59): 239 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 411 us MUL_MAT_ID(ffn_moe_down-60): 213 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 409 us MUL_MAT_ID(ffn_moe_down-61): 450 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 377 us MUL_MAT_ID(ffn_moe_down-62): 247 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 406 us MUL_MAT_ID(ffn_moe_down-63): 460 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 438 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 345 us MUL_MAT_ID(ffn_moe_down-65): 202 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 374 us MUL_MAT_ID(ffn_moe_down-66): 199 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 486 us MUL_MAT_ID(ffn_moe_down-67): 227 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 472 us MUL_MAT_ID(ffn_moe_down-68): 181 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 416 us MUL_MAT_ID(ffn_moe_down-69): 199 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 497 us MUL_MAT_ID(ffn_moe_down-70): 213 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 367 us MUL_MAT_ID(ffn_moe_down-71): 250 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 371 us MUL_MAT_ID(ffn_moe_down-72): 220 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 395 us MUL_MAT_ID(ffn_moe_down-73): 255 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 398 us MUL_MAT_ID(ffn_moe_down-74): 284 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 482 us MUL_MAT_ID(ffn_moe_down-75): 203 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 430 us MUL_MAT_ID(ffn_moe_down-76): 196 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 416 us MUL_MAT_ID(ffn_moe_down-77): 249 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 444 us MUL_MAT_ID(ffn_moe_down-78): 293 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 386 us MUL_MAT_ID(ffn_moe_down-79): 227 us ggml_barrier(...): 363 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 427 us MUL_MAT_ID(ffn_moe_down-80): 374 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 395 us MUL_MAT_ID(ffn_moe_down-81): 222 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 519 us MUL_MAT_ID(ffn_moe_down-82): 196 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 408 us MUL_MAT_ID(ffn_moe_down-83): 191 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 446 us MUL_MAT_ID(ffn_moe_down-84): 289 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 366 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 391 us MUL_MAT_ID(ffn_moe_down-86): 387 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 177 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 389 us MUL_MAT_ID(ffn_moe_down-88): 213 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 412 us MUL_MAT_ID(ffn_moe_down-89): 227 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 394 us MUL_MAT_ID(ffn_moe_down-90): 218 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 414 us MUL_MAT_ID(ffn_moe_down-91): 286 us ggml_barrier(...): 212 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 633 us MUL_MAT_ID(ffn_moe_down-25): 230 us ggml_barrier(...): 390 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 486 us MUL_MAT_ID(ffn_moe_down-26): 209 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 502 us MUL_MAT_ID(ffn_moe_down-27): 243 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 416 us MUL_MAT_ID(ffn_moe_down-28): 261 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 512 us MUL_MAT_ID(ffn_moe_down-29): 298 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 485 us MUL_MAT_ID(ffn_moe_down-30): 190 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 483 us MUL_MAT_ID(ffn_moe_down-31): 295 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 451 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 458 us MUL_MAT_ID(ffn_moe_down-33): 212 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 444 us MUL_MAT_ID(ffn_moe_down-34): 288 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 514 us MUL_MAT_ID(ffn_moe_down-35): 220 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 438 us MUL_MAT_ID(ffn_moe_down-36): 257 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 372 us MUL_MAT_ID(ffn_moe_down-37): 255 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 471 us MUL_MAT_ID(ffn_moe_down-38): 231 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 471 us MUL_MAT_ID(ffn_moe_down-39): 243 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 372 us MUL_MAT_ID(ffn_moe_down-40): 303 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 490 us MUL_MAT_ID(ffn_moe_down-41): 544 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 490 us MUL_MAT_ID(ffn_moe_down-42): 218 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 476 us MUL_MAT_ID(ffn_moe_down-43): 220 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 360 us MUL_MAT_ID(ffn_moe_down-44): 198 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 492 us MUL_MAT_ID(ffn_moe_down-45): 375 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 409 us MUL_MAT_ID(ffn_moe_down-46): 240 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 409 us MUL_MAT_ID(ffn_moe_down-47): 232 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 363 us MUL_MAT_ID(ffn_moe_down-48): 207 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 404 us MUL_MAT_ID(ffn_moe_down-49): 202 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 386 us MUL_MAT_ID(ffn_moe_down-50): 209 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 365 us MUL_MAT_ID(ffn_moe_down-51): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 395 us MUL_MAT_ID(ffn_moe_down-52): 181 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 384 us MUL_MAT_ID(ffn_moe_down-53): 198 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 368 us MUL_MAT_ID(ffn_moe_down-54): 205 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 386 us MUL_MAT_ID(ffn_moe_down-55): 206 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 375 us MUL_MAT_ID(ffn_moe_down-56): 247 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 415 us MUL_MAT_ID(ffn_moe_down-57): 273 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 465 us MUL_MAT_ID(ffn_moe_down-58): 244 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 377 us MUL_MAT_ID(ffn_moe_down-59): 240 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 378 us MUL_MAT_ID(ffn_moe_down-60): 238 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 514 us MUL_MAT_ID(ffn_moe_down-61): 183 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 384 us MUL_MAT_ID(ffn_moe_down-62): 232 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 376 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 407 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 442 us MUL_MAT_ID(ffn_moe_down-65): 172 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 382 us MUL_MAT_ID(ffn_moe_down-66): 241 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 375 us MUL_MAT_ID(ffn_moe_down-67): 206 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 376 us MUL_MAT_ID(ffn_moe_down-68): 259 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 386 us MUL_MAT_ID(ffn_moe_down-69): 212 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 400 us MUL_MAT_ID(ffn_moe_down-70): 241 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 495 us MUL_MAT_ID(ffn_moe_down-71): 206 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 406 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 397 us MUL_MAT_ID(ffn_moe_down-73): 512 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 369 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 391 us MUL_MAT_ID(ffn_moe_down-75): 225 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 376 us MUL_MAT_ID(ffn_moe_down-76): 184 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 398 us MUL_MAT_ID(ffn_moe_down-77): 234 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 411 us MUL_MAT_ID(ffn_moe_down-78): 205 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 375 us MUL_MAT_ID(ffn_moe_down-79): 221 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 388 us MUL_MAT_ID(ffn_moe_down-80): 306 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 451 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 415 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 377 us MUL_MAT_ID(ffn_moe_down-83): 240 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 400 us MUL_MAT_ID(ffn_moe_down-84): 305 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 390 us MUL_MAT_ID(ffn_moe_down-85): 227 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 369 us MUL_MAT_ID(ffn_moe_down-86): 210 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 405 us MUL_MAT_ID(ffn_moe_down-87): 184 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 392 us MUL_MAT_ID(ffn_moe_down-88): 531 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 405 us MUL_MAT_ID(ffn_moe_down-89): 231 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 364 us MUL_MAT_ID(ffn_moe_down-90): 227 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 461 us MUL_MAT_ID(ffn_moe_down-91): 215 us ggml_barrier(...): 215 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 569 us MUL_MAT_ID(ffn_moe_down-25): 290 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 464 us MUL_MAT_ID(ffn_moe_down-26): 313 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 415 us MUL_MAT_ID(ffn_moe_down-27): 216 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 373 us MUL_MAT_ID(ffn_moe_down-28): 312 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 441 us MUL_MAT_ID(ffn_moe_down-29): 203 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 389 us MUL_MAT_ID(ffn_moe_down-30): 551 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 529 us MUL_MAT_ID(ffn_moe_down-31): 279 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 367 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 475 us MUL_MAT_ID(ffn_moe_down-33): 183 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 367 us MUL_MAT_ID(ffn_moe_down-34): 202 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 398 us MUL_MAT_ID(ffn_moe_down-35): 314 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 392 us MUL_MAT_ID(ffn_moe_down-36): 369 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 406 us MUL_MAT_ID(ffn_moe_down-37): 333 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 440 us MUL_MAT_ID(ffn_moe_down-38): 170 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 413 us MUL_MAT_ID(ffn_moe_down-39): 242 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 479 us MUL_MAT_ID(ffn_moe_down-40): 241 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 429 us MUL_MAT_ID(ffn_moe_down-41): 210 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 428 us MUL_MAT_ID(ffn_moe_down-42): 259 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 454 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 439 us MUL_MAT_ID(ffn_moe_down-44): 176 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 384 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 400 us MUL_MAT_ID(ffn_moe_down-46): 205 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 413 us MUL_MAT_ID(ffn_moe_down-47): 253 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 390 us MUL_MAT_ID(ffn_moe_down-48): 216 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 389 us MUL_MAT_ID(ffn_moe_down-49): 294 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 376 us MUL_MAT_ID(ffn_moe_down-51): 208 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 371 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 350 us MUL_MAT_ID(ffn_moe_down-53): 354 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 528 us MUL_MAT_ID(ffn_moe_down-54): 217 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 419 us MUL_MAT_ID(ffn_moe_down-55): 257 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 412 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 358 us MUL_MAT_ID(ffn_moe_down-57): 181 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 427 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 372 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 392 us MUL_MAT_ID(ffn_moe_down-60): 184 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 441 us MUL_MAT_ID(ffn_moe_down-61): 205 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 414 us MUL_MAT_ID(ffn_moe_down-62): 205 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 428 us MUL_MAT_ID(ffn_moe_down-63): 365 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 376 us MUL_MAT_ID(ffn_moe_down-64): 221 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 389 us MUL_MAT_ID(ffn_moe_down-65): 203 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 278 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 410 us MUL_MAT_ID(ffn_moe_down-67): 560 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 405 us MUL_MAT_ID(ffn_moe_down-68): 204 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 441 us MUL_MAT_ID(ffn_moe_down-69): 243 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 405 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 496 us MUL_MAT_ID(ffn_moe_down-71): 208 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 466 us MUL_MAT_ID(ffn_moe_down-72): 212 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 419 us MUL_MAT_ID(ffn_moe_down-73): 197 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 382 us MUL_MAT_ID(ffn_moe_down-74): 225 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 395 us MUL_MAT_ID(ffn_moe_down-75): 191 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 367 us MUL_MAT_ID(ffn_moe_down-76): 184 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 375 us MUL_MAT_ID(ffn_moe_down-77): 188 us ggml_barrier(...): 345 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 377 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 402 us MUL_MAT_ID(ffn_moe_down-79): 188 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 389 us MUL_MAT_ID(ffn_moe_down-80): 201 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 382 us MUL_MAT_ID(ffn_moe_down-81): 523 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 398 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 398 us MUL_MAT_ID(ffn_moe_down-83): 242 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 379 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 356 us MUL_MAT_ID(ffn_moe_down-85): 223 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 385 us MUL_MAT_ID(ffn_moe_down-86): 225 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 472 us MUL_MAT_ID(ffn_moe_down-87): 189 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 383 us MUL_MAT_ID(ffn_moe_down-88): 258 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 344 us MUL_MAT_ID(ffn_moe_down-89): 421 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 414 us MUL_MAT_ID(ffn_moe_down-90): 230 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 452 us MUL_MAT_ID(ffn_moe_down-91): 290 us ggml_barrier(...): 134 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 379 us MUL_MAT_ID(ffn_moe_down-25): 184 us ggml_barrier(...): 539 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 411 us MUL_MAT_ID(ffn_moe_down-26): 202 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 399 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 419 us MUL_MAT_ID(ffn_moe_down-28): 210 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 424 us MUL_MAT_ID(ffn_moe_down-29): 187 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 408 us MUL_MAT_ID(ffn_moe_down-30): 221 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 392 us MUL_MAT_ID(ffn_moe_down-31): 213 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 439 us MUL_MAT_ID(ffn_moe_down-32): 261 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 395 us MUL_MAT_ID(ffn_moe_down-33): 223 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 411 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 396 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 399 us MUL_MAT_ID(ffn_moe_down-36): 225 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 396 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 378 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 489 us MUL_MAT_ID(ffn_moe_down-39): 269 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 364 us MUL_MAT_ID(ffn_moe_down-40): 206 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 478 us MUL_MAT_ID(ffn_moe_down-41): 197 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 404 us MUL_MAT_ID(ffn_moe_down-42): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 399 us MUL_MAT_ID(ffn_moe_down-43): 226 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 388 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 393 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 379 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 379 us MUL_MAT_ID(ffn_moe_down-47): 177 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 388 us MUL_MAT_ID(ffn_moe_down-48): 213 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 424 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 438 us MUL_MAT_ID(ffn_moe_down-50): 211 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 386 us MUL_MAT_ID(ffn_moe_down-51): 239 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 386 us MUL_MAT_ID(ffn_moe_down-52): 608 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 393 us MUL_MAT_ID(ffn_moe_down-53): 226 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 420 us MUL_MAT_ID(ffn_moe_down-54): 207 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 520 us MUL_MAT_ID(ffn_moe_down-55): 225 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 397 us MUL_MAT_ID(ffn_moe_down-56): 257 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 464 us MUL_MAT_ID(ffn_moe_down-57): 187 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 351 us MUL_MAT_ID(ffn_moe_down-58): 186 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 442 us MUL_MAT_ID(ffn_moe_down-59): 263 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 236 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 437 us MUL_MAT_ID(ffn_moe_down-61): 199 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 381 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 512 us MUL_MAT_ID(ffn_moe_down-63): 217 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 416 us MUL_MAT_ID(ffn_moe_down-64): 246 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 494 us MUL_MAT_ID(ffn_moe_down-65): 205 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 405 us MUL_MAT_ID(ffn_moe_down-66): 274 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 474 us MUL_MAT_ID(ffn_moe_down-67): 307 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 397 us MUL_MAT_ID(ffn_moe_down-68): 165 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 440 us MUL_MAT_ID(ffn_moe_down-69): 312 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 465 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 406 us MUL_MAT_ID(ffn_moe_down-71): 240 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 519 us MUL_MAT_ID(ffn_moe_down-72): 212 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 417 us MUL_MAT_ID(ffn_moe_down-73): 224 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 401 us MUL_MAT_ID(ffn_moe_down-74): 154 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 406 us MUL_MAT_ID(ffn_moe_down-75): 251 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 426 us MUL_MAT_ID(ffn_moe_down-76): 198 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 612 us MUL_MAT_ID(ffn_moe_down-77): 310 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 392 us MUL_MAT_ID(ffn_moe_down-78): 232 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 361 us MUL_MAT_ID(ffn_moe_down-79): 203 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 377 us MUL_MAT_ID(ffn_moe_down-80): 194 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 400 us MUL_MAT_ID(ffn_moe_down-81): 216 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 460 us MUL_MAT_ID(ffn_moe_down-82): 276 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 520 us MUL_MAT_ID(ffn_moe_down-83): 568 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 397 us MUL_MAT_ID(ffn_moe_down-84): 348 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 474 us MUL_MAT_ID(ffn_moe_down-85): 412 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 474 us MUL_MAT_ID(ffn_moe_down-86): 214 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 441 us MUL_MAT_ID(ffn_moe_down-87): 369 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 456 us MUL_MAT_ID(ffn_moe_down-88): 239 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 514 us MUL_MAT_ID(ffn_moe_down-89): 259 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 433 us MUL_MAT_ID(ffn_moe_down-90): 308 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 416 us MUL_MAT_ID(ffn_moe_down-91): 638 us ggml_barrier(...): 235 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 398 us MUL_MAT_ID(ffn_moe_down-25): 202 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 431 us MUL_MAT_ID(ffn_moe_down-26): 200 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 413 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 582 us MUL_MAT_ID(ffn_moe_down-28): 427 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 351 us MUL_MAT_ID(ffn_moe_down-29): 373 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 394 us MUL_MAT_ID(ffn_moe_down-30): 204 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 457 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 398 us MUL_MAT_ID(ffn_moe_down-32): 325 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 462 us MUL_MAT_ID(ffn_moe_down-33): 210 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 477 us MUL_MAT_ID(ffn_moe_down-34): 231 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 392 us MUL_MAT_ID(ffn_moe_down-35): 263 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 378 us MUL_MAT_ID(ffn_moe_down-36): 188 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 458 us MUL_MAT_ID(ffn_moe_down-37): 258 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 364 us MUL_MAT_ID(ffn_moe_down-38): 208 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 444 us MUL_MAT_ID(ffn_moe_down-39): 218 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 381 us MUL_MAT_ID(ffn_moe_down-40): 424 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 372 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 378 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 458 us MUL_MAT_ID(ffn_moe_down-43): 197 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 525 us MUL_MAT_ID(ffn_moe_down-44): 205 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 353 us MUL_MAT_ID(ffn_moe_down-45): 197 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 468 us MUL_MAT_ID(ffn_moe_down-46): 314 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 354 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 415 us MUL_MAT_ID(ffn_moe_down-48): 179 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 390 us MUL_MAT_ID(ffn_moe_down-49): 224 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 382 us MUL_MAT_ID(ffn_moe_down-50): 543 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 423 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 374 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 392 us MUL_MAT_ID(ffn_moe_down-53): 202 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 358 us MUL_MAT_ID(ffn_moe_down-54): 227 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 410 us MUL_MAT_ID(ffn_moe_down-55): 488 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 417 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 361 us MUL_MAT_ID(ffn_moe_down-57): 374 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 372 us MUL_MAT_ID(ffn_moe_down-58): 212 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 394 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 380 us MUL_MAT_ID(ffn_moe_down-60): 191 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 392 us MUL_MAT_ID(ffn_moe_down-61): 214 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 420 us MUL_MAT_ID(ffn_moe_down-62): 273 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 361 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 377 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 418 us MUL_MAT_ID(ffn_moe_down-65): 284 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 432 us MUL_MAT_ID(ffn_moe_down-66): 187 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 377 us MUL_MAT_ID(ffn_moe_down-67): 209 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 369 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 388 us MUL_MAT_ID(ffn_moe_down-69): 431 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 371 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 397 us MUL_MAT_ID(ffn_moe_down-71): 280 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 395 us MUL_MAT_ID(ffn_moe_down-73): 202 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 383 us MUL_MAT_ID(ffn_moe_down-74): 270 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 366 us MUL_MAT_ID(ffn_moe_down-75): 194 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 389 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 360 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 406 us MUL_MAT_ID(ffn_moe_down-78): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 368 us MUL_MAT_ID(ffn_moe_down-79): 289 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 359 us MUL_MAT_ID(ffn_moe_down-80): 373 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 432 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 381 us MUL_MAT_ID(ffn_moe_down-82): 300 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 426 us MUL_MAT_ID(ffn_moe_down-83): 293 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 434 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 389 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 494 us MUL_MAT_ID(ffn_moe_down-86): 236 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 396 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 373 us MUL_MAT_ID(ffn_moe_down-88): 186 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 423 us MUL_MAT_ID(ffn_moe_down-89): 208 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 376 us MUL_MAT_ID(ffn_moe_down-90): 227 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 408 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 22 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 430 us MUL_MAT_ID(ffn_moe_down-25): 251 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 471 us MUL_MAT_ID(ffn_moe_down-26): 468 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 489 us MUL_MAT_ID(ffn_moe_down-27): 334 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 461 us MUL_MAT_ID(ffn_moe_down-29): 279 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 389 us MUL_MAT_ID(ffn_moe_down-30): 244 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 466 us MUL_MAT_ID(ffn_moe_down-31): 202 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 392 us MUL_MAT_ID(ffn_moe_down-32): 250 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 379 us MUL_MAT_ID(ffn_moe_down-33): 273 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 471 us MUL_MAT_ID(ffn_moe_down-34): 247 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 476 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 447 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 477 us MUL_MAT_ID(ffn_moe_down-37): 327 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 403 us MUL_MAT_ID(ffn_moe_down-38): 421 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 427 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 397 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 505 us MUL_MAT_ID(ffn_moe_down-41): 227 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 478 us MUL_MAT_ID(ffn_moe_down-42): 214 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 436 us MUL_MAT_ID(ffn_moe_down-43): 261 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 457 us MUL_MAT_ID(ffn_moe_down-44): 232 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 386 us MUL_MAT_ID(ffn_moe_down-45): 238 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 473 us MUL_MAT_ID(ffn_moe_down-46): 181 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 403 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 453 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 366 us MUL_MAT_ID(ffn_moe_down-49): 226 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 426 us MUL_MAT_ID(ffn_moe_down-50): 219 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 374 us MUL_MAT_ID(ffn_moe_down-51): 435 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 430 us MUL_MAT_ID(ffn_moe_down-52): 547 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 416 us MUL_MAT_ID(ffn_moe_down-53): 215 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 410 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 429 us MUL_MAT_ID(ffn_moe_down-55): 205 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 469 us MUL_MAT_ID(ffn_moe_down-56): 258 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 452 us MUL_MAT_ID(ffn_moe_down-57): 279 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 504 us MUL_MAT_ID(ffn_moe_down-58): 248 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 506 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 421 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 366 us MUL_MAT_ID(ffn_moe_down-61): 194 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 397 us MUL_MAT_ID(ffn_moe_down-62): 180 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 521 us MUL_MAT_ID(ffn_moe_down-63): 214 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 410 us MUL_MAT_ID(ffn_moe_down-64): 227 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 430 us MUL_MAT_ID(ffn_moe_down-65): 224 us ggml_barrier(...): 344 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 440 us MUL_MAT_ID(ffn_moe_down-66): 290 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 392 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 470 us MUL_MAT_ID(ffn_moe_down-68): 202 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 497 us MUL_MAT_ID(ffn_moe_down-69): 182 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 420 us MUL_MAT_ID(ffn_moe_down-70): 208 us ggml_barrier(...): 512 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 404 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 396 us MUL_MAT_ID(ffn_moe_down-72): 273 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 491 us MUL_MAT_ID(ffn_moe_down-73): 450 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 465 us MUL_MAT_ID(ffn_moe_down-74): 225 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 373 us MUL_MAT_ID(ffn_moe_down-75): 198 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 363 us MUL_MAT_ID(ffn_moe_down-76): 188 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 445 us MUL_MAT_ID(ffn_moe_down-77): 234 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 477 us MUL_MAT_ID(ffn_moe_down-78): 185 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 399 us MUL_MAT_ID(ffn_moe_down-79): 223 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 466 us MUL_MAT_ID(ffn_moe_down-80): 276 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 401 us MUL_MAT_ID(ffn_moe_down-81): 445 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 453 us MUL_MAT_ID(ffn_moe_down-82): 217 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 413 us MUL_MAT_ID(ffn_moe_down-83): 232 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 383 us MUL_MAT_ID(ffn_moe_down-84): 428 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 416 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 542 us MUL_MAT_ID(ffn_moe_down-86): 236 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 420 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 405 us MUL_MAT_ID(ffn_moe_down-88): 315 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 386 us MUL_MAT_ID(ffn_moe_down-89): 233 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 350 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 556 us MUL_MAT_ID(ffn_moe_down-91): 266 us ggml_barrier(...): 97 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 439 us MUL_MAT_ID(ffn_moe_down-25): 247 us ggml_barrier(...): 474 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 430 us MUL_MAT_ID(ffn_moe_down-26): 214 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 460 us MUL_MAT_ID(ffn_moe_down-27): 216 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 420 us MUL_MAT_ID(ffn_moe_down-28): 185 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 398 us MUL_MAT_ID(ffn_moe_down-29): 551 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 419 us MUL_MAT_ID(ffn_moe_down-30): 211 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 373 us MUL_MAT_ID(ffn_moe_down-31): 187 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 390 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 370 us MUL_MAT_ID(ffn_moe_down-33): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 473 us MUL_MAT_ID(ffn_moe_down-34): 234 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 373 us MUL_MAT_ID(ffn_moe_down-35): 200 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 390 us MUL_MAT_ID(ffn_moe_down-36): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 454 us MUL_MAT_ID(ffn_moe_down-37): 322 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 383 us MUL_MAT_ID(ffn_moe_down-38): 365 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 390 us MUL_MAT_ID(ffn_moe_down-39): 269 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 375 us MUL_MAT_ID(ffn_moe_down-40): 418 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 485 us MUL_MAT_ID(ffn_moe_down-41): 372 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 381 us MUL_MAT_ID(ffn_moe_down-42): 207 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 373 us MUL_MAT_ID(ffn_moe_down-43): 209 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 362 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 374 us MUL_MAT_ID(ffn_moe_down-45): 189 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 356 us MUL_MAT_ID(ffn_moe_down-46): 227 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 489 us MUL_MAT_ID(ffn_moe_down-47): 172 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 377 us MUL_MAT_ID(ffn_moe_down-48): 199 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 416 us MUL_MAT_ID(ffn_moe_down-49): 214 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 385 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 402 us MUL_MAT_ID(ffn_moe_down-51): 361 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 372 us MUL_MAT_ID(ffn_moe_down-52): 310 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 390 us MUL_MAT_ID(ffn_moe_down-53): 241 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 363 us MUL_MAT_ID(ffn_moe_down-54): 539 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 500 us MUL_MAT_ID(ffn_moe_down-55): 168 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 458 us MUL_MAT_ID(ffn_moe_down-56): 226 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 353 us MUL_MAT_ID(ffn_moe_down-57): 174 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 391 us MUL_MAT_ID(ffn_moe_down-58): 228 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 360 us MUL_MAT_ID(ffn_moe_down-59): 171 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 362 us MUL_MAT_ID(ffn_moe_down-60): 180 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 358 us MUL_MAT_ID(ffn_moe_down-61): 225 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 427 us MUL_MAT_ID(ffn_moe_down-62): 168 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 376 us MUL_MAT_ID(ffn_moe_down-63): 202 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 372 us MUL_MAT_ID(ffn_moe_down-64): 225 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 378 us MUL_MAT_ID(ffn_moe_down-65): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 381 us MUL_MAT_ID(ffn_moe_down-66): 217 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 368 us MUL_MAT_ID(ffn_moe_down-67): 202 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 464 us MUL_MAT_ID(ffn_moe_down-68): 302 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 422 us MUL_MAT_ID(ffn_moe_down-69): 187 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 413 us MUL_MAT_ID(ffn_moe_down-70): 225 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 443 us MUL_MAT_ID(ffn_moe_down-71): 440 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 369 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 378 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 379 us MUL_MAT_ID(ffn_moe_down-74): 221 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 389 us MUL_MAT_ID(ffn_moe_down-75): 627 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 455 us MUL_MAT_ID(ffn_moe_down-76): 234 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 476 us MUL_MAT_ID(ffn_moe_down-77): 223 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 382 us MUL_MAT_ID(ffn_moe_down-78): 304 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 370 us MUL_MAT_ID(ffn_moe_down-79): 179 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 382 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 387 us MUL_MAT_ID(ffn_moe_down-81): 188 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 444 us MUL_MAT_ID(ffn_moe_down-82): 447 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 437 us MUL_MAT_ID(ffn_moe_down-83): 327 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 348 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 408 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 405 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 440 us MUL_MAT_ID(ffn_moe_down-87): 197 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 429 us MUL_MAT_ID(ffn_moe_down-88): 198 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 399 us MUL_MAT_ID(ffn_moe_down-89): 339 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 410 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 420 us MUL_MAT_ID(ffn_moe_down-91): 531 us ggml_barrier(...): 189 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 429 us MUL_MAT_ID(ffn_moe_down-25): 178 us ggml_barrier(...): 419 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 479 us MUL_MAT_ID(ffn_moe_down-26): 200 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 449 us MUL_MAT_ID(ffn_moe_down-27): 222 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 374 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 390 us MUL_MAT_ID(ffn_moe_down-29): 206 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 400 us MUL_MAT_ID(ffn_moe_down-30): 213 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 370 us MUL_MAT_ID(ffn_moe_down-31): 275 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 408 us MUL_MAT_ID(ffn_moe_down-32): 166 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 421 us MUL_MAT_ID(ffn_moe_down-33): 182 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 372 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 404 us MUL_MAT_ID(ffn_moe_down-35): 327 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 395 us MUL_MAT_ID(ffn_moe_down-36): 215 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 418 us MUL_MAT_ID(ffn_moe_down-37): 239 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 393 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 361 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 404 us MUL_MAT_ID(ffn_moe_down-40): 240 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 400 us MUL_MAT_ID(ffn_moe_down-41): 191 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 412 us MUL_MAT_ID(ffn_moe_down-42): 204 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 476 us MUL_MAT_ID(ffn_moe_down-43): 215 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 372 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 361 us MUL_MAT_ID(ffn_moe_down-45): 253 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 414 us MUL_MAT_ID(ffn_moe_down-46): 203 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 362 us MUL_MAT_ID(ffn_moe_down-47): 214 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 358 us MUL_MAT_ID(ffn_moe_down-48): 155 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 410 us MUL_MAT_ID(ffn_moe_down-49): 203 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 425 us MUL_MAT_ID(ffn_moe_down-50): 247 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 447 us MUL_MAT_ID(ffn_moe_down-51): 306 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 390 us MUL_MAT_ID(ffn_moe_down-52): 222 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 516 us MUL_MAT_ID(ffn_moe_down-53): 244 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 379 us MUL_MAT_ID(ffn_moe_down-54): 483 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 513 us MUL_MAT_ID(ffn_moe_down-55): 206 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 393 us MUL_MAT_ID(ffn_moe_down-56): 216 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 453 us MUL_MAT_ID(ffn_moe_down-57): 312 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 432 us MUL_MAT_ID(ffn_moe_down-58): 220 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 513 us MUL_MAT_ID(ffn_moe_down-59): 204 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 400 us MUL_MAT_ID(ffn_moe_down-60): 226 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 369 us MUL_MAT_ID(ffn_moe_down-61): 163 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 480 us MUL_MAT_ID(ffn_moe_down-62): 241 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 412 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 377 us MUL_MAT_ID(ffn_moe_down-64): 214 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 440 us MUL_MAT_ID(ffn_moe_down-65): 196 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 408 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 434 us MUL_MAT_ID(ffn_moe_down-67): 268 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 415 us MUL_MAT_ID(ffn_moe_down-68): 526 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 552 us MUL_MAT_ID(ffn_moe_down-69): 225 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 437 us MUL_MAT_ID(ffn_moe_down-70): 275 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 450 us MUL_MAT_ID(ffn_moe_down-71): 215 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 426 us MUL_MAT_ID(ffn_moe_down-72): 228 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 431 us MUL_MAT_ID(ffn_moe_down-73): 417 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 389 us MUL_MAT_ID(ffn_moe_down-74): 216 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 379 us MUL_MAT_ID(ffn_moe_down-75): 236 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 418 us MUL_MAT_ID(ffn_moe_down-76): 165 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 447 us MUL_MAT_ID(ffn_moe_down-77): 278 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 398 us MUL_MAT_ID(ffn_moe_down-78): 230 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 406 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 388 us MUL_MAT_ID(ffn_moe_down-80): 191 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 407 us MUL_MAT_ID(ffn_moe_down-81): 571 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 442 us MUL_MAT_ID(ffn_moe_down-82): 221 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 384 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 438 us MUL_MAT_ID(ffn_moe_down-84): 272 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 381 us MUL_MAT_ID(ffn_moe_down-85): 200 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 440 us MUL_MAT_ID(ffn_moe_down-86): 211 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 421 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 397 us MUL_MAT_ID(ffn_moe_down-88): 223 us ggml_barrier(...): 482 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 500 us MUL_MAT_ID(ffn_moe_down-89): 175 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 425 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 426 us MUL_MAT_ID(ffn_moe_down-91): 249 us ggml_barrier(...): 166 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 400 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 453 us MUL_MAT_ID(ffn_moe_down-26): 215 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 374 us MUL_MAT_ID(ffn_moe_down-27): 346 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 388 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 392 us MUL_MAT_ID(ffn_moe_down-29): 192 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 415 us MUL_MAT_ID(ffn_moe_down-30): 192 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 360 us MUL_MAT_ID(ffn_moe_down-31): 171 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 400 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 406 us MUL_MAT_ID(ffn_moe_down-33): 229 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 400 us MUL_MAT_ID(ffn_moe_down-34): 348 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 365 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 359 us MUL_MAT_ID(ffn_moe_down-36): 226 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 418 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 460 us MUL_MAT_ID(ffn_moe_down-38): 309 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 503 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 392 us MUL_MAT_ID(ffn_moe_down-40): 244 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 395 us MUL_MAT_ID(ffn_moe_down-41): 313 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 392 us MUL_MAT_ID(ffn_moe_down-42): 192 us ggml_barrier(...): 334 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 423 us MUL_MAT_ID(ffn_moe_down-43): 259 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 359 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 398 us MUL_MAT_ID(ffn_moe_down-45): 223 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 399 us MUL_MAT_ID(ffn_moe_down-46): 183 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 369 us MUL_MAT_ID(ffn_moe_down-47): 263 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 393 us MUL_MAT_ID(ffn_moe_down-48): 328 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 401 us MUL_MAT_ID(ffn_moe_down-49): 296 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 382 us MUL_MAT_ID(ffn_moe_down-50): 204 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 390 us MUL_MAT_ID(ffn_moe_down-51): 182 us ggml_barrier(...): 436 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 426 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 376 us MUL_MAT_ID(ffn_moe_down-53): 190 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 363 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 366 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 479 us MUL_MAT_ID(ffn_moe_down-56): 343 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 482 us MUL_MAT_ID(ffn_moe_down-57): 202 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 468 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 392 us MUL_MAT_ID(ffn_moe_down-59): 326 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 388 us MUL_MAT_ID(ffn_moe_down-60): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 379 us MUL_MAT_ID(ffn_moe_down-61): 271 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 378 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 370 us MUL_MAT_ID(ffn_moe_down-63): 301 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 410 us MUL_MAT_ID(ffn_moe_down-64): 215 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 408 us MUL_MAT_ID(ffn_moe_down-65): 208 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 388 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 417 us MUL_MAT_ID(ffn_moe_down-67): 227 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 392 us MUL_MAT_ID(ffn_moe_down-68): 253 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 383 us MUL_MAT_ID(ffn_moe_down-69): 183 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 380 us MUL_MAT_ID(ffn_moe_down-70): 330 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 373 us MUL_MAT_ID(ffn_moe_down-71): 259 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 429 us MUL_MAT_ID(ffn_moe_down-72): 299 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 407 us MUL_MAT_ID(ffn_moe_down-73): 271 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 365 us MUL_MAT_ID(ffn_moe_down-74): 221 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 390 us MUL_MAT_ID(ffn_moe_down-75): 188 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 374 us MUL_MAT_ID(ffn_moe_down-76): 386 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 374 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 473 us MUL_MAT_ID(ffn_moe_down-78): 187 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 431 us MUL_MAT_ID(ffn_moe_down-79): 242 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 444 us MUL_MAT_ID(ffn_moe_down-80): 189 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 485 us MUL_MAT_ID(ffn_moe_down-81): 285 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 380 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 426 us MUL_MAT_ID(ffn_moe_down-83): 404 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 373 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 456 us MUL_MAT_ID(ffn_moe_down-85): 183 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 465 us MUL_MAT_ID(ffn_moe_down-86): 393 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 402 us MUL_MAT_ID(ffn_moe_down-87): 176 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 401 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 403 us MUL_MAT_ID(ffn_moe_down-89): 373 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 372 us MUL_MAT_ID(ffn_moe_down-90): 262 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 362 us MUL_MAT_ID(ffn_moe_down-91): 294 us ggml_barrier(...): 52 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 397 us MUL_MAT_ID(ffn_moe_down-25): 296 us ggml_barrier(...): 436 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 510 us MUL_MAT_ID(ffn_moe_down-26): 188 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 399 us MUL_MAT_ID(ffn_moe_down-27): 292 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 404 us MUL_MAT_ID(ffn_moe_down-28): 210 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 437 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 369 us MUL_MAT_ID(ffn_moe_down-30): 578 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 392 us MUL_MAT_ID(ffn_moe_down-31): 237 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 418 us MUL_MAT_ID(ffn_moe_down-32): 213 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 396 us MUL_MAT_ID(ffn_moe_down-33): 186 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 379 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 411 us MUL_MAT_ID(ffn_moe_down-35): 282 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 479 us MUL_MAT_ID(ffn_moe_down-36): 221 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 428 us MUL_MAT_ID(ffn_moe_down-37): 334 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 452 us MUL_MAT_ID(ffn_moe_down-38): 220 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 414 us MUL_MAT_ID(ffn_moe_down-39): 278 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 440 us MUL_MAT_ID(ffn_moe_down-40): 198 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 378 us MUL_MAT_ID(ffn_moe_down-41): 374 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 384 us MUL_MAT_ID(ffn_moe_down-42): 168 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 439 us MUL_MAT_ID(ffn_moe_down-43): 252 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 375 us MUL_MAT_ID(ffn_moe_down-44): 274 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 360 us MUL_MAT_ID(ffn_moe_down-45): 376 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 502 us MUL_MAT_ID(ffn_moe_down-46): 239 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 364 us MUL_MAT_ID(ffn_moe_down-47): 238 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 397 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 375 us MUL_MAT_ID(ffn_moe_down-49): 179 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 394 us MUL_MAT_ID(ffn_moe_down-50): 262 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 452 us MUL_MAT_ID(ffn_moe_down-51): 323 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 392 us MUL_MAT_ID(ffn_moe_down-52): 279 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 367 us MUL_MAT_ID(ffn_moe_down-53): 343 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 420 us MUL_MAT_ID(ffn_moe_down-54): 211 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 393 us MUL_MAT_ID(ffn_moe_down-55): 362 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 316 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 405 us MUL_MAT_ID(ffn_moe_down-57): 357 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 357 us MUL_MAT_ID(ffn_moe_down-58): 249 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 399 us MUL_MAT_ID(ffn_moe_down-59): 233 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 451 us MUL_MAT_ID(ffn_moe_down-60): 177 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 428 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 479 us MUL_MAT_ID(ffn_moe_down-62): 176 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 418 us MUL_MAT_ID(ffn_moe_down-63): 409 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 402 us MUL_MAT_ID(ffn_moe_down-64): 226 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 404 us MUL_MAT_ID(ffn_moe_down-65): 225 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 357 us MUL_MAT_ID(ffn_moe_down-66): 665 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 384 us MUL_MAT_ID(ffn_moe_down-67): 191 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 455 us MUL_MAT_ID(ffn_moe_down-68): 223 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 363 us MUL_MAT_ID(ffn_moe_down-69): 162 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 361 us MUL_MAT_ID(ffn_moe_down-70): 188 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 395 us MUL_MAT_ID(ffn_moe_down-71): 240 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 366 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 382 us MUL_MAT_ID(ffn_moe_down-73): 204 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 483 us MUL_MAT_ID(ffn_moe_down-74): 163 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 384 us MUL_MAT_ID(ffn_moe_down-75): 430 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 358 us MUL_MAT_ID(ffn_moe_down-76): 207 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 375 us MUL_MAT_ID(ffn_moe_down-77): 225 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 407 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 419 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 349 us MUL_MAT_ID(ffn_moe_down-80): 236 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 400 us MUL_MAT_ID(ffn_moe_down-81): 227 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 373 us MUL_MAT_ID(ffn_moe_down-82): 266 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 390 us MUL_MAT_ID(ffn_moe_down-83): 241 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 372 us MUL_MAT_ID(ffn_moe_down-84): 429 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 405 us MUL_MAT_ID(ffn_moe_down-85): 661 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 415 us MUL_MAT_ID(ffn_moe_down-86): 185 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 378 us MUL_MAT_ID(ffn_moe_down-87): 255 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 390 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 379 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 412 us MUL_MAT_ID(ffn_moe_down-90): 241 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 392 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 21 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 601 us MUL_MAT_ID(ffn_moe_down-25): 184 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 407 us MUL_MAT_ID(ffn_moe_down-26): 213 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 526 us MUL_MAT_ID(ffn_moe_down-27): 478 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 379 us MUL_MAT_ID(ffn_moe_down-28): 215 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 435 us MUL_MAT_ID(ffn_moe_down-29): 314 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 411 us MUL_MAT_ID(ffn_moe_down-30): 181 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 391 us MUL_MAT_ID(ffn_moe_down-31): 215 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 402 us MUL_MAT_ID(ffn_moe_down-32): 248 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 405 us MUL_MAT_ID(ffn_moe_down-33): 209 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 427 us MUL_MAT_ID(ffn_moe_down-34): 226 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 385 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 410 us MUL_MAT_ID(ffn_moe_down-36): 298 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 380 us MUL_MAT_ID(ffn_moe_down-37): 303 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 434 us MUL_MAT_ID(ffn_moe_down-38): 243 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 430 us MUL_MAT_ID(ffn_moe_down-39): 281 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 370 us MUL_MAT_ID(ffn_moe_down-40): 228 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 437 us MUL_MAT_ID(ffn_moe_down-41): 374 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 416 us MUL_MAT_ID(ffn_moe_down-42): 275 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 442 us MUL_MAT_ID(ffn_moe_down-43): 287 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 414 us MUL_MAT_ID(ffn_moe_down-44): 413 us ggml_barrier(...): 351 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 426 us MUL_MAT_ID(ffn_moe_down-45): 300 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 379 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 449 us MUL_MAT_ID(ffn_moe_down-47): 291 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 378 us MUL_MAT_ID(ffn_moe_down-48): 256 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 395 us MUL_MAT_ID(ffn_moe_down-49): 305 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 407 us MUL_MAT_ID(ffn_moe_down-50): 270 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 417 us MUL_MAT_ID(ffn_moe_down-51): 237 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 417 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 360 us MUL_MAT_ID(ffn_moe_down-53): 227 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 420 us MUL_MAT_ID(ffn_moe_down-54): 259 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 437 us MUL_MAT_ID(ffn_moe_down-55): 251 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 353 us MUL_MAT_ID(ffn_moe_down-56): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 423 us MUL_MAT_ID(ffn_moe_down-57): 405 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 355 us MUL_MAT_ID(ffn_moe_down-58): 167 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 472 us MUL_MAT_ID(ffn_moe_down-59): 258 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 402 us MUL_MAT_ID(ffn_moe_down-60): 260 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 499 us MUL_MAT_ID(ffn_moe_down-61): 210 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 486 us MUL_MAT_ID(ffn_moe_down-62): 173 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 398 us MUL_MAT_ID(ffn_moe_down-63): 295 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 399 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 436 us MUL_MAT_ID(ffn_moe_down-65): 274 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 350 us MUL_MAT_ID(ffn_moe_down-66): 235 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 407 us MUL_MAT_ID(ffn_moe_down-67): 233 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 445 us MUL_MAT_ID(ffn_moe_down-68): 183 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 390 us MUL_MAT_ID(ffn_moe_down-69): 439 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 483 us MUL_MAT_ID(ffn_moe_down-70): 173 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 428 us MUL_MAT_ID(ffn_moe_down-71): 187 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 503 us MUL_MAT_ID(ffn_moe_down-72): 277 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 412 us MUL_MAT_ID(ffn_moe_down-73): 281 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 357 us MUL_MAT_ID(ffn_moe_down-74): 364 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 373 us MUL_MAT_ID(ffn_moe_down-75): 210 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 368 us MUL_MAT_ID(ffn_moe_down-76): 204 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 387 us MUL_MAT_ID(ffn_moe_down-77): 186 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 427 us MUL_MAT_ID(ffn_moe_down-78): 224 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 356 us MUL_MAT_ID(ffn_moe_down-79): 227 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 367 us MUL_MAT_ID(ffn_moe_down-80): 251 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 443 us MUL_MAT_ID(ffn_moe_down-81): 318 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 468 us MUL_MAT_ID(ffn_moe_down-82): 191 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 452 us MUL_MAT_ID(ffn_moe_down-83): 280 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 411 us MUL_MAT_ID(ffn_moe_down-84): 250 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 414 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 390 us MUL_MAT_ID(ffn_moe_down-86): 286 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 389 us MUL_MAT_ID(ffn_moe_down-87): 186 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 398 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 385 us MUL_MAT_ID(ffn_moe_down-89): 365 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 378 us MUL_MAT_ID(ffn_moe_down-90): 226 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 388 us MUL_MAT_ID(ffn_moe_down-91): 246 us ggml_barrier(...): 196 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 374 us MUL_MAT_ID(ffn_moe_down-25): 295 us ggml_barrier(...): 478 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 414 us MUL_MAT_ID(ffn_moe_down-26): 214 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 483 us MUL_MAT_ID(ffn_moe_down-27): 179 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 423 us MUL_MAT_ID(ffn_moe_down-28): 214 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 465 us MUL_MAT_ID(ffn_moe_down-29): 219 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 555 us MUL_MAT_ID(ffn_moe_down-30): 209 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 418 us MUL_MAT_ID(ffn_moe_down-31): 264 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 430 us MUL_MAT_ID(ffn_moe_down-32): 216 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 365 us MUL_MAT_ID(ffn_moe_down-33): 230 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 480 us MUL_MAT_ID(ffn_moe_down-34): 228 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 474 us MUL_MAT_ID(ffn_moe_down-35): 185 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 505 us MUL_MAT_ID(ffn_moe_down-36): 263 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 369 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 369 us MUL_MAT_ID(ffn_moe_down-38): 232 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 491 us MUL_MAT_ID(ffn_moe_down-39): 223 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 408 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 343 us MUL_MAT_ID(ffn_moe_down-41): 210 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 398 us MUL_MAT_ID(ffn_moe_down-42): 188 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 402 us MUL_MAT_ID(ffn_moe_down-43): 219 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 355 us MUL_MAT_ID(ffn_moe_down-44): 182 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 364 us MUL_MAT_ID(ffn_moe_down-45): 383 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 433 us MUL_MAT_ID(ffn_moe_down-46): 369 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 489 us MUL_MAT_ID(ffn_moe_down-47): 170 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 355 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 388 us MUL_MAT_ID(ffn_moe_down-49): 206 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 373 us MUL_MAT_ID(ffn_moe_down-50): 206 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 504 us MUL_MAT_ID(ffn_moe_down-51): 340 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 390 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 359 us MUL_MAT_ID(ffn_moe_down-53): 306 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 396 us MUL_MAT_ID(ffn_moe_down-54): 266 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 347 us MUL_MAT_ID(ffn_moe_down-55): 271 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 462 us MUL_MAT_ID(ffn_moe_down-56): 203 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 404 us MUL_MAT_ID(ffn_moe_down-57): 186 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 407 us MUL_MAT_ID(ffn_moe_down-58): 209 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 491 us MUL_MAT_ID(ffn_moe_down-59): 229 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 398 us MUL_MAT_ID(ffn_moe_down-60): 216 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 429 us MUL_MAT_ID(ffn_moe_down-61): 194 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 400 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 368 us MUL_MAT_ID(ffn_moe_down-63): 184 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 401 us MUL_MAT_ID(ffn_moe_down-64): 201 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 436 us MUL_MAT_ID(ffn_moe_down-65): 302 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 389 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 481 us MUL_MAT_ID(ffn_moe_down-67): 189 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 369 us MUL_MAT_ID(ffn_moe_down-68): 189 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 442 us MUL_MAT_ID(ffn_moe_down-69): 225 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 368 us MUL_MAT_ID(ffn_moe_down-70): 185 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 392 us MUL_MAT_ID(ffn_moe_down-71): 208 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 361 us MUL_MAT_ID(ffn_moe_down-72): 205 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 383 us MUL_MAT_ID(ffn_moe_down-73): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 363 us MUL_MAT_ID(ffn_moe_down-74): 200 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 374 us MUL_MAT_ID(ffn_moe_down-75): 206 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 367 us MUL_MAT_ID(ffn_moe_down-76): 176 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 391 us MUL_MAT_ID(ffn_moe_down-77): 302 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 482 us MUL_MAT_ID(ffn_moe_down-78): 219 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 397 us MUL_MAT_ID(ffn_moe_down-79): 382 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 356 us MUL_MAT_ID(ffn_moe_down-80): 193 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 436 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 384 us MUL_MAT_ID(ffn_moe_down-82): 180 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 397 us MUL_MAT_ID(ffn_moe_down-83): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 408 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 397 us MUL_MAT_ID(ffn_moe_down-85): 713 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 483 us MUL_MAT_ID(ffn_moe_down-86): 256 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 397 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 352 us MUL_MAT_ID(ffn_moe_down-88): 215 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 375 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 374 us MUL_MAT_ID(ffn_moe_down-90): 256 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 371 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 61 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 548 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 361 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 446 us MUL_MAT_ID(ffn_moe_down-26): 304 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 422 us MUL_MAT_ID(ffn_moe_down-27): 183 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 375 us MUL_MAT_ID(ffn_moe_down-28): 180 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 475 us MUL_MAT_ID(ffn_moe_down-29): 573 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 405 us MUL_MAT_ID(ffn_moe_down-30): 372 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 360 us MUL_MAT_ID(ffn_moe_down-31): 204 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 388 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 377 us MUL_MAT_ID(ffn_moe_down-33): 320 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 446 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 412 us MUL_MAT_ID(ffn_moe_down-35): 249 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 373 us MUL_MAT_ID(ffn_moe_down-36): 199 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 410 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 410 us MUL_MAT_ID(ffn_moe_down-38): 205 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 370 us MUL_MAT_ID(ffn_moe_down-39): 250 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 364 us MUL_MAT_ID(ffn_moe_down-40): 215 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 408 us MUL_MAT_ID(ffn_moe_down-41): 260 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 477 us MUL_MAT_ID(ffn_moe_down-42): 240 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 393 us MUL_MAT_ID(ffn_moe_down-43): 293 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 410 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 353 us MUL_MAT_ID(ffn_moe_down-45): 193 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 421 us MUL_MAT_ID(ffn_moe_down-46): 391 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 393 us MUL_MAT_ID(ffn_moe_down-47): 245 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 388 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 431 us MUL_MAT_ID(ffn_moe_down-49): 260 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 406 us MUL_MAT_ID(ffn_moe_down-50): 199 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 397 us MUL_MAT_ID(ffn_moe_down-51): 232 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 426 us MUL_MAT_ID(ffn_moe_down-52): 186 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 353 us MUL_MAT_ID(ffn_moe_down-53): 200 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 380 us MUL_MAT_ID(ffn_moe_down-54): 186 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 404 us MUL_MAT_ID(ffn_moe_down-55): 241 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 434 us MUL_MAT_ID(ffn_moe_down-56): 245 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 370 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 437 us MUL_MAT_ID(ffn_moe_down-58): 325 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 423 us MUL_MAT_ID(ffn_moe_down-59): 204 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 408 us MUL_MAT_ID(ffn_moe_down-60): 237 us ggml_barrier(...): 322 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 396 us MUL_MAT_ID(ffn_moe_down-61): 209 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 492 us MUL_MAT_ID(ffn_moe_down-62): 171 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 363 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 337 us MUL_MAT_ID(ffn_moe_down-64): 231 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 390 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 368 us MUL_MAT_ID(ffn_moe_down-66): 183 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 481 us MUL_MAT_ID(ffn_moe_down-67): 303 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 432 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 343 us MUL_MAT_ID(ffn_moe_down-69): 228 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 385 us MUL_MAT_ID(ffn_moe_down-70): 180 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 353 us MUL_MAT_ID(ffn_moe_down-71): 213 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 408 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 219 us ggml_barrier(...): 502 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 420 us MUL_MAT_ID(ffn_moe_down-74): 295 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 396 us MUL_MAT_ID(ffn_moe_down-75): 331 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 512 us MUL_MAT_ID(ffn_moe_down-76): 196 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 396 us MUL_MAT_ID(ffn_moe_down-77): 278 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 376 us MUL_MAT_ID(ffn_moe_down-78): 212 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 533 us MUL_MAT_ID(ffn_moe_down-79): 218 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 373 us MUL_MAT_ID(ffn_moe_down-80): 194 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 411 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 377 us MUL_MAT_ID(ffn_moe_down-82): 509 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 399 us MUL_MAT_ID(ffn_moe_down-83): 317 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 352 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 480 us MUL_MAT_ID(ffn_moe_down-85): 600 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 485 us MUL_MAT_ID(ffn_moe_down-86): 278 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 439 us MUL_MAT_ID(ffn_moe_down-87): 173 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 449 us MUL_MAT_ID(ffn_moe_down-88): 280 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 371 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 403 us MUL_MAT_ID(ffn_moe_down-90): 527 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 401 us MUL_MAT_ID(ffn_moe_down-91): 228 us ggml_barrier(...): 26 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 400 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 428 us MUL_MAT_ID(ffn_moe_down-26): 352 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 403 us MUL_MAT_ID(ffn_moe_down-27): 229 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 427 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 393 us MUL_MAT_ID(ffn_moe_down-29): 206 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 398 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 406 us MUL_MAT_ID(ffn_moe_down-31): 422 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 377 us MUL_MAT_ID(ffn_moe_down-32): 518 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 468 us MUL_MAT_ID(ffn_moe_down-33): 368 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 377 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 386 us MUL_MAT_ID(ffn_moe_down-35): 399 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 429 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 381 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 352 us MUL_MAT_ID(ffn_moe_down-38): 223 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 389 us MUL_MAT_ID(ffn_moe_down-39): 239 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 374 us MUL_MAT_ID(ffn_moe_down-40): 447 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 484 us MUL_MAT_ID(ffn_moe_down-41): 180 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 354 us MUL_MAT_ID(ffn_moe_down-42): 186 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 355 us MUL_MAT_ID(ffn_moe_down-43): 377 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 387 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 399 us MUL_MAT_ID(ffn_moe_down-45): 197 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 345 us MUL_MAT_ID(ffn_moe_down-46): 224 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 360 us MUL_MAT_ID(ffn_moe_down-47): 246 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 511 us MUL_MAT_ID(ffn_moe_down-48): 220 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 379 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 378 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 359 us MUL_MAT_ID(ffn_moe_down-51): 188 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 370 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 392 us MUL_MAT_ID(ffn_moe_down-53): 192 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 528 us MUL_MAT_ID(ffn_moe_down-54): 169 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 354 us MUL_MAT_ID(ffn_moe_down-55): 186 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 459 us MUL_MAT_ID(ffn_moe_down-56): 326 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 382 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 380 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 376 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 472 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 388 us MUL_MAT_ID(ffn_moe_down-61): 181 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 355 us MUL_MAT_ID(ffn_moe_down-62): 297 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 404 us MUL_MAT_ID(ffn_moe_down-63): 182 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 400 us MUL_MAT_ID(ffn_moe_down-64): 400 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 401 us MUL_MAT_ID(ffn_moe_down-65): 193 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 363 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 378 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 390 us MUL_MAT_ID(ffn_moe_down-68): 197 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 368 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 405 us MUL_MAT_ID(ffn_moe_down-70): 197 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 366 us MUL_MAT_ID(ffn_moe_down-71): 209 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 381 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 399 us MUL_MAT_ID(ffn_moe_down-73): 317 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 381 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 383 us MUL_MAT_ID(ffn_moe_down-75): 214 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 356 us MUL_MAT_ID(ffn_moe_down-76): 188 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 376 us MUL_MAT_ID(ffn_moe_down-77): 199 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 422 us MUL_MAT_ID(ffn_moe_down-78): 206 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 406 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 406 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 394 us MUL_MAT_ID(ffn_moe_down-81): 377 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 398 us MUL_MAT_ID(ffn_moe_down-82): 364 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 454 us MUL_MAT_ID(ffn_moe_down-83): 557 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 371 us MUL_MAT_ID(ffn_moe_down-84): 252 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 378 us MUL_MAT_ID(ffn_moe_down-85): 260 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 399 us MUL_MAT_ID(ffn_moe_down-86): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 457 us MUL_MAT_ID(ffn_moe_down-87): 217 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 189 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 389 us MUL_MAT_ID(ffn_moe_down-89): 209 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 433 us MUL_MAT_ID(ffn_moe_down-90): 246 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 407 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 14 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 399 us MUL_MAT_ID(ffn_moe_down-25): 296 us ggml_barrier(...): 519 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 373 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 371 us MUL_MAT_ID(ffn_moe_down-27): 200 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 378 us MUL_MAT_ID(ffn_moe_down-28): 364 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 361 us MUL_MAT_ID(ffn_moe_down-29): 212 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 353 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 385 us MUL_MAT_ID(ffn_moe_down-31): 202 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 357 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 449 us MUL_MAT_ID(ffn_moe_down-33): 172 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 410 us MUL_MAT_ID(ffn_moe_down-34): 190 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 450 us MUL_MAT_ID(ffn_moe_down-35): 307 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 487 us MUL_MAT_ID(ffn_moe_down-36): 218 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 399 us MUL_MAT_ID(ffn_moe_down-37): 252 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 405 us MUL_MAT_ID(ffn_moe_down-38): 189 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 374 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 420 us MUL_MAT_ID(ffn_moe_down-40): 220 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 383 us MUL_MAT_ID(ffn_moe_down-41): 538 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 375 us MUL_MAT_ID(ffn_moe_down-42): 258 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 402 us MUL_MAT_ID(ffn_moe_down-43): 184 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 375 us MUL_MAT_ID(ffn_moe_down-44): 186 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 411 us MUL_MAT_ID(ffn_moe_down-45): 195 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 404 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 430 us MUL_MAT_ID(ffn_moe_down-47): 248 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 374 us MUL_MAT_ID(ffn_moe_down-48): 210 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 362 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 362 us MUL_MAT_ID(ffn_moe_down-50): 187 us ggml_barrier(...): 402 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 495 us MUL_MAT_ID(ffn_moe_down-51): 215 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 406 us MUL_MAT_ID(ffn_moe_down-52): 184 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 398 us MUL_MAT_ID(ffn_moe_down-53): 173 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 413 us MUL_MAT_ID(ffn_moe_down-54): 204 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 388 us MUL_MAT_ID(ffn_moe_down-55): 212 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 409 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 425 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 374 us MUL_MAT_ID(ffn_moe_down-58): 228 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 395 us MUL_MAT_ID(ffn_moe_down-59): 202 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 214 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 374 us MUL_MAT_ID(ffn_moe_down-61): 374 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 391 us MUL_MAT_ID(ffn_moe_down-62): 237 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 449 us MUL_MAT_ID(ffn_moe_down-63): 225 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 382 us MUL_MAT_ID(ffn_moe_down-64): 222 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 428 us MUL_MAT_ID(ffn_moe_down-65): 221 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 425 us MUL_MAT_ID(ffn_moe_down-66): 268 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 444 us MUL_MAT_ID(ffn_moe_down-67): 264 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 347 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 492 us MUL_MAT_ID(ffn_moe_down-69): 219 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 436 us MUL_MAT_ID(ffn_moe_down-70): 181 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 396 us MUL_MAT_ID(ffn_moe_down-71): 373 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 503 us MUL_MAT_ID(ffn_moe_down-72): 184 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 416 us MUL_MAT_ID(ffn_moe_down-73): 217 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 394 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 324 us MUL_MAT_ID(ffn_moe_down-75): 268 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 392 us MUL_MAT_ID(ffn_moe_down-76): 251 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 344 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 417 us MUL_MAT_ID(ffn_moe_down-78): 250 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 474 us MUL_MAT_ID(ffn_moe_down-79): 188 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 497 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 413 us MUL_MAT_ID(ffn_moe_down-81): 222 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 335 us MUL_MAT_ID(ffn_moe_down-82): 306 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 395 us MUL_MAT_ID(ffn_moe_down-83): 170 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 378 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 488 us MUL_MAT_ID(ffn_moe_down-85): 277 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 404 us MUL_MAT_ID(ffn_moe_down-86): 240 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 388 us MUL_MAT_ID(ffn_moe_down-87): 218 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 405 us MUL_MAT_ID(ffn_moe_down-88): 364 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 381 us MUL_MAT_ID(ffn_moe_down-89): 353 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 412 us MUL_MAT_ID(ffn_moe_down-90): 236 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 397 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 13 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 398 us MUL_MAT_ID(ffn_moe_down-25): 196 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 389 us MUL_MAT_ID(ffn_moe_down-26): 422 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 436 us MUL_MAT_ID(ffn_moe_down-27): 291 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 388 us MUL_MAT_ID(ffn_moe_down-28): 217 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 438 us MUL_MAT_ID(ffn_moe_down-29): 304 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 437 us MUL_MAT_ID(ffn_moe_down-30): 325 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 390 us MUL_MAT_ID(ffn_moe_down-31): 299 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 435 us MUL_MAT_ID(ffn_moe_down-32): 255 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 469 us MUL_MAT_ID(ffn_moe_down-33): 296 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 447 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 369 us MUL_MAT_ID(ffn_moe_down-35): 275 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 387 us MUL_MAT_ID(ffn_moe_down-36): 372 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 379 us MUL_MAT_ID(ffn_moe_down-37): 281 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 394 us MUL_MAT_ID(ffn_moe_down-38): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 360 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 363 us MUL_MAT_ID(ffn_moe_down-40): 290 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 374 us MUL_MAT_ID(ffn_moe_down-41): 305 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 446 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 365 us MUL_MAT_ID(ffn_moe_down-43): 150 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 459 us MUL_MAT_ID(ffn_moe_down-44): 297 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 365 us MUL_MAT_ID(ffn_moe_down-45): 224 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 418 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 387 us MUL_MAT_ID(ffn_moe_down-47): 239 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 376 us MUL_MAT_ID(ffn_moe_down-48): 221 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 407 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 372 us MUL_MAT_ID(ffn_moe_down-50): 278 us ggml_barrier(...): 345 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 367 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 399 us MUL_MAT_ID(ffn_moe_down-52): 324 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 391 us MUL_MAT_ID(ffn_moe_down-53): 321 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 355 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 382 us MUL_MAT_ID(ffn_moe_down-55): 232 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 446 us MUL_MAT_ID(ffn_moe_down-56): 395 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 377 us MUL_MAT_ID(ffn_moe_down-57): 366 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 384 us MUL_MAT_ID(ffn_moe_down-58): 224 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 477 us MUL_MAT_ID(ffn_moe_down-59): 234 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 365 us MUL_MAT_ID(ffn_moe_down-60): 483 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 410 us MUL_MAT_ID(ffn_moe_down-61): 175 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 377 us MUL_MAT_ID(ffn_moe_down-62): 209 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 369 us MUL_MAT_ID(ffn_moe_down-63): 253 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 425 us MUL_MAT_ID(ffn_moe_down-64): 202 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 394 us MUL_MAT_ID(ffn_moe_down-65): 204 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 397 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 371 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 406 us MUL_MAT_ID(ffn_moe_down-68): 383 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 408 us MUL_MAT_ID(ffn_moe_down-69): 385 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 455 us MUL_MAT_ID(ffn_moe_down-70): 536 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 496 us MUL_MAT_ID(ffn_moe_down-71): 171 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 466 us MUL_MAT_ID(ffn_moe_down-72): 168 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 501 us MUL_MAT_ID(ffn_moe_down-73): 349 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 423 us MUL_MAT_ID(ffn_moe_down-74): 214 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 466 us MUL_MAT_ID(ffn_moe_down-75): 226 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 531 us MUL_MAT_ID(ffn_moe_down-76): 330 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 399 us MUL_MAT_ID(ffn_moe_down-77): 651 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 438 us MUL_MAT_ID(ffn_moe_down-78): 257 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 418 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 388 us MUL_MAT_ID(ffn_moe_down-80): 296 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 358 us MUL_MAT_ID(ffn_moe_down-81): 185 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 562 us MUL_MAT_ID(ffn_moe_down-82): 212 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 420 us MUL_MAT_ID(ffn_moe_down-83): 236 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 400 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 414 us MUL_MAT_ID(ffn_moe_down-85): 290 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 421 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 466 us MUL_MAT_ID(ffn_moe_down-87): 216 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 385 us MUL_MAT_ID(ffn_moe_down-88): 388 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 521 us MUL_MAT_ID(ffn_moe_down-89): 224 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 416 us MUL_MAT_ID(ffn_moe_down-90): 268 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 387 us MUL_MAT_ID(ffn_moe_down-91): 220 us ggml_barrier(...): 247 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 449 us MUL_MAT_ID(ffn_moe_down-25): 260 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 384 us MUL_MAT_ID(ffn_moe_down-26): 243 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 369 us MUL_MAT_ID(ffn_moe_down-27): 211 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 417 us MUL_MAT_ID(ffn_moe_down-28): 191 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 375 us MUL_MAT_ID(ffn_moe_down-29): 176 us ggml_barrier(...): 361 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 455 us MUL_MAT_ID(ffn_moe_down-30): 178 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 413 us MUL_MAT_ID(ffn_moe_down-31): 183 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 397 us MUL_MAT_ID(ffn_moe_down-32): 209 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 468 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 393 us MUL_MAT_ID(ffn_moe_down-34): 206 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 468 us MUL_MAT_ID(ffn_moe_down-35): 324 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 432 us MUL_MAT_ID(ffn_moe_down-36): 179 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 401 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 391 us MUL_MAT_ID(ffn_moe_down-38): 399 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 460 us MUL_MAT_ID(ffn_moe_down-39): 254 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 402 us MUL_MAT_ID(ffn_moe_down-40): 246 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 517 us MUL_MAT_ID(ffn_moe_down-41): 194 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 389 us MUL_MAT_ID(ffn_moe_down-42): 219 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 442 us MUL_MAT_ID(ffn_moe_down-43): 282 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 377 us MUL_MAT_ID(ffn_moe_down-44): 306 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 383 us MUL_MAT_ID(ffn_moe_down-45): 197 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 367 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 369 us MUL_MAT_ID(ffn_moe_down-47): 184 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 400 us MUL_MAT_ID(ffn_moe_down-48): 201 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 454 us MUL_MAT_ID(ffn_moe_down-49): 212 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 391 us MUL_MAT_ID(ffn_moe_down-50): 419 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 477 us MUL_MAT_ID(ffn_moe_down-51): 172 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 392 us MUL_MAT_ID(ffn_moe_down-52): 391 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 431 us MUL_MAT_ID(ffn_moe_down-53): 287 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 375 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 381 us MUL_MAT_ID(ffn_moe_down-56): 191 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 540 us MUL_MAT_ID(ffn_moe_down-57): 241 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 475 us MUL_MAT_ID(ffn_moe_down-58): 401 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 410 us MUL_MAT_ID(ffn_moe_down-59): 410 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 388 us MUL_MAT_ID(ffn_moe_down-60): 212 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 375 us MUL_MAT_ID(ffn_moe_down-61): 171 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 375 us MUL_MAT_ID(ffn_moe_down-62): 288 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 451 us MUL_MAT_ID(ffn_moe_down-63): 325 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 399 us MUL_MAT_ID(ffn_moe_down-64): 216 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 353 us MUL_MAT_ID(ffn_moe_down-65): 178 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 540 us MUL_MAT_ID(ffn_moe_down-66): 210 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 407 us MUL_MAT_ID(ffn_moe_down-67): 227 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 361 us MUL_MAT_ID(ffn_moe_down-68): 205 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 383 us MUL_MAT_ID(ffn_moe_down-69): 179 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 419 us MUL_MAT_ID(ffn_moe_down-70): 256 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 503 us MUL_MAT_ID(ffn_moe_down-71): 242 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 379 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 386 us MUL_MAT_ID(ffn_moe_down-73): 245 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 403 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 414 us MUL_MAT_ID(ffn_moe_down-75): 221 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 422 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 418 us MUL_MAT_ID(ffn_moe_down-77): 566 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 410 us MUL_MAT_ID(ffn_moe_down-78): 205 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 421 us MUL_MAT_ID(ffn_moe_down-79): 285 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 516 us MUL_MAT_ID(ffn_moe_down-80): 184 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 394 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 367 us MUL_MAT_ID(ffn_moe_down-82): 225 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 518 us MUL_MAT_ID(ffn_moe_down-83): 211 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 379 us MUL_MAT_ID(ffn_moe_down-84): 383 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 368 us MUL_MAT_ID(ffn_moe_down-85): 240 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 404 us MUL_MAT_ID(ffn_moe_down-86): 221 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 382 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 367 us MUL_MAT_ID(ffn_moe_down-88): 430 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 410 us MUL_MAT_ID(ffn_moe_down-89): 301 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 370 us MUL_MAT_ID(ffn_moe_down-90): 334 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 395 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 25 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 246 us ggml_barrier(...): 407 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 433 us MUL_MAT_ID(ffn_moe_down-26): 268 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 367 us MUL_MAT_ID(ffn_moe_down-27): 544 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 423 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 413 us MUL_MAT_ID(ffn_moe_down-29): 223 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 365 us MUL_MAT_ID(ffn_moe_down-30): 211 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 400 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 374 us MUL_MAT_ID(ffn_moe_down-32): 248 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 402 us MUL_MAT_ID(ffn_moe_down-33): 188 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 361 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 408 us MUL_MAT_ID(ffn_moe_down-35): 502 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 347 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 364 us MUL_MAT_ID(ffn_moe_down-37): 248 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 395 us MUL_MAT_ID(ffn_moe_down-38): 205 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 396 us MUL_MAT_ID(ffn_moe_down-39): 231 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 548 us MUL_MAT_ID(ffn_moe_down-40): 245 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 410 us MUL_MAT_ID(ffn_moe_down-41): 207 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 417 us MUL_MAT_ID(ffn_moe_down-42): 189 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 403 us MUL_MAT_ID(ffn_moe_down-43): 277 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 346 us MUL_MAT_ID(ffn_moe_down-44): 275 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 387 us MUL_MAT_ID(ffn_moe_down-45): 617 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 423 us MUL_MAT_ID(ffn_moe_down-46): 354 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 372 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 412 us MUL_MAT_ID(ffn_moe_down-48): 510 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 468 us MUL_MAT_ID(ffn_moe_down-49): 455 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 395 us MUL_MAT_ID(ffn_moe_down-50): 213 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 386 us MUL_MAT_ID(ffn_moe_down-51): 404 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 408 us MUL_MAT_ID(ffn_moe_down-52): 399 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 481 us MUL_MAT_ID(ffn_moe_down-53): 361 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 378 us MUL_MAT_ID(ffn_moe_down-54): 209 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 404 us MUL_MAT_ID(ffn_moe_down-55): 274 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 391 us MUL_MAT_ID(ffn_moe_down-56): 214 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 496 us MUL_MAT_ID(ffn_moe_down-57): 292 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 378 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 405 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 512 us MUL_MAT_ID(ffn_moe_down-60): 213 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 483 us MUL_MAT_ID(ffn_moe_down-61): 251 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 399 us MUL_MAT_ID(ffn_moe_down-62): 216 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 360 us MUL_MAT_ID(ffn_moe_down-63): 194 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 400 us MUL_MAT_ID(ffn_moe_down-64): 213 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 365 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 420 us MUL_MAT_ID(ffn_moe_down-66): 399 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 366 us MUL_MAT_ID(ffn_moe_down-67): 251 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 458 us MUL_MAT_ID(ffn_moe_down-68): 567 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 392 us MUL_MAT_ID(ffn_moe_down-69): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 401 us MUL_MAT_ID(ffn_moe_down-70): 564 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 433 us MUL_MAT_ID(ffn_moe_down-71): 291 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 209 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 401 us MUL_MAT_ID(ffn_moe_down-73): 279 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 476 us MUL_MAT_ID(ffn_moe_down-74): 186 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 395 us MUL_MAT_ID(ffn_moe_down-75): 217 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 355 us MUL_MAT_ID(ffn_moe_down-76): 184 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 415 us MUL_MAT_ID(ffn_moe_down-77): 218 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 392 us MUL_MAT_ID(ffn_moe_down-78): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 371 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 431 us MUL_MAT_ID(ffn_moe_down-80): 269 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 466 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 202 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 355 us MUL_MAT_ID(ffn_moe_down-83): 188 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 429 us MUL_MAT_ID(ffn_moe_down-84): 280 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 370 us MUL_MAT_ID(ffn_moe_down-85): 257 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 463 us MUL_MAT_ID(ffn_moe_down-86): 220 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 415 us MUL_MAT_ID(ffn_moe_down-87): 179 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 390 us MUL_MAT_ID(ffn_moe_down-88): 211 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 375 us MUL_MAT_ID(ffn_moe_down-89): 556 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 394 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 368 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 60 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 387 us MUL_MAT_ID(ffn_moe_down-25): 183 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 369 us MUL_MAT_ID(ffn_moe_down-26): 279 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 379 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 392 us MUL_MAT_ID(ffn_moe_down-28): 195 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 187 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 394 us MUL_MAT_ID(ffn_moe_down-30): 209 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 412 us MUL_MAT_ID(ffn_moe_down-31): 190 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 393 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 375 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 401 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 366 us MUL_MAT_ID(ffn_moe_down-35): 229 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 394 us MUL_MAT_ID(ffn_moe_down-36): 532 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 383 us MUL_MAT_ID(ffn_moe_down-37): 292 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 364 us MUL_MAT_ID(ffn_moe_down-38): 199 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 367 us MUL_MAT_ID(ffn_moe_down-39): 229 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 404 us MUL_MAT_ID(ffn_moe_down-40): 261 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 379 us MUL_MAT_ID(ffn_moe_down-41): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 378 us MUL_MAT_ID(ffn_moe_down-42): 558 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 520 us MUL_MAT_ID(ffn_moe_down-43): 370 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 370 us MUL_MAT_ID(ffn_moe_down-44): 551 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 393 us MUL_MAT_ID(ffn_moe_down-45): 216 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 366 us MUL_MAT_ID(ffn_moe_down-46): 267 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 354 us MUL_MAT_ID(ffn_moe_down-47): 263 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 430 us MUL_MAT_ID(ffn_moe_down-48): 247 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 428 us MUL_MAT_ID(ffn_moe_down-49): 185 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 399 us MUL_MAT_ID(ffn_moe_down-50): 286 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 424 us MUL_MAT_ID(ffn_moe_down-51): 293 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 402 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 379 us MUL_MAT_ID(ffn_moe_down-53): 224 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 368 us MUL_MAT_ID(ffn_moe_down-54): 208 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 378 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 483 us MUL_MAT_ID(ffn_moe_down-56): 205 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 435 us MUL_MAT_ID(ffn_moe_down-57): 201 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 395 us MUL_MAT_ID(ffn_moe_down-58): 346 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 367 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 526 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 420 us MUL_MAT_ID(ffn_moe_down-61): 207 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 380 us MUL_MAT_ID(ffn_moe_down-62): 237 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 390 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 449 us MUL_MAT_ID(ffn_moe_down-64): 159 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 433 us MUL_MAT_ID(ffn_moe_down-65): 297 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 392 us MUL_MAT_ID(ffn_moe_down-66): 253 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 369 us MUL_MAT_ID(ffn_moe_down-67): 206 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 229 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 477 us MUL_MAT_ID(ffn_moe_down-69): 314 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 412 us MUL_MAT_ID(ffn_moe_down-70): 207 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 408 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 413 us MUL_MAT_ID(ffn_moe_down-72): 238 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 391 us MUL_MAT_ID(ffn_moe_down-73): 234 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 346 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 443 us MUL_MAT_ID(ffn_moe_down-75): 287 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 451 us MUL_MAT_ID(ffn_moe_down-76): 204 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 423 us MUL_MAT_ID(ffn_moe_down-77): 253 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 401 us MUL_MAT_ID(ffn_moe_down-78): 431 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 374 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 336 us MUL_MAT_ID(ffn_moe_down-80): 316 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 405 us MUL_MAT_ID(ffn_moe_down-81): 244 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 420 us MUL_MAT_ID(ffn_moe_down-82): 261 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 388 us MUL_MAT_ID(ffn_moe_down-83): 184 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 417 us MUL_MAT_ID(ffn_moe_down-84): 257 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 377 us MUL_MAT_ID(ffn_moe_down-85): 360 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 346 us MUL_MAT_ID(ffn_moe_down-86): 185 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 436 us MUL_MAT_ID(ffn_moe_down-87): 176 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 377 us MUL_MAT_ID(ffn_moe_down-88): 257 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 373 us MUL_MAT_ID(ffn_moe_down-89): 213 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 372 us MUL_MAT_ID(ffn_moe_down-90): 223 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 398 us MUL_MAT_ID(ffn_moe_down-91): 195 us ggml_barrier(...): 371 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 391 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 357 us MUL_MAT_ID(ffn_moe_down-26): 197 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 387 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 419 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 361 us MUL_MAT_ID(ffn_moe_down-29): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 366 us MUL_MAT_ID(ffn_moe_down-30): 521 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 386 us MUL_MAT_ID(ffn_moe_down-31): 185 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 363 us MUL_MAT_ID(ffn_moe_down-32): 474 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 426 us MUL_MAT_ID(ffn_moe_down-33): 217 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 521 us MUL_MAT_ID(ffn_moe_down-34): 256 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 425 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 448 us MUL_MAT_ID(ffn_moe_down-36): 549 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 421 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 498 us MUL_MAT_ID(ffn_moe_down-38): 233 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 387 us MUL_MAT_ID(ffn_moe_down-39): 242 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 388 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 399 us MUL_MAT_ID(ffn_moe_down-41): 217 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 361 us MUL_MAT_ID(ffn_moe_down-42): 186 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 379 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 366 us MUL_MAT_ID(ffn_moe_down-44): 509 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 383 us MUL_MAT_ID(ffn_moe_down-45): 218 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 390 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 398 us MUL_MAT_ID(ffn_moe_down-47): 326 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 396 us MUL_MAT_ID(ffn_moe_down-48): 198 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 372 us MUL_MAT_ID(ffn_moe_down-49): 174 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 344 us MUL_MAT_ID(ffn_moe_down-50): 187 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 370 us MUL_MAT_ID(ffn_moe_down-51): 512 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 390 us MUL_MAT_ID(ffn_moe_down-52): 185 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 378 us MUL_MAT_ID(ffn_moe_down-53): 210 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 356 us MUL_MAT_ID(ffn_moe_down-54): 181 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 412 us MUL_MAT_ID(ffn_moe_down-55): 553 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 378 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 364 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 398 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 344 us MUL_MAT_ID(ffn_moe_down-59): 361 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 465 us MUL_MAT_ID(ffn_moe_down-60): 271 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 423 us MUL_MAT_ID(ffn_moe_down-61): 270 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 358 us MUL_MAT_ID(ffn_moe_down-62): 210 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 390 us MUL_MAT_ID(ffn_moe_down-63): 222 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 395 us MUL_MAT_ID(ffn_moe_down-64): 202 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 379 us MUL_MAT_ID(ffn_moe_down-65): 200 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 376 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 476 us MUL_MAT_ID(ffn_moe_down-67): 215 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 387 us MUL_MAT_ID(ffn_moe_down-68): 224 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 388 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 384 us MUL_MAT_ID(ffn_moe_down-70): 206 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 429 us MUL_MAT_ID(ffn_moe_down-71): 214 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 355 us MUL_MAT_ID(ffn_moe_down-72): 201 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 409 us MUL_MAT_ID(ffn_moe_down-73): 291 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 383 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 392 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 431 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 397 us MUL_MAT_ID(ffn_moe_down-77): 251 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 353 us MUL_MAT_ID(ffn_moe_down-78): 191 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 380 us MUL_MAT_ID(ffn_moe_down-79): 280 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 354 us MUL_MAT_ID(ffn_moe_down-80): 169 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 367 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 362 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 355 us MUL_MAT_ID(ffn_moe_down-83): 365 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 426 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 392 us MUL_MAT_ID(ffn_moe_down-85): 249 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 400 us MUL_MAT_ID(ffn_moe_down-86): 222 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 413 us MUL_MAT_ID(ffn_moe_down-87): 206 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 374 us MUL_MAT_ID(ffn_moe_down-88): 193 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 372 us MUL_MAT_ID(ffn_moe_down-89): 186 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 395 us MUL_MAT_ID(ffn_moe_down-90): 222 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 376 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 39 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 382 us MUL_MAT_ID(ffn_moe_down-25): 301 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 382 us MUL_MAT_ID(ffn_moe_down-26): 369 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 424 us MUL_MAT_ID(ffn_moe_down-27): 178 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 372 us MUL_MAT_ID(ffn_moe_down-28): 353 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 519 us MUL_MAT_ID(ffn_moe_down-29): 220 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 417 us MUL_MAT_ID(ffn_moe_down-30): 225 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 486 us MUL_MAT_ID(ffn_moe_down-31): 457 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 384 us MUL_MAT_ID(ffn_moe_down-32): 211 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 371 us MUL_MAT_ID(ffn_moe_down-33): 529 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 372 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 370 us MUL_MAT_ID(ffn_moe_down-35): 222 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 373 us MUL_MAT_ID(ffn_moe_down-36): 302 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 377 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 385 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 475 us MUL_MAT_ID(ffn_moe_down-39): 257 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 378 us MUL_MAT_ID(ffn_moe_down-40): 663 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 518 us MUL_MAT_ID(ffn_moe_down-41): 229 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 384 us MUL_MAT_ID(ffn_moe_down-42): 236 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 365 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 348 us MUL_MAT_ID(ffn_moe_down-44): 194 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 440 us MUL_MAT_ID(ffn_moe_down-45): 395 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 357 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 381 us MUL_MAT_ID(ffn_moe_down-47): 525 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 398 us MUL_MAT_ID(ffn_moe_down-48): 290 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 422 us MUL_MAT_ID(ffn_moe_down-49): 183 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 439 us MUL_MAT_ID(ffn_moe_down-50): 276 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 366 us MUL_MAT_ID(ffn_moe_down-51): 297 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 394 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 398 us MUL_MAT_ID(ffn_moe_down-53): 241 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 368 us MUL_MAT_ID(ffn_moe_down-54): 537 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 371 us MUL_MAT_ID(ffn_moe_down-55): 211 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 386 us MUL_MAT_ID(ffn_moe_down-56): 209 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 422 us MUL_MAT_ID(ffn_moe_down-57): 395 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 407 us MUL_MAT_ID(ffn_moe_down-58): 188 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 432 us MUL_MAT_ID(ffn_moe_down-59): 250 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 390 us MUL_MAT_ID(ffn_moe_down-60): 304 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 382 us MUL_MAT_ID(ffn_moe_down-61): 218 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 365 us MUL_MAT_ID(ffn_moe_down-62): 208 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 379 us MUL_MAT_ID(ffn_moe_down-63): 237 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 465 us MUL_MAT_ID(ffn_moe_down-64): 182 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 398 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 366 us MUL_MAT_ID(ffn_moe_down-66): 384 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 401 us MUL_MAT_ID(ffn_moe_down-67): 309 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 373 us MUL_MAT_ID(ffn_moe_down-68): 200 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 406 us MUL_MAT_ID(ffn_moe_down-69): 501 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 367 us MUL_MAT_ID(ffn_moe_down-70): 243 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 432 us MUL_MAT_ID(ffn_moe_down-71): 228 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 420 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 394 us MUL_MAT_ID(ffn_moe_down-73): 411 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 390 us MUL_MAT_ID(ffn_moe_down-74): 247 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 365 us MUL_MAT_ID(ffn_moe_down-75): 187 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 424 us MUL_MAT_ID(ffn_moe_down-76): 301 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 380 us MUL_MAT_ID(ffn_moe_down-77): 200 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 419 us MUL_MAT_ID(ffn_moe_down-78): 219 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 426 us MUL_MAT_ID(ffn_moe_down-79): 265 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 490 us MUL_MAT_ID(ffn_moe_down-80): 187 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 428 us MUL_MAT_ID(ffn_moe_down-81): 376 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 486 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 361 us MUL_MAT_ID(ffn_moe_down-83): 208 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 371 us MUL_MAT_ID(ffn_moe_down-84): 192 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 382 us MUL_MAT_ID(ffn_moe_down-85): 239 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 505 us MUL_MAT_ID(ffn_moe_down-86): 428 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 419 us MUL_MAT_ID(ffn_moe_down-87): 270 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 390 us MUL_MAT_ID(ffn_moe_down-88): 207 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 399 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 399 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 404 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 19 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 168 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 388 us MUL_MAT_ID(ffn_moe_down-26): 181 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 380 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 460 us MUL_MAT_ID(ffn_moe_down-28): 276 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 370 us MUL_MAT_ID(ffn_moe_down-29): 224 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 463 us MUL_MAT_ID(ffn_moe_down-30): 277 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 381 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 469 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 391 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 407 us MUL_MAT_ID(ffn_moe_down-34): 190 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 370 us MUL_MAT_ID(ffn_moe_down-35): 305 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 476 us MUL_MAT_ID(ffn_moe_down-36): 239 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 431 us MUL_MAT_ID(ffn_moe_down-37): 263 us ggml_barrier(...): 344 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 420 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 356 us MUL_MAT_ID(ffn_moe_down-39): 349 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 508 us MUL_MAT_ID(ffn_moe_down-40): 335 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 447 us MUL_MAT_ID(ffn_moe_down-41): 220 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 412 us MUL_MAT_ID(ffn_moe_down-42): 184 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 449 us MUL_MAT_ID(ffn_moe_down-43): 183 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 350 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 389 us MUL_MAT_ID(ffn_moe_down-45): 182 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 400 us MUL_MAT_ID(ffn_moe_down-46): 252 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 367 us MUL_MAT_ID(ffn_moe_down-47): 172 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 394 us MUL_MAT_ID(ffn_moe_down-48): 527 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 413 us MUL_MAT_ID(ffn_moe_down-49): 348 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 368 us MUL_MAT_ID(ffn_moe_down-50): 166 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 393 us MUL_MAT_ID(ffn_moe_down-51): 323 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 404 us MUL_MAT_ID(ffn_moe_down-52): 199 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 455 us MUL_MAT_ID(ffn_moe_down-53): 281 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 393 us MUL_MAT_ID(ffn_moe_down-54): 270 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 371 us MUL_MAT_ID(ffn_moe_down-55): 184 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 447 us MUL_MAT_ID(ffn_moe_down-56): 203 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 499 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 496 us MUL_MAT_ID(ffn_moe_down-58): 258 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 369 us MUL_MAT_ID(ffn_moe_down-59): 503 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 368 us MUL_MAT_ID(ffn_moe_down-60): 175 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 479 us MUL_MAT_ID(ffn_moe_down-61): 233 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 420 us MUL_MAT_ID(ffn_moe_down-62): 183 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 428 us MUL_MAT_ID(ffn_moe_down-63): 222 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 404 us MUL_MAT_ID(ffn_moe_down-64): 241 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 428 us MUL_MAT_ID(ffn_moe_down-65): 286 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 377 us MUL_MAT_ID(ffn_moe_down-66): 170 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 389 us MUL_MAT_ID(ffn_moe_down-67): 270 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 561 us MUL_MAT_ID(ffn_moe_down-68): 190 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 362 us MUL_MAT_ID(ffn_moe_down-69): 232 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 409 us MUL_MAT_ID(ffn_moe_down-70): 658 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 375 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 426 us MUL_MAT_ID(ffn_moe_down-72): 241 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 427 us MUL_MAT_ID(ffn_moe_down-73): 201 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 435 us MUL_MAT_ID(ffn_moe_down-74): 412 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 416 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 428 us MUL_MAT_ID(ffn_moe_down-76): 230 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 387 us MUL_MAT_ID(ffn_moe_down-77): 171 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 433 us MUL_MAT_ID(ffn_moe_down-78): 186 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 390 us MUL_MAT_ID(ffn_moe_down-79): 218 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 387 us MUL_MAT_ID(ffn_moe_down-80): 243 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 425 us MUL_MAT_ID(ffn_moe_down-81): 462 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 405 us MUL_MAT_ID(ffn_moe_down-82): 256 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 457 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 494 us MUL_MAT_ID(ffn_moe_down-84): 246 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 390 us MUL_MAT_ID(ffn_moe_down-85): 681 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 507 us MUL_MAT_ID(ffn_moe_down-86): 275 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 391 us MUL_MAT_ID(ffn_moe_down-87): 184 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 481 us MUL_MAT_ID(ffn_moe_down-88): 199 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 510 us MUL_MAT_ID(ffn_moe_down-89): 226 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 404 us MUL_MAT_ID(ffn_moe_down-90): 261 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 374 us MUL_MAT_ID(ffn_moe_down-91): 199 us ggml_barrier(...): 312 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 510 us MUL_MAT_ID(ffn_moe_down-25): 165 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 378 us MUL_MAT_ID(ffn_moe_down-26): 197 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 399 us MUL_MAT_ID(ffn_moe_down-27): 262 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 361 us MUL_MAT_ID(ffn_moe_down-28): 207 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 367 us MUL_MAT_ID(ffn_moe_down-29): 208 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 513 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 499 us MUL_MAT_ID(ffn_moe_down-31): 241 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 405 us MUL_MAT_ID(ffn_moe_down-32): 385 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 393 us MUL_MAT_ID(ffn_moe_down-33): 400 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 404 us MUL_MAT_ID(ffn_moe_down-34): 189 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 354 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 363 us MUL_MAT_ID(ffn_moe_down-36): 202 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 439 us MUL_MAT_ID(ffn_moe_down-37): 205 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 452 us MUL_MAT_ID(ffn_moe_down-38): 321 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 468 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 458 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 455 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 430 us MUL_MAT_ID(ffn_moe_down-42): 188 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 451 us MUL_MAT_ID(ffn_moe_down-43): 288 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 418 us MUL_MAT_ID(ffn_moe_down-44): 267 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 394 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 417 us MUL_MAT_ID(ffn_moe_down-46): 183 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 492 us MUL_MAT_ID(ffn_moe_down-47): 184 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 509 us MUL_MAT_ID(ffn_moe_down-48): 312 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 390 us MUL_MAT_ID(ffn_moe_down-49): 173 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 479 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 454 us MUL_MAT_ID(ffn_moe_down-51): 211 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 362 us MUL_MAT_ID(ffn_moe_down-52): 245 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 426 us MUL_MAT_ID(ffn_moe_down-53): 215 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 484 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 400 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 388 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 393 us MUL_MAT_ID(ffn_moe_down-57): 218 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 446 us MUL_MAT_ID(ffn_moe_down-58): 246 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 404 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 400 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 354 us MUL_MAT_ID(ffn_moe_down-61): 340 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 415 us MUL_MAT_ID(ffn_moe_down-62): 189 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 397 us MUL_MAT_ID(ffn_moe_down-63): 277 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 390 us MUL_MAT_ID(ffn_moe_down-64): 238 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 349 us MUL_MAT_ID(ffn_moe_down-65): 297 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 365 us MUL_MAT_ID(ffn_moe_down-66): 314 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 415 us MUL_MAT_ID(ffn_moe_down-67): 270 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 475 us MUL_MAT_ID(ffn_moe_down-68): 274 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 422 us MUL_MAT_ID(ffn_moe_down-69): 304 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 462 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 388 us MUL_MAT_ID(ffn_moe_down-71): 202 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 373 us MUL_MAT_ID(ffn_moe_down-73): 208 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 370 us MUL_MAT_ID(ffn_moe_down-74): 232 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 369 us MUL_MAT_ID(ffn_moe_down-75): 507 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 406 us MUL_MAT_ID(ffn_moe_down-76): 260 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 352 us MUL_MAT_ID(ffn_moe_down-77): 185 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 430 us MUL_MAT_ID(ffn_moe_down-78): 171 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 483 us MUL_MAT_ID(ffn_moe_down-79): 291 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 405 us MUL_MAT_ID(ffn_moe_down-80): 206 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 384 us MUL_MAT_ID(ffn_moe_down-81): 297 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 381 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 379 us MUL_MAT_ID(ffn_moe_down-83): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 416 us MUL_MAT_ID(ffn_moe_down-84): 295 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 377 us MUL_MAT_ID(ffn_moe_down-85): 242 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 493 us MUL_MAT_ID(ffn_moe_down-86): 226 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 493 us MUL_MAT_ID(ffn_moe_down-87): 218 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 365 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 443 us MUL_MAT_ID(ffn_moe_down-89): 287 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 502 us MUL_MAT_ID(ffn_moe_down-90): 253 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 415 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 20 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 413 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 444 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 363 us MUL_MAT_ID(ffn_moe_down-27): 240 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 182 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 401 us MUL_MAT_ID(ffn_moe_down-30): 167 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 579 us MUL_MAT_ID(ffn_moe_down-31): 214 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 422 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 413 us MUL_MAT_ID(ffn_moe_down-33): 202 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 395 us MUL_MAT_ID(ffn_moe_down-34): 188 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 395 us MUL_MAT_ID(ffn_moe_down-35): 183 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 425 us MUL_MAT_ID(ffn_moe_down-36): 190 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 402 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 394 us MUL_MAT_ID(ffn_moe_down-38): 217 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 513 us MUL_MAT_ID(ffn_moe_down-39): 192 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 508 us MUL_MAT_ID(ffn_moe_down-40): 221 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 516 us MUL_MAT_ID(ffn_moe_down-41): 226 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 448 us MUL_MAT_ID(ffn_moe_down-42): 266 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 367 us MUL_MAT_ID(ffn_moe_down-43): 234 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 462 us MUL_MAT_ID(ffn_moe_down-44): 175 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 408 us MUL_MAT_ID(ffn_moe_down-45): 197 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 387 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 377 us MUL_MAT_ID(ffn_moe_down-47): 280 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 500 us MUL_MAT_ID(ffn_moe_down-48): 183 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 390 us MUL_MAT_ID(ffn_moe_down-49): 401 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 368 us MUL_MAT_ID(ffn_moe_down-50): 245 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 404 us MUL_MAT_ID(ffn_moe_down-51): 550 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 395 us MUL_MAT_ID(ffn_moe_down-52): 188 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 530 us MUL_MAT_ID(ffn_moe_down-53): 355 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 438 us MUL_MAT_ID(ffn_moe_down-54): 273 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 404 us MUL_MAT_ID(ffn_moe_down-55): 278 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 501 us MUL_MAT_ID(ffn_moe_down-56): 306 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 368 us MUL_MAT_ID(ffn_moe_down-57): 185 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 492 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 415 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 379 us MUL_MAT_ID(ffn_moe_down-60): 169 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 503 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 395 us MUL_MAT_ID(ffn_moe_down-62): 247 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 435 us MUL_MAT_ID(ffn_moe_down-63): 249 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 523 us MUL_MAT_ID(ffn_moe_down-64): 563 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 377 us MUL_MAT_ID(ffn_moe_down-65): 205 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 427 us MUL_MAT_ID(ffn_moe_down-66): 235 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 454 us MUL_MAT_ID(ffn_moe_down-67): 296 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 377 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 392 us MUL_MAT_ID(ffn_moe_down-69): 204 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 384 us MUL_MAT_ID(ffn_moe_down-70): 198 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 379 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 363 us MUL_MAT_ID(ffn_moe_down-72): 205 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 378 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 390 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 375 us MUL_MAT_ID(ffn_moe_down-75): 200 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 385 us MUL_MAT_ID(ffn_moe_down-76): 200 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 385 us MUL_MAT_ID(ffn_moe_down-77): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 451 us MUL_MAT_ID(ffn_moe_down-78): 187 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 403 us MUL_MAT_ID(ffn_moe_down-79): 186 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 359 us MUL_MAT_ID(ffn_moe_down-80): 170 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 499 us MUL_MAT_ID(ffn_moe_down-81): 179 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 358 us MUL_MAT_ID(ffn_moe_down-82): 314 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 384 us MUL_MAT_ID(ffn_moe_down-83): 253 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 363 us MUL_MAT_ID(ffn_moe_down-84): 286 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 359 us MUL_MAT_ID(ffn_moe_down-85): 234 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 392 us MUL_MAT_ID(ffn_moe_down-86): 481 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 428 us MUL_MAT_ID(ffn_moe_down-87): 253 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 401 us MUL_MAT_ID(ffn_moe_down-88): 247 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 390 us MUL_MAT_ID(ffn_moe_down-89): 440 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 486 us MUL_MAT_ID(ffn_moe_down-90): 279 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 371 us MUL_MAT_ID(ffn_moe_down-91): 233 us ggml_barrier(...): 51 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 536 us MUL_MAT_ID(ffn_moe_down-25): 275 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 435 us MUL_MAT_ID(ffn_moe_down-26): 313 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 403 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 469 us MUL_MAT_ID(ffn_moe_down-28): 195 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 436 us MUL_MAT_ID(ffn_moe_down-29): 248 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 411 us MUL_MAT_ID(ffn_moe_down-30): 254 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 394 us MUL_MAT_ID(ffn_moe_down-31): 284 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 458 us MUL_MAT_ID(ffn_moe_down-32): 223 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 406 us MUL_MAT_ID(ffn_moe_down-33): 222 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 412 us MUL_MAT_ID(ffn_moe_down-34): 474 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 406 us MUL_MAT_ID(ffn_moe_down-35): 278 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 411 us MUL_MAT_ID(ffn_moe_down-36): 313 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 484 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 247 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 413 us MUL_MAT_ID(ffn_moe_down-39): 316 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 433 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 448 us MUL_MAT_ID(ffn_moe_down-41): 302 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 480 us MUL_MAT_ID(ffn_moe_down-42): 271 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 399 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 385 us MUL_MAT_ID(ffn_moe_down-44): 221 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 389 us MUL_MAT_ID(ffn_moe_down-45): 227 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 366 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 405 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 387 us MUL_MAT_ID(ffn_moe_down-47): 370 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 385 us MUL_MAT_ID(ffn_moe_down-48): 262 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 423 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 392 us MUL_MAT_ID(ffn_moe_down-50): 425 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 366 us MUL_MAT_ID(ffn_moe_down-51): 188 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 368 us MUL_MAT_ID(ffn_moe_down-52): 380 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 405 us MUL_MAT_ID(ffn_moe_down-53): 224 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 409 us MUL_MAT_ID(ffn_moe_down-54): 192 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 419 us MUL_MAT_ID(ffn_moe_down-55): 208 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 450 us MUL_MAT_ID(ffn_moe_down-56): 177 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 349 us MUL_MAT_ID(ffn_moe_down-57): 177 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 393 us MUL_MAT_ID(ffn_moe_down-58): 166 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 492 us MUL_MAT_ID(ffn_moe_down-59): 212 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 522 us MUL_MAT_ID(ffn_moe_down-60): 227 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 418 us MUL_MAT_ID(ffn_moe_down-61): 234 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 444 us MUL_MAT_ID(ffn_moe_down-62): 298 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 381 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 429 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 376 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 408 us MUL_MAT_ID(ffn_moe_down-66): 408 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 478 us MUL_MAT_ID(ffn_moe_down-67): 214 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 427 us MUL_MAT_ID(ffn_moe_down-68): 192 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 416 us MUL_MAT_ID(ffn_moe_down-69): 217 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 450 us MUL_MAT_ID(ffn_moe_down-70): 198 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 417 us MUL_MAT_ID(ffn_moe_down-71): 267 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 374 us MUL_MAT_ID(ffn_moe_down-72): 248 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 384 us MUL_MAT_ID(ffn_moe_down-73): 299 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 358 us MUL_MAT_ID(ffn_moe_down-74): 309 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 365 us MUL_MAT_ID(ffn_moe_down-75): 246 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 413 us MUL_MAT_ID(ffn_moe_down-76): 267 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 480 us MUL_MAT_ID(ffn_moe_down-77): 225 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 404 us MUL_MAT_ID(ffn_moe_down-78): 399 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 396 us MUL_MAT_ID(ffn_moe_down-79): 178 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 359 us MUL_MAT_ID(ffn_moe_down-80): 203 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 395 us MUL_MAT_ID(ffn_moe_down-81): 302 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 424 us MUL_MAT_ID(ffn_moe_down-82): 190 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 370 us MUL_MAT_ID(ffn_moe_down-83): 174 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 367 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 423 us MUL_MAT_ID(ffn_moe_down-85): 240 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 384 us MUL_MAT_ID(ffn_moe_down-86): 221 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 397 us MUL_MAT_ID(ffn_moe_down-88): 281 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 389 us MUL_MAT_ID(ffn_moe_down-89): 211 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 351 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 402 us MUL_MAT_ID(ffn_moe_down-91): 288 us ggml_barrier(...): 19 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 380 us MUL_MAT_ID(ffn_moe_down-25): 183 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 399 us MUL_MAT_ID(ffn_moe_down-26): 191 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 431 us MUL_MAT_ID(ffn_moe_down-27): 226 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 406 us MUL_MAT_ID(ffn_moe_down-28): 377 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 423 us MUL_MAT_ID(ffn_moe_down-29): 204 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 383 us MUL_MAT_ID(ffn_moe_down-30): 212 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 363 us MUL_MAT_ID(ffn_moe_down-31): 243 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 405 us MUL_MAT_ID(ffn_moe_down-32): 185 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 394 us MUL_MAT_ID(ffn_moe_down-33): 212 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 433 us MUL_MAT_ID(ffn_moe_down-34): 168 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 380 us MUL_MAT_ID(ffn_moe_down-35): 215 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 366 us MUL_MAT_ID(ffn_moe_down-36): 189 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 382 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 391 us MUL_MAT_ID(ffn_moe_down-38): 535 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 359 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 373 us MUL_MAT_ID(ffn_moe_down-40): 270 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 366 us MUL_MAT_ID(ffn_moe_down-41): 206 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 362 us MUL_MAT_ID(ffn_moe_down-42): 201 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 393 us MUL_MAT_ID(ffn_moe_down-43): 240 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 393 us MUL_MAT_ID(ffn_moe_down-44): 187 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 367 us MUL_MAT_ID(ffn_moe_down-45): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 393 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 375 us MUL_MAT_ID(ffn_moe_down-47): 188 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 341 us MUL_MAT_ID(ffn_moe_down-48): 345 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 372 us MUL_MAT_ID(ffn_moe_down-49): 273 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 386 us MUL_MAT_ID(ffn_moe_down-50): 183 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 393 us MUL_MAT_ID(ffn_moe_down-51): 388 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 742 us MUL_MAT_ID(ffn_moe_down-52): 327 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 458 us MUL_MAT_ID(ffn_moe_down-53): 287 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 369 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 408 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 541 us MUL_MAT_ID(ffn_moe_down-56): 175 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 362 us MUL_MAT_ID(ffn_moe_down-57): 183 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 424 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 394 us MUL_MAT_ID(ffn_moe_down-59): 419 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 367 us MUL_MAT_ID(ffn_moe_down-60): 212 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 352 us MUL_MAT_ID(ffn_moe_down-61): 555 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 361 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 511 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 363 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 365 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 185 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 378 us MUL_MAT_ID(ffn_moe_down-67): 205 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 423 us MUL_MAT_ID(ffn_moe_down-68): 386 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 382 us MUL_MAT_ID(ffn_moe_down-69): 227 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 444 us MUL_MAT_ID(ffn_moe_down-70): 183 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 388 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 339 us MUL_MAT_ID(ffn_moe_down-72): 186 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 377 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 401 us MUL_MAT_ID(ffn_moe_down-74): 288 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 415 us MUL_MAT_ID(ffn_moe_down-75): 290 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 395 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 390 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 404 us MUL_MAT_ID(ffn_moe_down-78): 399 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 497 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 415 us MUL_MAT_ID(ffn_moe_down-80): 273 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 364 us MUL_MAT_ID(ffn_moe_down-81): 355 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 410 us MUL_MAT_ID(ffn_moe_down-82): 199 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 389 us MUL_MAT_ID(ffn_moe_down-83): 167 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 391 us MUL_MAT_ID(ffn_moe_down-84): 273 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 464 us MUL_MAT_ID(ffn_moe_down-85): 203 us ggml_barrier(...): 376 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 391 us MUL_MAT_ID(ffn_moe_down-86): 208 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 477 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 515 us MUL_MAT_ID(ffn_moe_down-88): 199 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 360 us MUL_MAT_ID(ffn_moe_down-89): 429 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 374 us MUL_MAT_ID(ffn_moe_down-90): 248 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 404 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 71 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 360 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 388 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 530 us MUL_MAT_ID(ffn_moe_down-27): 189 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 384 us MUL_MAT_ID(ffn_moe_down-28): 191 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 432 us MUL_MAT_ID(ffn_moe_down-29): 179 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 369 us MUL_MAT_ID(ffn_moe_down-30): 188 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 405 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 383 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 395 us MUL_MAT_ID(ffn_moe_down-33): 296 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 385 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 397 us MUL_MAT_ID(ffn_moe_down-35): 392 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 385 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 397 us MUL_MAT_ID(ffn_moe_down-38): 286 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 392 us MUL_MAT_ID(ffn_moe_down-39): 422 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 418 us MUL_MAT_ID(ffn_moe_down-40): 202 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 359 us MUL_MAT_ID(ffn_moe_down-41): 205 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 389 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 392 us MUL_MAT_ID(ffn_moe_down-43): 192 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 405 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 378 us MUL_MAT_ID(ffn_moe_down-45): 190 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 396 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 350 us MUL_MAT_ID(ffn_moe_down-47): 209 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 388 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 405 us MUL_MAT_ID(ffn_moe_down-49): 312 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 367 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 471 us MUL_MAT_ID(ffn_moe_down-51): 206 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 394 us MUL_MAT_ID(ffn_moe_down-52): 197 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 371 us MUL_MAT_ID(ffn_moe_down-53): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 388 us MUL_MAT_ID(ffn_moe_down-54): 216 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 409 us MUL_MAT_ID(ffn_moe_down-55): 384 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 404 us MUL_MAT_ID(ffn_moe_down-56): 246 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 368 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 435 us MUL_MAT_ID(ffn_moe_down-58): 253 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 383 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 384 us MUL_MAT_ID(ffn_moe_down-60): 211 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 367 us MUL_MAT_ID(ffn_moe_down-61): 341 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 373 us MUL_MAT_ID(ffn_moe_down-62): 205 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 394 us MUL_MAT_ID(ffn_moe_down-63): 219 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 403 us MUL_MAT_ID(ffn_moe_down-64): 187 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 383 us MUL_MAT_ID(ffn_moe_down-65): 200 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 212 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 416 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 423 us MUL_MAT_ID(ffn_moe_down-68): 380 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 357 us MUL_MAT_ID(ffn_moe_down-69): 222 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 399 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 403 us MUL_MAT_ID(ffn_moe_down-71): 223 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 374 us MUL_MAT_ID(ffn_moe_down-72): 243 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 384 us MUL_MAT_ID(ffn_moe_down-73): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 366 us MUL_MAT_ID(ffn_moe_down-74): 416 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 382 us MUL_MAT_ID(ffn_moe_down-75): 422 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 396 us MUL_MAT_ID(ffn_moe_down-76): 185 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 364 us MUL_MAT_ID(ffn_moe_down-77): 431 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 403 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 446 us MUL_MAT_ID(ffn_moe_down-79): 227 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 369 us MUL_MAT_ID(ffn_moe_down-80): 394 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 370 us MUL_MAT_ID(ffn_moe_down-81): 199 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 413 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 357 us MUL_MAT_ID(ffn_moe_down-83): 544 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 367 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 359 us MUL_MAT_ID(ffn_moe_down-85): 730 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 354 us MUL_MAT_ID(ffn_moe_down-86): 218 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 398 us MUL_MAT_ID(ffn_moe_down-87): 183 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 381 us MUL_MAT_ID(ffn_moe_down-88): 196 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 346 us MUL_MAT_ID(ffn_moe_down-89): 216 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 355 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 391 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 22 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 428 us MUL_MAT_ID(ffn_moe_down-25): 203 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 438 us MUL_MAT_ID(ffn_moe_down-26): 257 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 499 us MUL_MAT_ID(ffn_moe_down-27): 222 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 427 us MUL_MAT_ID(ffn_moe_down-28): 161 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 486 us MUL_MAT_ID(ffn_moe_down-29): 472 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 523 us MUL_MAT_ID(ffn_moe_down-30): 185 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 389 us MUL_MAT_ID(ffn_moe_down-31): 392 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 406 us MUL_MAT_ID(ffn_moe_down-32): 488 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 220 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 543 us MUL_MAT_ID(ffn_moe_down-34): 278 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 434 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 455 us MUL_MAT_ID(ffn_moe_down-36): 237 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 450 us MUL_MAT_ID(ffn_moe_down-37): 215 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 447 us MUL_MAT_ID(ffn_moe_down-38): 165 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 398 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 473 us MUL_MAT_ID(ffn_moe_down-40): 264 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 400 us MUL_MAT_ID(ffn_moe_down-41): 220 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 406 us MUL_MAT_ID(ffn_moe_down-42): 193 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 482 us MUL_MAT_ID(ffn_moe_down-43): 560 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 485 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 506 us MUL_MAT_ID(ffn_moe_down-45): 191 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 379 us MUL_MAT_ID(ffn_moe_down-46): 169 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 379 us MUL_MAT_ID(ffn_moe_down-47): 216 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 367 us MUL_MAT_ID(ffn_moe_down-48): 235 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 395 us MUL_MAT_ID(ffn_moe_down-49): 187 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 210 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 447 us MUL_MAT_ID(ffn_moe_down-51): 251 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 402 us MUL_MAT_ID(ffn_moe_down-52): 210 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 356 us MUL_MAT_ID(ffn_moe_down-53): 247 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 378 us MUL_MAT_ID(ffn_moe_down-54): 267 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 393 us MUL_MAT_ID(ffn_moe_down-55): 214 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 376 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 373 us MUL_MAT_ID(ffn_moe_down-57): 180 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 407 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 379 us MUL_MAT_ID(ffn_moe_down-59): 183 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 184 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 351 us MUL_MAT_ID(ffn_moe_down-61): 186 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 424 us MUL_MAT_ID(ffn_moe_down-62): 269 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 358 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 407 us MUL_MAT_ID(ffn_moe_down-64): 205 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 478 us MUL_MAT_ID(ffn_moe_down-65): 203 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 457 us MUL_MAT_ID(ffn_moe_down-66): 269 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 375 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 486 us MUL_MAT_ID(ffn_moe_down-68): 218 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 411 us MUL_MAT_ID(ffn_moe_down-69): 188 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 383 us MUL_MAT_ID(ffn_moe_down-70): 270 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 461 us MUL_MAT_ID(ffn_moe_down-71): 301 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 387 us MUL_MAT_ID(ffn_moe_down-72): 348 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 508 us MUL_MAT_ID(ffn_moe_down-73): 173 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 353 us MUL_MAT_ID(ffn_moe_down-74): 219 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 350 us MUL_MAT_ID(ffn_moe_down-75): 213 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 350 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 396 us MUL_MAT_ID(ffn_moe_down-77): 162 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 476 us MUL_MAT_ID(ffn_moe_down-78): 173 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 397 us MUL_MAT_ID(ffn_moe_down-79): 205 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 391 us MUL_MAT_ID(ffn_moe_down-80): 357 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 373 us MUL_MAT_ID(ffn_moe_down-81): 188 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 483 us MUL_MAT_ID(ffn_moe_down-82): 179 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 381 us MUL_MAT_ID(ffn_moe_down-83): 186 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 366 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 398 us MUL_MAT_ID(ffn_moe_down-85): 515 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 362 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 344 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 382 us MUL_MAT_ID(ffn_moe_down-88): 184 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 423 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 389 us MUL_MAT_ID(ffn_moe_down-90): 227 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 367 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 42 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 355 us MUL_MAT_ID(ffn_moe_down-25): 246 us ggml_barrier(...): 509 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 373 us MUL_MAT_ID(ffn_moe_down-26): 255 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 406 us MUL_MAT_ID(ffn_moe_down-27): 202 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 382 us MUL_MAT_ID(ffn_moe_down-28): 295 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 399 us MUL_MAT_ID(ffn_moe_down-29): 191 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 422 us MUL_MAT_ID(ffn_moe_down-30): 267 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 399 us MUL_MAT_ID(ffn_moe_down-31): 204 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 403 us MUL_MAT_ID(ffn_moe_down-32): 184 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 358 us MUL_MAT_ID(ffn_moe_down-33): 223 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 400 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 387 us MUL_MAT_ID(ffn_moe_down-35): 181 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 369 us MUL_MAT_ID(ffn_moe_down-36): 205 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 373 us MUL_MAT_ID(ffn_moe_down-37): 266 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 398 us MUL_MAT_ID(ffn_moe_down-38): 167 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 356 us MUL_MAT_ID(ffn_moe_down-39): 241 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 391 us MUL_MAT_ID(ffn_moe_down-40): 251 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 454 us MUL_MAT_ID(ffn_moe_down-41): 179 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 382 us MUL_MAT_ID(ffn_moe_down-42): 233 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 381 us MUL_MAT_ID(ffn_moe_down-43): 263 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 376 us MUL_MAT_ID(ffn_moe_down-44): 377 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 394 us MUL_MAT_ID(ffn_moe_down-45): 247 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 361 us MUL_MAT_ID(ffn_moe_down-46): 185 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 355 us MUL_MAT_ID(ffn_moe_down-47): 410 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 384 us MUL_MAT_ID(ffn_moe_down-48): 388 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 419 us MUL_MAT_ID(ffn_moe_down-49): 290 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 392 us MUL_MAT_ID(ffn_moe_down-51): 286 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 551 us MUL_MAT_ID(ffn_moe_down-52): 287 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 532 us MUL_MAT_ID(ffn_moe_down-53): 337 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 365 us MUL_MAT_ID(ffn_moe_down-54): 528 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 383 us MUL_MAT_ID(ffn_moe_down-55): 179 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 390 us MUL_MAT_ID(ffn_moe_down-56): 232 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 486 us MUL_MAT_ID(ffn_moe_down-57): 212 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 400 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 387 us MUL_MAT_ID(ffn_moe_down-59): 226 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 433 us MUL_MAT_ID(ffn_moe_down-60): 400 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 381 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 441 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 388 us MUL_MAT_ID(ffn_moe_down-63): 231 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 382 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 411 us MUL_MAT_ID(ffn_moe_down-65): 225 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 408 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 413 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 365 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 387 us MUL_MAT_ID(ffn_moe_down-69): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 371 us MUL_MAT_ID(ffn_moe_down-70): 379 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 390 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 372 us MUL_MAT_ID(ffn_moe_down-72): 206 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 546 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 417 us MUL_MAT_ID(ffn_moe_down-74): 248 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 368 us MUL_MAT_ID(ffn_moe_down-75): 223 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 400 us MUL_MAT_ID(ffn_moe_down-76): 385 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 380 us MUL_MAT_ID(ffn_moe_down-77): 185 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 408 us MUL_MAT_ID(ffn_moe_down-78): 307 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 390 us MUL_MAT_ID(ffn_moe_down-79): 303 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 394 us MUL_MAT_ID(ffn_moe_down-80): 379 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 375 us MUL_MAT_ID(ffn_moe_down-81): 223 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 394 us MUL_MAT_ID(ffn_moe_down-82): 184 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 383 us MUL_MAT_ID(ffn_moe_down-83): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 391 us MUL_MAT_ID(ffn_moe_down-84): 632 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 395 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 393 us MUL_MAT_ID(ffn_moe_down-86): 345 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 416 us MUL_MAT_ID(ffn_moe_down-87): 232 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 391 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 374 us MUL_MAT_ID(ffn_moe_down-89): 403 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 362 us MUL_MAT_ID(ffn_moe_down-90): 237 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 346 us MUL_MAT_ID(ffn_moe_down-91): 382 us ggml_barrier(...): 63 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 454 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 394 us MUL_MAT_ID(ffn_moe_down-26): 447 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 432 us MUL_MAT_ID(ffn_moe_down-27): 157 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 449 us MUL_MAT_ID(ffn_moe_down-28): 264 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 421 us MUL_MAT_ID(ffn_moe_down-29): 173 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 404 us MUL_MAT_ID(ffn_moe_down-30): 191 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 393 us MUL_MAT_ID(ffn_moe_down-31): 202 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 384 us MUL_MAT_ID(ffn_moe_down-32): 213 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 393 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 457 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 423 us MUL_MAT_ID(ffn_moe_down-35): 204 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 412 us MUL_MAT_ID(ffn_moe_down-36): 202 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 481 us MUL_MAT_ID(ffn_moe_down-37): 287 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 468 us MUL_MAT_ID(ffn_moe_down-38): 204 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 393 us MUL_MAT_ID(ffn_moe_down-39): 251 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 421 us MUL_MAT_ID(ffn_moe_down-40): 213 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 418 us MUL_MAT_ID(ffn_moe_down-41): 195 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 372 us MUL_MAT_ID(ffn_moe_down-42): 199 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 394 us MUL_MAT_ID(ffn_moe_down-43): 534 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 398 us MUL_MAT_ID(ffn_moe_down-44): 187 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 478 us MUL_MAT_ID(ffn_moe_down-45): 200 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 355 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 398 us MUL_MAT_ID(ffn_moe_down-47): 208 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 334 us MUL_MAT_ID(ffn_moe_down-48): 173 us ggml_barrier(...): 340 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 389 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 365 us MUL_MAT_ID(ffn_moe_down-50): 481 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 379 us MUL_MAT_ID(ffn_moe_down-51): 234 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 404 us MUL_MAT_ID(ffn_moe_down-52): 379 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 389 us MUL_MAT_ID(ffn_moe_down-53): 294 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 485 us MUL_MAT_ID(ffn_moe_down-54): 201 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 370 us MUL_MAT_ID(ffn_moe_down-55): 244 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 379 us MUL_MAT_ID(ffn_moe_down-56): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 398 us MUL_MAT_ID(ffn_moe_down-57): 180 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 465 us MUL_MAT_ID(ffn_moe_down-58): 216 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 411 us MUL_MAT_ID(ffn_moe_down-59): 411 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 356 us MUL_MAT_ID(ffn_moe_down-60): 171 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 447 us MUL_MAT_ID(ffn_moe_down-61): 187 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 364 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 399 us MUL_MAT_ID(ffn_moe_down-63): 245 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 379 us MUL_MAT_ID(ffn_moe_down-64): 212 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 379 us MUL_MAT_ID(ffn_moe_down-65): 250 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 407 us MUL_MAT_ID(ffn_moe_down-66): 224 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 379 us MUL_MAT_ID(ffn_moe_down-67): 289 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 340 us MUL_MAT_ID(ffn_moe_down-68): 233 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 371 us MUL_MAT_ID(ffn_moe_down-69): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 374 us MUL_MAT_ID(ffn_moe_down-70): 348 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 465 us MUL_MAT_ID(ffn_moe_down-71): 266 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 378 us MUL_MAT_ID(ffn_moe_down-72): 382 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 391 us MUL_MAT_ID(ffn_moe_down-73): 217 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 355 us MUL_MAT_ID(ffn_moe_down-74): 228 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 380 us MUL_MAT_ID(ffn_moe_down-75): 204 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 393 us MUL_MAT_ID(ffn_moe_down-76): 331 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 404 us MUL_MAT_ID(ffn_moe_down-77): 212 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 652 us MUL_MAT_ID(ffn_moe_down-78): 219 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 390 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 397 us MUL_MAT_ID(ffn_moe_down-80): 186 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 421 us MUL_MAT_ID(ffn_moe_down-81): 306 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 377 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 397 us MUL_MAT_ID(ffn_moe_down-83): 274 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 431 us MUL_MAT_ID(ffn_moe_down-84): 407 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 425 us MUL_MAT_ID(ffn_moe_down-85): 240 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 414 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 367 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 409 us MUL_MAT_ID(ffn_moe_down-88): 204 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 467 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 376 us MUL_MAT_ID(ffn_moe_down-90): 219 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 427 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 28 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 396 us MUL_MAT_ID(ffn_moe_down-25): 182 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 409 us MUL_MAT_ID(ffn_moe_down-26): 220 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 446 us MUL_MAT_ID(ffn_moe_down-27): 289 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 379 us MUL_MAT_ID(ffn_moe_down-28): 217 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 421 us MUL_MAT_ID(ffn_moe_down-29): 177 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 418 us MUL_MAT_ID(ffn_moe_down-30): 185 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 366 us MUL_MAT_ID(ffn_moe_down-31): 206 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 459 us MUL_MAT_ID(ffn_moe_down-32): 256 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 424 us MUL_MAT_ID(ffn_moe_down-33): 204 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 513 us MUL_MAT_ID(ffn_moe_down-34): 202 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 473 us MUL_MAT_ID(ffn_moe_down-35): 171 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 379 us MUL_MAT_ID(ffn_moe_down-36): 412 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 390 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 475 us MUL_MAT_ID(ffn_moe_down-38): 269 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 380 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 363 us MUL_MAT_ID(ffn_moe_down-40): 416 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 521 us MUL_MAT_ID(ffn_moe_down-41): 332 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 617 us MUL_MAT_ID(ffn_moe_down-42): 314 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 398 us MUL_MAT_ID(ffn_moe_down-43): 204 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 415 us MUL_MAT_ID(ffn_moe_down-44): 270 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 380 us MUL_MAT_ID(ffn_moe_down-45): 355 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 388 us MUL_MAT_ID(ffn_moe_down-46): 247 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 362 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 404 us MUL_MAT_ID(ffn_moe_down-48): 209 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 407 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 388 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 378 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 501 us MUL_MAT_ID(ffn_moe_down-52): 219 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 371 us MUL_MAT_ID(ffn_moe_down-53): 320 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 395 us MUL_MAT_ID(ffn_moe_down-54): 266 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 359 us MUL_MAT_ID(ffn_moe_down-55): 184 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 375 us MUL_MAT_ID(ffn_moe_down-56): 157 us ggml_barrier(...): 322 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 416 us MUL_MAT_ID(ffn_moe_down-57): 180 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 515 us MUL_MAT_ID(ffn_moe_down-58): 183 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 387 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 496 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 395 us MUL_MAT_ID(ffn_moe_down-61): 185 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 369 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 381 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 422 us MUL_MAT_ID(ffn_moe_down-64): 223 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 407 us MUL_MAT_ID(ffn_moe_down-65): 175 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 432 us MUL_MAT_ID(ffn_moe_down-66): 290 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 351 us MUL_MAT_ID(ffn_moe_down-67): 358 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 356 us MUL_MAT_ID(ffn_moe_down-68): 237 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 513 us MUL_MAT_ID(ffn_moe_down-69): 200 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 395 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 358 us MUL_MAT_ID(ffn_moe_down-71): 270 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 372 us MUL_MAT_ID(ffn_moe_down-72): 178 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 410 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 388 us MUL_MAT_ID(ffn_moe_down-74): 187 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 383 us MUL_MAT_ID(ffn_moe_down-75): 277 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 373 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 403 us MUL_MAT_ID(ffn_moe_down-77): 222 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 402 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 434 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 520 us MUL_MAT_ID(ffn_moe_down-80): 204 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 356 us MUL_MAT_ID(ffn_moe_down-81): 182 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 409 us MUL_MAT_ID(ffn_moe_down-82): 446 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 381 us MUL_MAT_ID(ffn_moe_down-83): 384 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 433 us MUL_MAT_ID(ffn_moe_down-84): 385 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 394 us MUL_MAT_ID(ffn_moe_down-85): 653 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 387 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 398 us MUL_MAT_ID(ffn_moe_down-87): 189 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 558 us MUL_MAT_ID(ffn_moe_down-88): 208 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 366 us MUL_MAT_ID(ffn_moe_down-89): 186 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 393 us MUL_MAT_ID(ffn_moe_down-90): 248 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 451 us MUL_MAT_ID(ffn_moe_down-91): 251 us ggml_barrier(...): 206 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 435 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 405 us MUL_MAT_ID(ffn_moe_down-26): 227 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 395 us MUL_MAT_ID(ffn_moe_down-27): 200 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 358 us MUL_MAT_ID(ffn_moe_down-28): 180 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 377 us MUL_MAT_ID(ffn_moe_down-29): 204 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 462 us MUL_MAT_ID(ffn_moe_down-30): 311 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 400 us MUL_MAT_ID(ffn_moe_down-31): 269 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 397 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 422 us MUL_MAT_ID(ffn_moe_down-33): 208 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 428 us MUL_MAT_ID(ffn_moe_down-34): 184 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 395 us MUL_MAT_ID(ffn_moe_down-35): 207 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 353 us MUL_MAT_ID(ffn_moe_down-36): 536 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 379 us MUL_MAT_ID(ffn_moe_down-37): 293 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 356 us MUL_MAT_ID(ffn_moe_down-38): 202 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 371 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 363 us MUL_MAT_ID(ffn_moe_down-40): 729 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 395 us MUL_MAT_ID(ffn_moe_down-41): 222 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 415 us MUL_MAT_ID(ffn_moe_down-42): 328 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 388 us MUL_MAT_ID(ffn_moe_down-43): 214 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 468 us MUL_MAT_ID(ffn_moe_down-44): 215 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 401 us MUL_MAT_ID(ffn_moe_down-45): 203 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 406 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 380 us MUL_MAT_ID(ffn_moe_down-47): 536 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 402 us MUL_MAT_ID(ffn_moe_down-48): 215 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 385 us MUL_MAT_ID(ffn_moe_down-49): 212 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 379 us MUL_MAT_ID(ffn_moe_down-50): 214 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 384 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 443 us MUL_MAT_ID(ffn_moe_down-52): 538 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 384 us MUL_MAT_ID(ffn_moe_down-53): 452 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 424 us MUL_MAT_ID(ffn_moe_down-54): 216 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 377 us MUL_MAT_ID(ffn_moe_down-55): 180 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 518 us MUL_MAT_ID(ffn_moe_down-56): 171 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 432 us MUL_MAT_ID(ffn_moe_down-57): 326 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 376 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 440 us MUL_MAT_ID(ffn_moe_down-59): 217 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 384 us MUL_MAT_ID(ffn_moe_down-60): 212 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 425 us MUL_MAT_ID(ffn_moe_down-61): 218 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 491 us MUL_MAT_ID(ffn_moe_down-62): 236 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 362 us MUL_MAT_ID(ffn_moe_down-63): 218 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 405 us MUL_MAT_ID(ffn_moe_down-64): 196 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 409 us MUL_MAT_ID(ffn_moe_down-65): 178 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 383 us MUL_MAT_ID(ffn_moe_down-66): 227 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 420 us MUL_MAT_ID(ffn_moe_down-67): 246 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 383 us MUL_MAT_ID(ffn_moe_down-68): 203 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 351 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 385 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 400 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 381 us MUL_MAT_ID(ffn_moe_down-72): 278 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 362 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 368 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 505 us MUL_MAT_ID(ffn_moe_down-75): 263 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 380 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 388 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 515 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 404 us MUL_MAT_ID(ffn_moe_down-79): 229 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 373 us MUL_MAT_ID(ffn_moe_down-80): 336 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 379 us MUL_MAT_ID(ffn_moe_down-81): 207 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 386 us MUL_MAT_ID(ffn_moe_down-82): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 384 us MUL_MAT_ID(ffn_moe_down-83): 202 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 407 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 481 us MUL_MAT_ID(ffn_moe_down-85): 281 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 398 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 381 us MUL_MAT_ID(ffn_moe_down-87): 168 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 398 us MUL_MAT_ID(ffn_moe_down-88): 481 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 415 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 354 us MUL_MAT_ID(ffn_moe_down-90): 240 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 497 us MUL_MAT_ID(ffn_moe_down-91): 274 us ggml_barrier(...): 75 us GET_ROWS(inp_embd): 11 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 413 us MUL_MAT_ID(ffn_moe_down-25): 188 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 445 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 510 us MUL_MAT_ID(ffn_moe_down-27): 169 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 429 us MUL_MAT_ID(ffn_moe_down-28): 186 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 454 us MUL_MAT_ID(ffn_moe_down-29): 265 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 376 us MUL_MAT_ID(ffn_moe_down-30): 180 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 448 us MUL_MAT_ID(ffn_moe_down-31): 278 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 511 us MUL_MAT_ID(ffn_moe_down-32): 203 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 391 us MUL_MAT_ID(ffn_moe_down-33): 253 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 448 us MUL_MAT_ID(ffn_moe_down-34): 373 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 372 us MUL_MAT_ID(ffn_moe_down-35): 208 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 395 us MUL_MAT_ID(ffn_moe_down-36): 206 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 351 us MUL_MAT_ID(ffn_moe_down-37): 261 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 487 us MUL_MAT_ID(ffn_moe_down-38): 189 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 380 us MUL_MAT_ID(ffn_moe_down-39): 289 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 372 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 489 us MUL_MAT_ID(ffn_moe_down-41): 214 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 373 us MUL_MAT_ID(ffn_moe_down-42): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 475 us MUL_MAT_ID(ffn_moe_down-43): 302 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 410 us MUL_MAT_ID(ffn_moe_down-44): 187 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 408 us MUL_MAT_ID(ffn_moe_down-45): 220 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 370 us MUL_MAT_ID(ffn_moe_down-46): 205 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 402 us MUL_MAT_ID(ffn_moe_down-47): 573 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 370 us MUL_MAT_ID(ffn_moe_down-48): 284 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 392 us MUL_MAT_ID(ffn_moe_down-49): 263 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 683 us MUL_MAT_ID(ffn_moe_down-50): 291 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 348 us MUL_MAT_ID(ffn_moe_down-51): 331 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 396 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 369 us MUL_MAT_ID(ffn_moe_down-53): 215 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 354 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 379 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 381 us MUL_MAT_ID(ffn_moe_down-56): 217 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 356 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 405 us MUL_MAT_ID(ffn_moe_down-58): 204 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 503 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 364 us MUL_MAT_ID(ffn_moe_down-60): 207 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 409 us MUL_MAT_ID(ffn_moe_down-61): 164 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 357 us MUL_MAT_ID(ffn_moe_down-62): 226 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 379 us MUL_MAT_ID(ffn_moe_down-63): 198 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 400 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 382 us MUL_MAT_ID(ffn_moe_down-65): 259 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 406 us MUL_MAT_ID(ffn_moe_down-66): 377 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 518 us MUL_MAT_ID(ffn_moe_down-67): 506 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 395 us MUL_MAT_ID(ffn_moe_down-68): 230 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 500 us MUL_MAT_ID(ffn_moe_down-69): 203 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 385 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 466 us MUL_MAT_ID(ffn_moe_down-71): 298 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 403 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 364 us MUL_MAT_ID(ffn_moe_down-73): 174 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 419 us MUL_MAT_ID(ffn_moe_down-74): 178 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 374 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 416 us MUL_MAT_ID(ffn_moe_down-76): 184 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 360 us MUL_MAT_ID(ffn_moe_down-77): 215 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 420 us MUL_MAT_ID(ffn_moe_down-78): 198 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 426 us MUL_MAT_ID(ffn_moe_down-79): 206 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 378 us MUL_MAT_ID(ffn_moe_down-80): 209 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 347 us MUL_MAT_ID(ffn_moe_down-81): 534 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 391 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 404 us MUL_MAT_ID(ffn_moe_down-83): 215 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 371 us MUL_MAT_ID(ffn_moe_down-84): 223 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 432 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 369 us MUL_MAT_ID(ffn_moe_down-86): 198 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 394 us MUL_MAT_ID(ffn_moe_down-87): 354 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 361 us MUL_MAT_ID(ffn_moe_down-88): 180 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 418 us MUL_MAT_ID(ffn_moe_down-89): 562 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 393 us MUL_MAT_ID(ffn_moe_down-90): 220 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 571 us MUL_MAT_ID(ffn_moe_down-91): 278 us ggml_barrier(...): 83 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 419 us MUL_MAT_ID(ffn_moe_down-25): 188 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 364 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 461 us MUL_MAT_ID(ffn_moe_down-27): 328 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 354 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 396 us MUL_MAT_ID(ffn_moe_down-29): 349 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 422 us MUL_MAT_ID(ffn_moe_down-30): 529 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 394 us MUL_MAT_ID(ffn_moe_down-31): 179 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 378 us MUL_MAT_ID(ffn_moe_down-32): 182 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 392 us MUL_MAT_ID(ffn_moe_down-33): 242 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 371 us MUL_MAT_ID(ffn_moe_down-34): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 366 us MUL_MAT_ID(ffn_moe_down-35): 183 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 351 us MUL_MAT_ID(ffn_moe_down-36): 184 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 363 us MUL_MAT_ID(ffn_moe_down-37): 242 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 444 us MUL_MAT_ID(ffn_moe_down-38): 184 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 365 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 398 us MUL_MAT_ID(ffn_moe_down-40): 227 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 425 us MUL_MAT_ID(ffn_moe_down-41): 183 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 378 us MUL_MAT_ID(ffn_moe_down-42): 178 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 391 us MUL_MAT_ID(ffn_moe_down-43): 390 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 473 us MUL_MAT_ID(ffn_moe_down-44): 284 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 419 us MUL_MAT_ID(ffn_moe_down-45): 189 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 355 us MUL_MAT_ID(ffn_moe_down-46): 268 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 364 us MUL_MAT_ID(ffn_moe_down-47): 179 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 376 us MUL_MAT_ID(ffn_moe_down-48): 212 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 367 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 396 us MUL_MAT_ID(ffn_moe_down-50): 216 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 401 us MUL_MAT_ID(ffn_moe_down-51): 432 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 375 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 387 us MUL_MAT_ID(ffn_moe_down-53): 213 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 426 us MUL_MAT_ID(ffn_moe_down-54): 192 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 401 us MUL_MAT_ID(ffn_moe_down-55): 277 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 394 us MUL_MAT_ID(ffn_moe_down-56): 201 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 363 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 508 us MUL_MAT_ID(ffn_moe_down-58): 314 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 371 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 397 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 390 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 397 us MUL_MAT_ID(ffn_moe_down-62): 365 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 387 us MUL_MAT_ID(ffn_moe_down-63): 184 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 493 us MUL_MAT_ID(ffn_moe_down-64): 179 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 382 us MUL_MAT_ID(ffn_moe_down-65): 176 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 419 us MUL_MAT_ID(ffn_moe_down-66): 310 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 405 us MUL_MAT_ID(ffn_moe_down-67): 214 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 376 us MUL_MAT_ID(ffn_moe_down-68): 258 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 412 us MUL_MAT_ID(ffn_moe_down-69): 192 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 389 us MUL_MAT_ID(ffn_moe_down-70): 158 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 401 us MUL_MAT_ID(ffn_moe_down-71): 205 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 388 us MUL_MAT_ID(ffn_moe_down-72): 368 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 397 us MUL_MAT_ID(ffn_moe_down-73): 215 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 498 us MUL_MAT_ID(ffn_moe_down-74): 175 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 414 us MUL_MAT_ID(ffn_moe_down-75): 181 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 369 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 385 us MUL_MAT_ID(ffn_moe_down-77): 187 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 407 us MUL_MAT_ID(ffn_moe_down-78): 211 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 367 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 411 us MUL_MAT_ID(ffn_moe_down-80): 250 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 387 us MUL_MAT_ID(ffn_moe_down-81): 163 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 407 us MUL_MAT_ID(ffn_moe_down-82): 235 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 469 us MUL_MAT_ID(ffn_moe_down-83): 175 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 472 us MUL_MAT_ID(ffn_moe_down-84): 263 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 421 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 366 us MUL_MAT_ID(ffn_moe_down-86): 216 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 381 us MUL_MAT_ID(ffn_moe_down-87): 215 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 464 us MUL_MAT_ID(ffn_moe_down-88): 288 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 360 us MUL_MAT_ID(ffn_moe_down-89): 218 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 412 us MUL_MAT_ID(ffn_moe_down-90): 277 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 487 us MUL_MAT_ID(ffn_moe_down-91): 762 us ggml_barrier(...): 135 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 411 us MUL_MAT_ID(ffn_moe_down-25): 206 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 482 us MUL_MAT_ID(ffn_moe_down-26): 278 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 414 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 404 us MUL_MAT_ID(ffn_moe_down-28): 262 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 438 us MUL_MAT_ID(ffn_moe_down-29): 159 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 386 us MUL_MAT_ID(ffn_moe_down-30): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 381 us MUL_MAT_ID(ffn_moe_down-31): 187 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 423 us MUL_MAT_ID(ffn_moe_down-32): 172 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 382 us MUL_MAT_ID(ffn_moe_down-33): 187 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 467 us MUL_MAT_ID(ffn_moe_down-34): 233 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 393 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 362 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 432 us MUL_MAT_ID(ffn_moe_down-37): 253 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 349 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 386 us MUL_MAT_ID(ffn_moe_down-39): 246 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 385 us MUL_MAT_ID(ffn_moe_down-40): 647 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 400 us MUL_MAT_ID(ffn_moe_down-41): 195 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 507 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 405 us MUL_MAT_ID(ffn_moe_down-43): 458 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 438 us MUL_MAT_ID(ffn_moe_down-44): 184 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 360 us MUL_MAT_ID(ffn_moe_down-45): 235 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 380 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 382 us MUL_MAT_ID(ffn_moe_down-47): 184 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 390 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 410 us MUL_MAT_ID(ffn_moe_down-49): 213 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 187 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 393 us MUL_MAT_ID(ffn_moe_down-51): 191 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 400 us MUL_MAT_ID(ffn_moe_down-52): 203 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 400 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 348 us MUL_MAT_ID(ffn_moe_down-54): 165 us ggml_barrier(...): 400 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 451 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 402 us MUL_MAT_ID(ffn_moe_down-56): 614 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 371 us MUL_MAT_ID(ffn_moe_down-57): 364 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 495 us MUL_MAT_ID(ffn_moe_down-58): 371 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 371 us MUL_MAT_ID(ffn_moe_down-59): 381 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 402 us MUL_MAT_ID(ffn_moe_down-60): 191 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 374 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 365 us MUL_MAT_ID(ffn_moe_down-62): 283 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 374 us MUL_MAT_ID(ffn_moe_down-63): 331 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 391 us MUL_MAT_ID(ffn_moe_down-64): 241 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 402 us MUL_MAT_ID(ffn_moe_down-65): 389 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 371 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 361 us MUL_MAT_ID(ffn_moe_down-67): 178 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 348 us MUL_MAT_ID(ffn_moe_down-68): 374 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 427 us MUL_MAT_ID(ffn_moe_down-69): 167 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 363 us MUL_MAT_ID(ffn_moe_down-70): 216 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 370 us MUL_MAT_ID(ffn_moe_down-71): 210 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 385 us MUL_MAT_ID(ffn_moe_down-72): 324 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 380 us MUL_MAT_ID(ffn_moe_down-73): 184 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 388 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 356 us MUL_MAT_ID(ffn_moe_down-75): 209 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 372 us MUL_MAT_ID(ffn_moe_down-76): 197 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 445 us MUL_MAT_ID(ffn_moe_down-77): 285 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 450 us MUL_MAT_ID(ffn_moe_down-78): 198 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 404 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 435 us MUL_MAT_ID(ffn_moe_down-80): 319 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 403 us MUL_MAT_ID(ffn_moe_down-81): 182 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 379 us MUL_MAT_ID(ffn_moe_down-82): 367 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 404 us MUL_MAT_ID(ffn_moe_down-83): 403 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 391 us MUL_MAT_ID(ffn_moe_down-84): 248 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 382 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 393 us MUL_MAT_ID(ffn_moe_down-86): 185 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 383 us MUL_MAT_ID(ffn_moe_down-87): 508 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 442 us MUL_MAT_ID(ffn_moe_down-88): 572 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 377 us MUL_MAT_ID(ffn_moe_down-89): 361 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 434 us MUL_MAT_ID(ffn_moe_down-90): 230 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 407 us MUL_MAT_ID(ffn_moe_down-91): 274 us ggml_barrier(...): 194 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 398 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 387 us MUL_MAT_ID(ffn_moe_down-26): 248 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 383 us MUL_MAT_ID(ffn_moe_down-27): 217 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 384 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 385 us MUL_MAT_ID(ffn_moe_down-29): 211 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 405 us MUL_MAT_ID(ffn_moe_down-30): 188 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 394 us MUL_MAT_ID(ffn_moe_down-31): 235 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 381 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 375 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 348 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 467 us MUL_MAT_ID(ffn_moe_down-36): 177 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 406 us MUL_MAT_ID(ffn_moe_down-37): 246 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 348 us MUL_MAT_ID(ffn_moe_down-38): 527 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 423 us MUL_MAT_ID(ffn_moe_down-39): 688 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 428 us MUL_MAT_ID(ffn_moe_down-40): 241 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 380 us MUL_MAT_ID(ffn_moe_down-41): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 418 us MUL_MAT_ID(ffn_moe_down-42): 344 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 486 us MUL_MAT_ID(ffn_moe_down-43): 160 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 394 us MUL_MAT_ID(ffn_moe_down-44): 219 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 364 us MUL_MAT_ID(ffn_moe_down-45): 246 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 387 us MUL_MAT_ID(ffn_moe_down-46): 211 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 377 us MUL_MAT_ID(ffn_moe_down-47): 206 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 349 us MUL_MAT_ID(ffn_moe_down-48): 236 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 381 us MUL_MAT_ID(ffn_moe_down-49): 412 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 392 us MUL_MAT_ID(ffn_moe_down-50): 221 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 364 us MUL_MAT_ID(ffn_moe_down-51): 235 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 396 us MUL_MAT_ID(ffn_moe_down-52): 207 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 369 us MUL_MAT_ID(ffn_moe_down-53): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 485 us MUL_MAT_ID(ffn_moe_down-54): 247 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 399 us MUL_MAT_ID(ffn_moe_down-55): 217 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 478 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 416 us MUL_MAT_ID(ffn_moe_down-57): 161 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 389 us MUL_MAT_ID(ffn_moe_down-58): 176 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 376 us MUL_MAT_ID(ffn_moe_down-59): 295 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 460 us MUL_MAT_ID(ffn_moe_down-60): 392 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 394 us MUL_MAT_ID(ffn_moe_down-61): 203 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 372 us MUL_MAT_ID(ffn_moe_down-62): 277 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 461 us MUL_MAT_ID(ffn_moe_down-63): 200 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 407 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 344 us MUL_MAT_ID(ffn_moe_down-65): 193 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 370 us MUL_MAT_ID(ffn_moe_down-66): 439 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 381 us MUL_MAT_ID(ffn_moe_down-67): 218 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 434 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 424 us MUL_MAT_ID(ffn_moe_down-69): 195 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 366 us MUL_MAT_ID(ffn_moe_down-70): 399 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 389 us MUL_MAT_ID(ffn_moe_down-71): 271 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 425 us MUL_MAT_ID(ffn_moe_down-72): 461 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 400 us MUL_MAT_ID(ffn_moe_down-73): 389 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 349 us MUL_MAT_ID(ffn_moe_down-74): 206 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 389 us MUL_MAT_ID(ffn_moe_down-75): 314 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 366 us MUL_MAT_ID(ffn_moe_down-76): 187 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 372 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 418 us MUL_MAT_ID(ffn_moe_down-78): 239 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 380 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 383 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 370 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 476 us MUL_MAT_ID(ffn_moe_down-82): 160 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 349 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 440 us MUL_MAT_ID(ffn_moe_down-84): 257 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 388 us MUL_MAT_ID(ffn_moe_down-85): 239 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 364 us MUL_MAT_ID(ffn_moe_down-86): 439 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 381 us MUL_MAT_ID(ffn_moe_down-87): 188 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 436 us MUL_MAT_ID(ffn_moe_down-88): 197 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 390 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 383 us MUL_MAT_ID(ffn_moe_down-90): 213 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 388 us MUL_MAT_ID(ffn_moe_down-91): 229 us ggml_barrier(...): 58 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 452 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 390 us MUL_MAT_ID(ffn_moe_down-26): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 497 us MUL_MAT_ID(ffn_moe_down-27): 525 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 391 us MUL_MAT_ID(ffn_moe_down-28): 196 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 368 us MUL_MAT_ID(ffn_moe_down-29): 224 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 466 us MUL_MAT_ID(ffn_moe_down-30): 273 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 426 us MUL_MAT_ID(ffn_moe_down-31): 298 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 435 us MUL_MAT_ID(ffn_moe_down-32): 177 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 353 us MUL_MAT_ID(ffn_moe_down-33): 187 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 388 us MUL_MAT_ID(ffn_moe_down-34): 202 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 414 us MUL_MAT_ID(ffn_moe_down-35): 250 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 398 us MUL_MAT_ID(ffn_moe_down-36): 380 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 402 us MUL_MAT_ID(ffn_moe_down-37): 245 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 363 us MUL_MAT_ID(ffn_moe_down-38): 532 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 400 us MUL_MAT_ID(ffn_moe_down-39): 247 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 452 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 397 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 412 us MUL_MAT_ID(ffn_moe_down-42): 545 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 391 us MUL_MAT_ID(ffn_moe_down-43): 188 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 388 us MUL_MAT_ID(ffn_moe_down-44): 200 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 406 us MUL_MAT_ID(ffn_moe_down-45): 193 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 348 us MUL_MAT_ID(ffn_moe_down-46): 198 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 387 us MUL_MAT_ID(ffn_moe_down-47): 466 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 435 us MUL_MAT_ID(ffn_moe_down-48): 275 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 437 us MUL_MAT_ID(ffn_moe_down-49): 226 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 375 us MUL_MAT_ID(ffn_moe_down-50): 222 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 364 us MUL_MAT_ID(ffn_moe_down-51): 210 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 425 us MUL_MAT_ID(ffn_moe_down-52): 437 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 517 us MUL_MAT_ID(ffn_moe_down-53): 231 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 433 us MUL_MAT_ID(ffn_moe_down-54): 206 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 401 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 377 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 381 us MUL_MAT_ID(ffn_moe_down-57): 236 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 470 us MUL_MAT_ID(ffn_moe_down-58): 244 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 369 us MUL_MAT_ID(ffn_moe_down-59): 297 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 402 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 379 us MUL_MAT_ID(ffn_moe_down-61): 227 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 467 us MUL_MAT_ID(ffn_moe_down-62): 175 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 375 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 368 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 457 us MUL_MAT_ID(ffn_moe_down-65): 299 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 493 us MUL_MAT_ID(ffn_moe_down-66): 218 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 376 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 394 us MUL_MAT_ID(ffn_moe_down-68): 379 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 425 us MUL_MAT_ID(ffn_moe_down-69): 254 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 429 us MUL_MAT_ID(ffn_moe_down-70): 194 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 423 us MUL_MAT_ID(ffn_moe_down-71): 201 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 383 us MUL_MAT_ID(ffn_moe_down-72): 199 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 395 us MUL_MAT_ID(ffn_moe_down-73): 182 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 384 us MUL_MAT_ID(ffn_moe_down-74): 213 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 333 us MUL_MAT_ID(ffn_moe_down-75): 209 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 376 us MUL_MAT_ID(ffn_moe_down-76): 206 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 374 us MUL_MAT_ID(ffn_moe_down-77): 207 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 410 us MUL_MAT_ID(ffn_moe_down-78): 314 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 373 us MUL_MAT_ID(ffn_moe_down-79): 197 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 390 us MUL_MAT_ID(ffn_moe_down-80): 277 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 373 us MUL_MAT_ID(ffn_moe_down-81): 207 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 387 us MUL_MAT_ID(ffn_moe_down-82): 182 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 357 us MUL_MAT_ID(ffn_moe_down-83): 190 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 389 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 373 us MUL_MAT_ID(ffn_moe_down-85): 704 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 388 us MUL_MAT_ID(ffn_moe_down-86): 481 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 373 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 370 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 392 us MUL_MAT_ID(ffn_moe_down-89): 352 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 381 us MUL_MAT_ID(ffn_moe_down-90): 237 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 391 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 34 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 429 us MUL_MAT_ID(ffn_moe_down-25): 292 us ggml_barrier(...): 494 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 398 us MUL_MAT_ID(ffn_moe_down-26): 209 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 423 us MUL_MAT_ID(ffn_moe_down-27): 187 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 371 us MUL_MAT_ID(ffn_moe_down-28): 187 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 437 us MUL_MAT_ID(ffn_moe_down-29): 280 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 408 us MUL_MAT_ID(ffn_moe_down-30): 275 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 420 us MUL_MAT_ID(ffn_moe_down-31): 372 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 393 us MUL_MAT_ID(ffn_moe_down-32): 362 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 398 us MUL_MAT_ID(ffn_moe_down-33): 217 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 389 us MUL_MAT_ID(ffn_moe_down-34): 206 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 435 us MUL_MAT_ID(ffn_moe_down-35): 256 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 363 us MUL_MAT_ID(ffn_moe_down-36): 188 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 406 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 372 us MUL_MAT_ID(ffn_moe_down-38): 189 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 395 us MUL_MAT_ID(ffn_moe_down-39): 567 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 403 us MUL_MAT_ID(ffn_moe_down-40): 226 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 390 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 390 us MUL_MAT_ID(ffn_moe_down-42): 257 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 424 us MUL_MAT_ID(ffn_moe_down-43): 213 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 439 us MUL_MAT_ID(ffn_moe_down-44): 207 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 365 us MUL_MAT_ID(ffn_moe_down-45): 219 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 404 us MUL_MAT_ID(ffn_moe_down-46): 343 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 462 us MUL_MAT_ID(ffn_moe_down-47): 226 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 502 us MUL_MAT_ID(ffn_moe_down-48): 232 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 499 us MUL_MAT_ID(ffn_moe_down-49): 247 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 415 us MUL_MAT_ID(ffn_moe_down-50): 378 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 417 us MUL_MAT_ID(ffn_moe_down-51): 382 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 364 us MUL_MAT_ID(ffn_moe_down-52): 184 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 366 us MUL_MAT_ID(ffn_moe_down-53): 173 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 486 us MUL_MAT_ID(ffn_moe_down-54): 284 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 399 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 490 us MUL_MAT_ID(ffn_moe_down-56): 332 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 379 us MUL_MAT_ID(ffn_moe_down-57): 200 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 472 us MUL_MAT_ID(ffn_moe_down-58): 237 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 389 us MUL_MAT_ID(ffn_moe_down-59): 253 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 354 us MUL_MAT_ID(ffn_moe_down-60): 193 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 376 us MUL_MAT_ID(ffn_moe_down-61): 199 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 367 us MUL_MAT_ID(ffn_moe_down-62): 186 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 387 us MUL_MAT_ID(ffn_moe_down-63): 184 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 545 us MUL_MAT_ID(ffn_moe_down-64): 225 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 369 us MUL_MAT_ID(ffn_moe_down-65): 285 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 387 us MUL_MAT_ID(ffn_moe_down-66): 201 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 360 us MUL_MAT_ID(ffn_moe_down-67): 212 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 389 us MUL_MAT_ID(ffn_moe_down-68): 202 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 379 us MUL_MAT_ID(ffn_moe_down-69): 222 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 379 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 379 us MUL_MAT_ID(ffn_moe_down-71): 219 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 393 us MUL_MAT_ID(ffn_moe_down-72): 380 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 374 us MUL_MAT_ID(ffn_moe_down-73): 213 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 334 us MUL_MAT_ID(ffn_moe_down-74): 226 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 430 us MUL_MAT_ID(ffn_moe_down-75): 185 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 407 us MUL_MAT_ID(ffn_moe_down-76): 274 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 352 us MUL_MAT_ID(ffn_moe_down-77): 328 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 450 us MUL_MAT_ID(ffn_moe_down-78): 198 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 378 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 401 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 354 us MUL_MAT_ID(ffn_moe_down-81): 214 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 478 us MUL_MAT_ID(ffn_moe_down-82): 178 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 470 us MUL_MAT_ID(ffn_moe_down-83): 210 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 381 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 401 us MUL_MAT_ID(ffn_moe_down-85): 243 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 351 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 389 us MUL_MAT_ID(ffn_moe_down-87): 199 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 427 us MUL_MAT_ID(ffn_moe_down-88): 272 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 377 us MUL_MAT_ID(ffn_moe_down-89): 284 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 377 us MUL_MAT_ID(ffn_moe_down-90): 223 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 397 us MUL_MAT_ID(ffn_moe_down-91): 250 us ggml_barrier(...): 238 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 404 us MUL_MAT_ID(ffn_moe_down-25): 208 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 405 us MUL_MAT_ID(ffn_moe_down-26): 317 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 390 us MUL_MAT_ID(ffn_moe_down-27): 286 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 391 us MUL_MAT_ID(ffn_moe_down-28): 392 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 406 us MUL_MAT_ID(ffn_moe_down-29): 247 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 506 us MUL_MAT_ID(ffn_moe_down-30): 192 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 381 us MUL_MAT_ID(ffn_moe_down-31): 243 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 440 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 365 us MUL_MAT_ID(ffn_moe_down-33): 236 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 382 us MUL_MAT_ID(ffn_moe_down-34): 291 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 510 us MUL_MAT_ID(ffn_moe_down-35): 365 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 363 us MUL_MAT_ID(ffn_moe_down-36): 223 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 373 us MUL_MAT_ID(ffn_moe_down-37): 275 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 485 us MUL_MAT_ID(ffn_moe_down-38): 197 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 363 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 387 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 416 us MUL_MAT_ID(ffn_moe_down-41): 221 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 475 us MUL_MAT_ID(ffn_moe_down-42): 483 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 394 us MUL_MAT_ID(ffn_moe_down-43): 366 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 368 us MUL_MAT_ID(ffn_moe_down-44): 184 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 369 us MUL_MAT_ID(ffn_moe_down-45): 379 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 357 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 348 us MUL_MAT_ID(ffn_moe_down-47): 185 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 383 us MUL_MAT_ID(ffn_moe_down-48): 224 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 396 us MUL_MAT_ID(ffn_moe_down-49): 187 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 368 us MUL_MAT_ID(ffn_moe_down-50): 357 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 452 us MUL_MAT_ID(ffn_moe_down-51): 237 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 410 us MUL_MAT_ID(ffn_moe_down-52): 202 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 373 us MUL_MAT_ID(ffn_moe_down-53): 183 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 678 us MUL_MAT_ID(ffn_moe_down-54): 300 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 353 us MUL_MAT_ID(ffn_moe_down-55): 221 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 378 us MUL_MAT_ID(ffn_moe_down-56): 229 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 412 us MUL_MAT_ID(ffn_moe_down-57): 186 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 378 us MUL_MAT_ID(ffn_moe_down-58): 181 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 370 us MUL_MAT_ID(ffn_moe_down-59): 180 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 338 us MUL_MAT_ID(ffn_moe_down-60): 516 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 390 us MUL_MAT_ID(ffn_moe_down-61): 325 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 423 us MUL_MAT_ID(ffn_moe_down-62): 354 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 365 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 429 us MUL_MAT_ID(ffn_moe_down-64): 269 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 386 us MUL_MAT_ID(ffn_moe_down-65): 206 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 408 us MUL_MAT_ID(ffn_moe_down-66): 205 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 407 us MUL_MAT_ID(ffn_moe_down-67): 177 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 343 us MUL_MAT_ID(ffn_moe_down-68): 203 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 374 us MUL_MAT_ID(ffn_moe_down-69): 200 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 385 us MUL_MAT_ID(ffn_moe_down-70): 217 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 354 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 214 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 413 us MUL_MAT_ID(ffn_moe_down-73): 373 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 393 us MUL_MAT_ID(ffn_moe_down-74): 239 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 393 us MUL_MAT_ID(ffn_moe_down-75): 191 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 417 us MUL_MAT_ID(ffn_moe_down-76): 184 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 361 us MUL_MAT_ID(ffn_moe_down-77): 196 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 409 us MUL_MAT_ID(ffn_moe_down-78): 245 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 366 us MUL_MAT_ID(ffn_moe_down-79): 212 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 419 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 397 us MUL_MAT_ID(ffn_moe_down-81): 519 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 413 us MUL_MAT_ID(ffn_moe_down-82): 339 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 409 us MUL_MAT_ID(ffn_moe_down-83): 190 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 354 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 362 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 377 us MUL_MAT_ID(ffn_moe_down-86): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 408 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 342 us MUL_MAT_ID(ffn_moe_down-88): 217 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 392 us MUL_MAT_ID(ffn_moe_down-89): 330 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 393 us MUL_MAT_ID(ffn_moe_down-90): 412 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 365 us MUL_MAT_ID(ffn_moe_down-91): 252 us ggml_barrier(...): 56 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 415 us MUL_MAT_ID(ffn_moe_down-25): 238 us ggml_barrier(...): 420 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 383 us MUL_MAT_ID(ffn_moe_down-26): 291 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 391 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 371 us MUL_MAT_ID(ffn_moe_down-28): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 384 us MUL_MAT_ID(ffn_moe_down-29): 202 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 373 us MUL_MAT_ID(ffn_moe_down-31): 201 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 421 us MUL_MAT_ID(ffn_moe_down-32): 173 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 359 us MUL_MAT_ID(ffn_moe_down-33): 207 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 381 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 408 us MUL_MAT_ID(ffn_moe_down-35): 348 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 374 us MUL_MAT_ID(ffn_moe_down-36): 360 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 354 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 362 us MUL_MAT_ID(ffn_moe_down-38): 382 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 459 us MUL_MAT_ID(ffn_moe_down-39): 214 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 404 us MUL_MAT_ID(ffn_moe_down-40): 649 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 398 us MUL_MAT_ID(ffn_moe_down-41): 189 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 428 us MUL_MAT_ID(ffn_moe_down-42): 189 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 384 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 491 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 343 us MUL_MAT_ID(ffn_moe_down-45): 250 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 488 us MUL_MAT_ID(ffn_moe_down-46): 168 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 395 us MUL_MAT_ID(ffn_moe_down-47): 229 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 441 us MUL_MAT_ID(ffn_moe_down-48): 222 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 358 us MUL_MAT_ID(ffn_moe_down-49): 239 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 362 us MUL_MAT_ID(ffn_moe_down-50): 184 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 458 us MUL_MAT_ID(ffn_moe_down-51): 274 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 407 us MUL_MAT_ID(ffn_moe_down-52): 197 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 388 us MUL_MAT_ID(ffn_moe_down-53): 187 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 437 us MUL_MAT_ID(ffn_moe_down-54): 238 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 187 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 374 us MUL_MAT_ID(ffn_moe_down-57): 193 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 408 us MUL_MAT_ID(ffn_moe_down-58): 393 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 402 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 380 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 526 us MUL_MAT_ID(ffn_moe_down-61): 174 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 376 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 374 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 403 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 344 us MUL_MAT_ID(ffn_moe_down-65): 277 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 394 us MUL_MAT_ID(ffn_moe_down-66): 344 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 345 us MUL_MAT_ID(ffn_moe_down-67): 185 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 474 us MUL_MAT_ID(ffn_moe_down-68): 173 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 408 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 333 us MUL_MAT_ID(ffn_moe_down-70): 359 us ggml_barrier(...): 352 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 400 us MUL_MAT_ID(ffn_moe_down-71): 187 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 345 us MUL_MAT_ID(ffn_moe_down-72): 181 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 396 us MUL_MAT_ID(ffn_moe_down-73): 180 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 362 us MUL_MAT_ID(ffn_moe_down-74): 374 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 392 us MUL_MAT_ID(ffn_moe_down-75): 249 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 416 us MUL_MAT_ID(ffn_moe_down-76): 208 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 399 us MUL_MAT_ID(ffn_moe_down-77): 257 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 366 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 351 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 377 us MUL_MAT_ID(ffn_moe_down-80): 190 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 355 us MUL_MAT_ID(ffn_moe_down-81): 222 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 373 us MUL_MAT_ID(ffn_moe_down-82): 296 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 391 us MUL_MAT_ID(ffn_moe_down-83): 207 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 407 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 371 us MUL_MAT_ID(ffn_moe_down-85): 243 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 384 us MUL_MAT_ID(ffn_moe_down-86): 236 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 354 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 410 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 394 us MUL_MAT_ID(ffn_moe_down-89): 225 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 406 us MUL_MAT_ID(ffn_moe_down-90): 546 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 362 us MUL_MAT_ID(ffn_moe_down-91): 248 us ggml_barrier(...): 62 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 402 us MUL_MAT_ID(ffn_moe_down-25): 185 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 385 us MUL_MAT_ID(ffn_moe_down-26): 264 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 390 us MUL_MAT_ID(ffn_moe_down-27): 285 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 347 us MUL_MAT_ID(ffn_moe_down-28): 350 us ggml_barrier(...): 336 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 401 us MUL_MAT_ID(ffn_moe_down-29): 202 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 420 us MUL_MAT_ID(ffn_moe_down-30): 210 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 401 us MUL_MAT_ID(ffn_moe_down-31): 233 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 376 us MUL_MAT_ID(ffn_moe_down-32): 403 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 422 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 210 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 365 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 396 us MUL_MAT_ID(ffn_moe_down-36): 238 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 394 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 388 us MUL_MAT_ID(ffn_moe_down-38): 236 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 402 us MUL_MAT_ID(ffn_moe_down-39): 205 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 441 us MUL_MAT_ID(ffn_moe_down-40): 251 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 401 us MUL_MAT_ID(ffn_moe_down-41): 420 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 397 us MUL_MAT_ID(ffn_moe_down-42): 260 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 380 us MUL_MAT_ID(ffn_moe_down-43): 203 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 431 us MUL_MAT_ID(ffn_moe_down-44): 266 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 441 us MUL_MAT_ID(ffn_moe_down-45): 214 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 397 us MUL_MAT_ID(ffn_moe_down-46): 210 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 419 us MUL_MAT_ID(ffn_moe_down-47): 244 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 486 us MUL_MAT_ID(ffn_moe_down-48): 204 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 398 us MUL_MAT_ID(ffn_moe_down-49): 204 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 375 us MUL_MAT_ID(ffn_moe_down-50): 167 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 370 us MUL_MAT_ID(ffn_moe_down-51): 202 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 469 us MUL_MAT_ID(ffn_moe_down-52): 213 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 409 us MUL_MAT_ID(ffn_moe_down-53): 236 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 397 us MUL_MAT_ID(ffn_moe_down-54): 183 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 426 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 387 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 440 us MUL_MAT_ID(ffn_moe_down-57): 171 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 418 us MUL_MAT_ID(ffn_moe_down-58): 305 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 418 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 404 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 456 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 475 us MUL_MAT_ID(ffn_moe_down-62): 182 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 416 us MUL_MAT_ID(ffn_moe_down-63): 200 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 471 us MUL_MAT_ID(ffn_moe_down-64): 183 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 417 us MUL_MAT_ID(ffn_moe_down-65): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 378 us MUL_MAT_ID(ffn_moe_down-66): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 348 us MUL_MAT_ID(ffn_moe_down-67): 215 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 202 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 403 us MUL_MAT_ID(ffn_moe_down-69): 206 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 378 us MUL_MAT_ID(ffn_moe_down-70): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 388 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 363 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 380 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 426 us MUL_MAT_ID(ffn_moe_down-75): 206 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 370 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 383 us MUL_MAT_ID(ffn_moe_down-77): 263 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 370 us MUL_MAT_ID(ffn_moe_down-78): 207 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 380 us MUL_MAT_ID(ffn_moe_down-79): 368 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 359 us MUL_MAT_ID(ffn_moe_down-80): 187 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 397 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 189 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 405 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 410 us MUL_MAT_ID(ffn_moe_down-84): 248 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 437 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 399 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 543 us MUL_MAT_ID(ffn_moe_down-87): 222 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 372 us MUL_MAT_ID(ffn_moe_down-88): 222 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 364 us MUL_MAT_ID(ffn_moe_down-89): 196 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 479 us MUL_MAT_ID(ffn_moe_down-90): 279 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 365 us MUL_MAT_ID(ffn_moe_down-91): 241 us ggml_barrier(...): 42 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 452 us MUL_MAT_ID(ffn_moe_down-25): 388 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 363 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 381 us MUL_MAT_ID(ffn_moe_down-27): 253 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 360 us MUL_MAT_ID(ffn_moe_down-28): 176 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 418 us MUL_MAT_ID(ffn_moe_down-29): 454 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 416 us MUL_MAT_ID(ffn_moe_down-30): 209 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 438 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 366 us MUL_MAT_ID(ffn_moe_down-32): 348 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 351 us MUL_MAT_ID(ffn_moe_down-33): 208 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 611 us MUL_MAT_ID(ffn_moe_down-35): 205 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 382 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 375 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 371 us MUL_MAT_ID(ffn_moe_down-38): 186 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 392 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 363 us MUL_MAT_ID(ffn_moe_down-40): 705 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 371 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 410 us MUL_MAT_ID(ffn_moe_down-42): 209 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 381 us MUL_MAT_ID(ffn_moe_down-43): 208 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 392 us MUL_MAT_ID(ffn_moe_down-44): 220 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 390 us MUL_MAT_ID(ffn_moe_down-45): 210 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 409 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 378 us MUL_MAT_ID(ffn_moe_down-47): 202 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 406 us MUL_MAT_ID(ffn_moe_down-48): 214 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 417 us MUL_MAT_ID(ffn_moe_down-49): 228 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 400 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 383 us MUL_MAT_ID(ffn_moe_down-51): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 371 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 366 us MUL_MAT_ID(ffn_moe_down-53): 424 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 364 us MUL_MAT_ID(ffn_moe_down-54): 203 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 186 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 386 us MUL_MAT_ID(ffn_moe_down-56): 171 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 394 us MUL_MAT_ID(ffn_moe_down-57): 184 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 397 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 385 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 374 us MUL_MAT_ID(ffn_moe_down-60): 184 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 371 us MUL_MAT_ID(ffn_moe_down-61): 179 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 366 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 381 us MUL_MAT_ID(ffn_moe_down-63): 205 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 393 us MUL_MAT_ID(ffn_moe_down-64): 540 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 367 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 359 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 413 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 393 us MUL_MAT_ID(ffn_moe_down-68): 526 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 389 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 455 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 442 us MUL_MAT_ID(ffn_moe_down-71): 174 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 364 us MUL_MAT_ID(ffn_moe_down-72): 232 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 459 us MUL_MAT_ID(ffn_moe_down-73): 271 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 375 us MUL_MAT_ID(ffn_moe_down-74): 203 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 373 us MUL_MAT_ID(ffn_moe_down-75): 206 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 416 us MUL_MAT_ID(ffn_moe_down-76): 244 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 402 us MUL_MAT_ID(ffn_moe_down-77): 187 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 390 us MUL_MAT_ID(ffn_moe_down-78): 368 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 382 us MUL_MAT_ID(ffn_moe_down-79): 343 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 397 us MUL_MAT_ID(ffn_moe_down-80): 218 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 404 us MUL_MAT_ID(ffn_moe_down-81): 178 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 392 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 368 us MUL_MAT_ID(ffn_moe_down-83): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 416 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 371 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 386 us MUL_MAT_ID(ffn_moe_down-86): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 374 us MUL_MAT_ID(ffn_moe_down-87): 277 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 513 us MUL_MAT_ID(ffn_moe_down-88): 216 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 381 us MUL_MAT_ID(ffn_moe_down-89): 215 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 379 us MUL_MAT_ID(ffn_moe_down-90): 227 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 379 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 46 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 441 us MUL_MAT_ID(ffn_moe_down-25): 207 us ggml_barrier(...): 412 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 399 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 444 us MUL_MAT_ID(ffn_moe_down-27): 155 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 405 us MUL_MAT_ID(ffn_moe_down-28): 187 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 394 us MUL_MAT_ID(ffn_moe_down-29): 187 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 399 us MUL_MAT_ID(ffn_moe_down-30): 209 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 506 us MUL_MAT_ID(ffn_moe_down-31): 201 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 406 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 395 us MUL_MAT_ID(ffn_moe_down-33): 212 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 387 us MUL_MAT_ID(ffn_moe_down-34): 188 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 383 us MUL_MAT_ID(ffn_moe_down-35): 189 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 393 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 490 us MUL_MAT_ID(ffn_moe_down-37): 284 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 475 us MUL_MAT_ID(ffn_moe_down-38): 173 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 354 us MUL_MAT_ID(ffn_moe_down-39): 231 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 425 us MUL_MAT_ID(ffn_moe_down-40): 366 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 415 us MUL_MAT_ID(ffn_moe_down-41): 644 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 484 us MUL_MAT_ID(ffn_moe_down-42): 207 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 406 us MUL_MAT_ID(ffn_moe_down-43): 198 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 364 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 387 us MUL_MAT_ID(ffn_moe_down-45): 193 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 414 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 391 us MUL_MAT_ID(ffn_moe_down-47): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 433 us MUL_MAT_ID(ffn_moe_down-48): 177 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 404 us MUL_MAT_ID(ffn_moe_down-49): 180 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 387 us MUL_MAT_ID(ffn_moe_down-50): 223 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 401 us MUL_MAT_ID(ffn_moe_down-51): 203 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 382 us MUL_MAT_ID(ffn_moe_down-52): 206 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 376 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 369 us MUL_MAT_ID(ffn_moe_down-54): 532 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 368 us MUL_MAT_ID(ffn_moe_down-55): 205 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 378 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 412 us MUL_MAT_ID(ffn_moe_down-57): 361 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 394 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 380 us MUL_MAT_ID(ffn_moe_down-59): 276 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 377 us MUL_MAT_ID(ffn_moe_down-60): 204 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 394 us MUL_MAT_ID(ffn_moe_down-61): 204 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 409 us MUL_MAT_ID(ffn_moe_down-62): 252 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 384 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 387 us MUL_MAT_ID(ffn_moe_down-64): 278 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 349 us MUL_MAT_ID(ffn_moe_down-65): 190 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 383 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 367 us MUL_MAT_ID(ffn_moe_down-67): 200 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 345 us MUL_MAT_ID(ffn_moe_down-68): 204 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 398 us MUL_MAT_ID(ffn_moe_down-69): 238 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 376 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 461 us MUL_MAT_ID(ffn_moe_down-71): 201 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 424 us MUL_MAT_ID(ffn_moe_down-72): 231 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 358 us MUL_MAT_ID(ffn_moe_down-73): 210 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 677 us MUL_MAT_ID(ffn_moe_down-74): 500 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 378 us MUL_MAT_ID(ffn_moe_down-75): 215 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 494 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 362 us MUL_MAT_ID(ffn_moe_down-77): 256 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 417 us MUL_MAT_ID(ffn_moe_down-78): 396 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 377 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 418 us MUL_MAT_ID(ffn_moe_down-80): 231 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 473 us MUL_MAT_ID(ffn_moe_down-81): 163 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 490 us MUL_MAT_ID(ffn_moe_down-82): 215 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 407 us MUL_MAT_ID(ffn_moe_down-83): 417 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 401 us MUL_MAT_ID(ffn_moe_down-84): 235 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 367 us MUL_MAT_ID(ffn_moe_down-85): 239 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 389 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 379 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 370 us MUL_MAT_ID(ffn_moe_down-88): 191 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 363 us MUL_MAT_ID(ffn_moe_down-89): 235 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 433 us MUL_MAT_ID(ffn_moe_down-90): 256 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 398 us MUL_MAT_ID(ffn_moe_down-91): 408 us ggml_barrier(...): 26 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 365 us MUL_MAT_ID(ffn_moe_down-25): 191 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 377 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 491 us MUL_MAT_ID(ffn_moe_down-27): 176 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 385 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 445 us MUL_MAT_ID(ffn_moe_down-29): 180 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 370 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 405 us MUL_MAT_ID(ffn_moe_down-31): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 398 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 363 us MUL_MAT_ID(ffn_moe_down-33): 202 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 375 us MUL_MAT_ID(ffn_moe_down-34): 210 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 367 us MUL_MAT_ID(ffn_moe_down-35): 208 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 362 us MUL_MAT_ID(ffn_moe_down-36): 250 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 359 us MUL_MAT_ID(ffn_moe_down-37): 250 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 384 us MUL_MAT_ID(ffn_moe_down-38): 183 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 381 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 395 us MUL_MAT_ID(ffn_moe_down-40): 238 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 368 us MUL_MAT_ID(ffn_moe_down-41): 225 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 464 us MUL_MAT_ID(ffn_moe_down-42): 210 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 446 us MUL_MAT_ID(ffn_moe_down-43): 184 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 392 us MUL_MAT_ID(ffn_moe_down-44): 213 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 366 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 381 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 387 us MUL_MAT_ID(ffn_moe_down-47): 198 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 371 us MUL_MAT_ID(ffn_moe_down-48): 178 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 375 us MUL_MAT_ID(ffn_moe_down-49): 504 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 347 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 443 us MUL_MAT_ID(ffn_moe_down-51): 289 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 389 us MUL_MAT_ID(ffn_moe_down-52): 197 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 353 us MUL_MAT_ID(ffn_moe_down-53): 187 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 391 us MUL_MAT_ID(ffn_moe_down-54): 203 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 391 us MUL_MAT_ID(ffn_moe_down-55): 233 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 395 us MUL_MAT_ID(ffn_moe_down-56): 206 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 416 us MUL_MAT_ID(ffn_moe_down-57): 325 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 400 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 359 us MUL_MAT_ID(ffn_moe_down-59): 383 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 422 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 369 us MUL_MAT_ID(ffn_moe_down-61): 340 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 537 us MUL_MAT_ID(ffn_moe_down-62): 487 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 382 us MUL_MAT_ID(ffn_moe_down-63): 203 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 383 us MUL_MAT_ID(ffn_moe_down-64): 187 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 356 us MUL_MAT_ID(ffn_moe_down-65): 231 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 354 us MUL_MAT_ID(ffn_moe_down-66): 420 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 386 us MUL_MAT_ID(ffn_moe_down-67): 361 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 504 us MUL_MAT_ID(ffn_moe_down-68): 232 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 394 us MUL_MAT_ID(ffn_moe_down-69): 186 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 404 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 389 us MUL_MAT_ID(ffn_moe_down-71): 208 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 360 us MUL_MAT_ID(ffn_moe_down-72): 547 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 422 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 386 us MUL_MAT_ID(ffn_moe_down-74): 194 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 345 us MUL_MAT_ID(ffn_moe_down-75): 244 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 386 us MUL_MAT_ID(ffn_moe_down-76): 204 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 370 us MUL_MAT_ID(ffn_moe_down-77): 216 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 460 us MUL_MAT_ID(ffn_moe_down-78): 206 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 435 us MUL_MAT_ID(ffn_moe_down-79): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 366 us MUL_MAT_ID(ffn_moe_down-80): 246 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 386 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 360 us MUL_MAT_ID(ffn_moe_down-82): 201 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 403 us MUL_MAT_ID(ffn_moe_down-83): 384 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 385 us MUL_MAT_ID(ffn_moe_down-84): 224 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 380 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 361 us MUL_MAT_ID(ffn_moe_down-86): 189 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 427 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 408 us MUL_MAT_ID(ffn_moe_down-88): 230 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 387 us MUL_MAT_ID(ffn_moe_down-89): 222 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 498 us MUL_MAT_ID(ffn_moe_down-90): 257 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 380 us MUL_MAT_ID(ffn_moe_down-91): 241 us ggml_barrier(...): 235 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 399 us MUL_MAT_ID(ffn_moe_down-25): 203 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 384 us MUL_MAT_ID(ffn_moe_down-26): 560 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 387 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 372 us MUL_MAT_ID(ffn_moe_down-28): 216 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 507 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 490 us MUL_MAT_ID(ffn_moe_down-30): 207 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 398 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 389 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 381 us MUL_MAT_ID(ffn_moe_down-33): 202 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 375 us MUL_MAT_ID(ffn_moe_down-34): 205 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 370 us MUL_MAT_ID(ffn_moe_down-35): 184 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 361 us MUL_MAT_ID(ffn_moe_down-36): 224 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 647 us MUL_MAT_ID(ffn_moe_down-37): 336 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 374 us MUL_MAT_ID(ffn_moe_down-38): 331 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 371 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 346 us MUL_MAT_ID(ffn_moe_down-40): 193 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 390 us MUL_MAT_ID(ffn_moe_down-41): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 411 us MUL_MAT_ID(ffn_moe_down-42): 406 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 393 us MUL_MAT_ID(ffn_moe_down-43): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 352 us MUL_MAT_ID(ffn_moe_down-44): 289 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 377 us MUL_MAT_ID(ffn_moe_down-45): 191 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 390 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 378 us MUL_MAT_ID(ffn_moe_down-47): 182 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 351 us MUL_MAT_ID(ffn_moe_down-48): 181 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 384 us MUL_MAT_ID(ffn_moe_down-49): 184 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 372 us MUL_MAT_ID(ffn_moe_down-50): 182 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 346 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 516 us MUL_MAT_ID(ffn_moe_down-52): 286 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 426 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 358 us MUL_MAT_ID(ffn_moe_down-54): 204 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 385 us MUL_MAT_ID(ffn_moe_down-55): 273 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 385 us MUL_MAT_ID(ffn_moe_down-56): 180 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 350 us MUL_MAT_ID(ffn_moe_down-57): 204 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 364 us MUL_MAT_ID(ffn_moe_down-58): 280 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 404 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 428 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 376 us MUL_MAT_ID(ffn_moe_down-61): 212 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 377 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 202 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 381 us MUL_MAT_ID(ffn_moe_down-64): 208 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 345 us MUL_MAT_ID(ffn_moe_down-65): 184 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 392 us MUL_MAT_ID(ffn_moe_down-66): 207 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 484 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 384 us MUL_MAT_ID(ffn_moe_down-68): 190 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 365 us MUL_MAT_ID(ffn_moe_down-69): 177 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 363 us MUL_MAT_ID(ffn_moe_down-70): 217 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 351 us MUL_MAT_ID(ffn_moe_down-71): 210 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 412 us MUL_MAT_ID(ffn_moe_down-72): 188 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 358 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 378 us MUL_MAT_ID(ffn_moe_down-74): 215 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 410 us MUL_MAT_ID(ffn_moe_down-75): 203 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 393 us MUL_MAT_ID(ffn_moe_down-76): 489 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 381 us MUL_MAT_ID(ffn_moe_down-77): 291 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 360 us MUL_MAT_ID(ffn_moe_down-78): 204 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 386 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 500 us MUL_MAT_ID(ffn_moe_down-80): 303 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 395 us MUL_MAT_ID(ffn_moe_down-81): 180 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 349 us MUL_MAT_ID(ffn_moe_down-82): 182 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 407 us MUL_MAT_ID(ffn_moe_down-83): 192 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 353 us MUL_MAT_ID(ffn_moe_down-84): 240 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 496 us MUL_MAT_ID(ffn_moe_down-85): 195 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 384 us MUL_MAT_ID(ffn_moe_down-86): 189 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 394 us MUL_MAT_ID(ffn_moe_down-87): 529 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 370 us MUL_MAT_ID(ffn_moe_down-88): 184 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 358 us MUL_MAT_ID(ffn_moe_down-89): 191 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 349 us MUL_MAT_ID(ffn_moe_down-90): 226 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 358 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 69 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 414 us MUL_MAT_ID(ffn_moe_down-25): 188 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 387 us MUL_MAT_ID(ffn_moe_down-26): 294 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 416 us MUL_MAT_ID(ffn_moe_down-27): 303 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 378 us MUL_MAT_ID(ffn_moe_down-28): 342 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 371 us MUL_MAT_ID(ffn_moe_down-29): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 402 us MUL_MAT_ID(ffn_moe_down-30): 218 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 361 us MUL_MAT_ID(ffn_moe_down-31): 248 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 415 us MUL_MAT_ID(ffn_moe_down-32): 185 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 366 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 390 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 411 us MUL_MAT_ID(ffn_moe_down-35): 228 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 472 us MUL_MAT_ID(ffn_moe_down-36): 181 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 485 us MUL_MAT_ID(ffn_moe_down-37): 342 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 398 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 396 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 406 us MUL_MAT_ID(ffn_moe_down-40): 406 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 381 us MUL_MAT_ID(ffn_moe_down-41): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 362 us MUL_MAT_ID(ffn_moe_down-42): 189 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 492 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 386 us MUL_MAT_ID(ffn_moe_down-44): 355 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 367 us MUL_MAT_ID(ffn_moe_down-45): 220 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 381 us MUL_MAT_ID(ffn_moe_down-46): 188 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 403 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 410 us MUL_MAT_ID(ffn_moe_down-48): 363 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 366 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 410 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 398 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 372 us MUL_MAT_ID(ffn_moe_down-52): 188 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 442 us MUL_MAT_ID(ffn_moe_down-53): 241 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 352 us MUL_MAT_ID(ffn_moe_down-54): 213 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 394 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 207 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 352 us MUL_MAT_ID(ffn_moe_down-57): 202 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 378 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 395 us MUL_MAT_ID(ffn_moe_down-59): 240 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 443 us MUL_MAT_ID(ffn_moe_down-60): 165 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 435 us MUL_MAT_ID(ffn_moe_down-61): 280 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 372 us MUL_MAT_ID(ffn_moe_down-62): 350 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 353 us MUL_MAT_ID(ffn_moe_down-63): 213 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 377 us MUL_MAT_ID(ffn_moe_down-64): 200 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 369 us MUL_MAT_ID(ffn_moe_down-65): 179 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 368 us MUL_MAT_ID(ffn_moe_down-66): 208 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 356 us MUL_MAT_ID(ffn_moe_down-67): 187 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 382 us MUL_MAT_ID(ffn_moe_down-68): 189 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 368 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 404 us MUL_MAT_ID(ffn_moe_down-70): 581 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 367 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 369 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 389 us MUL_MAT_ID(ffn_moe_down-73): 210 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 359 us MUL_MAT_ID(ffn_moe_down-74): 216 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 348 us MUL_MAT_ID(ffn_moe_down-75): 190 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 381 us MUL_MAT_ID(ffn_moe_down-76): 196 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 422 us MUL_MAT_ID(ffn_moe_down-77): 208 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 380 us MUL_MAT_ID(ffn_moe_down-78): 505 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 350 us MUL_MAT_ID(ffn_moe_down-79): 174 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 377 us MUL_MAT_ID(ffn_moe_down-80): 205 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 364 us MUL_MAT_ID(ffn_moe_down-81): 430 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 396 us MUL_MAT_ID(ffn_moe_down-82): 210 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 391 us MUL_MAT_ID(ffn_moe_down-83): 223 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 371 us MUL_MAT_ID(ffn_moe_down-84): 248 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 357 us MUL_MAT_ID(ffn_moe_down-85): 224 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 396 us MUL_MAT_ID(ffn_moe_down-86): 205 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 368 us MUL_MAT_ID(ffn_moe_down-87): 217 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 403 us MUL_MAT_ID(ffn_moe_down-88): 371 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 400 us MUL_MAT_ID(ffn_moe_down-89): 203 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 361 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 391 us MUL_MAT_ID(ffn_moe_down-91): 251 us ggml_barrier(...): 21 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 392 us MUL_MAT_ID(ffn_moe_down-25): 181 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 390 us MUL_MAT_ID(ffn_moe_down-26): 207 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 355 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 352 us MUL_MAT_ID(ffn_moe_down-28): 196 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 360 us MUL_MAT_ID(ffn_moe_down-29): 265 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 358 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 362 us MUL_MAT_ID(ffn_moe_down-31): 190 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 372 us MUL_MAT_ID(ffn_moe_down-32): 226 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 379 us MUL_MAT_ID(ffn_moe_down-33): 204 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 456 us MUL_MAT_ID(ffn_moe_down-34): 174 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 351 us MUL_MAT_ID(ffn_moe_down-35): 197 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 371 us MUL_MAT_ID(ffn_moe_down-36): 199 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 401 us MUL_MAT_ID(ffn_moe_down-37): 293 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 339 us MUL_MAT_ID(ffn_moe_down-38): 243 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 406 us MUL_MAT_ID(ffn_moe_down-39): 526 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 344 us MUL_MAT_ID(ffn_moe_down-40): 722 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 410 us MUL_MAT_ID(ffn_moe_down-41): 193 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 393 us MUL_MAT_ID(ffn_moe_down-42): 195 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 407 us MUL_MAT_ID(ffn_moe_down-43): 549 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 402 us MUL_MAT_ID(ffn_moe_down-44): 215 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 417 us MUL_MAT_ID(ffn_moe_down-45): 254 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 414 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 402 us MUL_MAT_ID(ffn_moe_down-47): 373 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 330 us MUL_MAT_ID(ffn_moe_down-48): 183 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 394 us MUL_MAT_ID(ffn_moe_down-49): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 392 us MUL_MAT_ID(ffn_moe_down-50): 338 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 411 us MUL_MAT_ID(ffn_moe_down-51): 209 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 389 us MUL_MAT_ID(ffn_moe_down-52): 207 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 395 us MUL_MAT_ID(ffn_moe_down-53): 252 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 453 us MUL_MAT_ID(ffn_moe_down-54): 218 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 369 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 409 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 379 us MUL_MAT_ID(ffn_moe_down-57): 187 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 369 us MUL_MAT_ID(ffn_moe_down-58): 463 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 393 us MUL_MAT_ID(ffn_moe_down-59): 184 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 344 us MUL_MAT_ID(ffn_moe_down-60): 260 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 403 us MUL_MAT_ID(ffn_moe_down-61): 240 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 403 us MUL_MAT_ID(ffn_moe_down-62): 289 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 374 us MUL_MAT_ID(ffn_moe_down-63): 199 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 393 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 370 us MUL_MAT_ID(ffn_moe_down-65): 193 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 406 us MUL_MAT_ID(ffn_moe_down-66): 395 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 392 us MUL_MAT_ID(ffn_moe_down-67): 216 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 446 us MUL_MAT_ID(ffn_moe_down-68): 383 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 382 us MUL_MAT_ID(ffn_moe_down-69): 213 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 399 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 405 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 347 us MUL_MAT_ID(ffn_moe_down-72): 193 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 366 us MUL_MAT_ID(ffn_moe_down-73): 302 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 427 us MUL_MAT_ID(ffn_moe_down-74): 533 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 368 us MUL_MAT_ID(ffn_moe_down-75): 335 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 371 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 366 us MUL_MAT_ID(ffn_moe_down-77): 236 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 392 us MUL_MAT_ID(ffn_moe_down-78): 221 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 398 us MUL_MAT_ID(ffn_moe_down-79): 187 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 398 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 398 us MUL_MAT_ID(ffn_moe_down-82): 212 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 412 us MUL_MAT_ID(ffn_moe_down-83): 220 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 367 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 394 us MUL_MAT_ID(ffn_moe_down-85): 234 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 425 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 368 us MUL_MAT_ID(ffn_moe_down-87): 216 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 377 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 371 us MUL_MAT_ID(ffn_moe_down-89): 196 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 381 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 630 us MUL_MAT_ID(ffn_moe_down-91): 270 us ggml_barrier(...): 37 us GET_ROWS(inp_embd): 22 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 391 us MUL_MAT_ID(ffn_moe_down-25): 356 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 420 us MUL_MAT_ID(ffn_moe_down-26): 232 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 430 us MUL_MAT_ID(ffn_moe_down-27): 244 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 409 us MUL_MAT_ID(ffn_moe_down-28): 217 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 399 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 416 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 485 us MUL_MAT_ID(ffn_moe_down-31): 271 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 376 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 374 us MUL_MAT_ID(ffn_moe_down-33): 196 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 434 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 376 us MUL_MAT_ID(ffn_moe_down-35): 205 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 435 us MUL_MAT_ID(ffn_moe_down-36): 189 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 408 us MUL_MAT_ID(ffn_moe_down-37): 242 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 377 us MUL_MAT_ID(ffn_moe_down-38): 197 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 374 us MUL_MAT_ID(ffn_moe_down-39): 249 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 472 us MUL_MAT_ID(ffn_moe_down-40): 211 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 417 us MUL_MAT_ID(ffn_moe_down-41): 266 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 420 us MUL_MAT_ID(ffn_moe_down-42): 417 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 395 us MUL_MAT_ID(ffn_moe_down-43): 211 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 381 us MUL_MAT_ID(ffn_moe_down-44): 520 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 477 us MUL_MAT_ID(ffn_moe_down-45): 423 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 399 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 491 us MUL_MAT_ID(ffn_moe_down-47): 259 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 419 us MUL_MAT_ID(ffn_moe_down-48): 360 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 385 us MUL_MAT_ID(ffn_moe_down-49): 228 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 467 us MUL_MAT_ID(ffn_moe_down-50): 214 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 403 us MUL_MAT_ID(ffn_moe_down-51): 363 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 371 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 406 us MUL_MAT_ID(ffn_moe_down-53): 267 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 366 us MUL_MAT_ID(ffn_moe_down-54): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 390 us MUL_MAT_ID(ffn_moe_down-55): 174 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 400 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 399 us MUL_MAT_ID(ffn_moe_down-57): 346 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 402 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 433 us MUL_MAT_ID(ffn_moe_down-59): 212 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 494 us MUL_MAT_ID(ffn_moe_down-60): 225 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 398 us MUL_MAT_ID(ffn_moe_down-61): 200 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 366 us MUL_MAT_ID(ffn_moe_down-62): 347 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 398 us MUL_MAT_ID(ffn_moe_down-63): 207 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 361 us MUL_MAT_ID(ffn_moe_down-64): 292 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 454 us MUL_MAT_ID(ffn_moe_down-65): 361 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 426 us MUL_MAT_ID(ffn_moe_down-66): 391 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 387 us MUL_MAT_ID(ffn_moe_down-67): 393 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 388 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 349 us MUL_MAT_ID(ffn_moe_down-69): 368 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 427 us MUL_MAT_ID(ffn_moe_down-70): 169 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 391 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 349 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 488 us MUL_MAT_ID(ffn_moe_down-73): 425 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 384 us MUL_MAT_ID(ffn_moe_down-74): 200 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 374 us MUL_MAT_ID(ffn_moe_down-75): 200 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 389 us MUL_MAT_ID(ffn_moe_down-76): 198 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 357 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 355 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 351 us MUL_MAT_ID(ffn_moe_down-79): 415 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 339 us MUL_MAT_ID(ffn_moe_down-80): 193 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 397 us MUL_MAT_ID(ffn_moe_down-81): 289 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 448 us MUL_MAT_ID(ffn_moe_down-82): 173 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 359 us MUL_MAT_ID(ffn_moe_down-83): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 400 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 383 us MUL_MAT_ID(ffn_moe_down-85): 258 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 382 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 397 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 370 us MUL_MAT_ID(ffn_moe_down-88): 209 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 374 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 379 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 394 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 23 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 380 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 381 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 369 us MUL_MAT_ID(ffn_moe_down-27): 372 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 406 us MUL_MAT_ID(ffn_moe_down-28): 200 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 379 us MUL_MAT_ID(ffn_moe_down-29): 209 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 366 us MUL_MAT_ID(ffn_moe_down-30): 331 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 382 us MUL_MAT_ID(ffn_moe_down-31): 200 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 411 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 399 us MUL_MAT_ID(ffn_moe_down-33): 193 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 391 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 449 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 383 us MUL_MAT_ID(ffn_moe_down-36): 208 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 394 us MUL_MAT_ID(ffn_moe_down-37): 312 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 396 us MUL_MAT_ID(ffn_moe_down-38): 184 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 403 us MUL_MAT_ID(ffn_moe_down-39): 203 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 361 us MUL_MAT_ID(ffn_moe_down-40): 244 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 386 us MUL_MAT_ID(ffn_moe_down-41): 513 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 409 us MUL_MAT_ID(ffn_moe_down-42): 251 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 384 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 384 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 386 us MUL_MAT_ID(ffn_moe_down-45): 218 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 369 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 376 us MUL_MAT_ID(ffn_moe_down-47): 208 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 392 us MUL_MAT_ID(ffn_moe_down-48): 412 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 377 us MUL_MAT_ID(ffn_moe_down-49): 201 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 388 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 371 us MUL_MAT_ID(ffn_moe_down-51): 189 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 626 us MUL_MAT_ID(ffn_moe_down-52): 211 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 373 us MUL_MAT_ID(ffn_moe_down-53): 187 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 546 us MUL_MAT_ID(ffn_moe_down-54): 179 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 387 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 433 us MUL_MAT_ID(ffn_moe_down-56): 273 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 365 us MUL_MAT_ID(ffn_moe_down-57): 234 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 405 us MUL_MAT_ID(ffn_moe_down-58): 189 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 401 us MUL_MAT_ID(ffn_moe_down-59): 293 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 343 us MUL_MAT_ID(ffn_moe_down-60): 244 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 453 us MUL_MAT_ID(ffn_moe_down-61): 194 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 379 us MUL_MAT_ID(ffn_moe_down-62): 331 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 639 us MUL_MAT_ID(ffn_moe_down-63): 240 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 385 us MUL_MAT_ID(ffn_moe_down-64): 284 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 489 us MUL_MAT_ID(ffn_moe_down-65): 219 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 448 us MUL_MAT_ID(ffn_moe_down-66): 239 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 492 us MUL_MAT_ID(ffn_moe_down-67): 314 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 396 us MUL_MAT_ID(ffn_moe_down-68): 206 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 389 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 368 us MUL_MAT_ID(ffn_moe_down-70): 186 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 326 us MUL_MAT_ID(ffn_moe_down-71): 295 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 453 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 382 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 429 us MUL_MAT_ID(ffn_moe_down-74): 183 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 499 us MUL_MAT_ID(ffn_moe_down-75): 181 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 414 us MUL_MAT_ID(ffn_moe_down-76): 385 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 400 us MUL_MAT_ID(ffn_moe_down-77): 375 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 365 us MUL_MAT_ID(ffn_moe_down-78): 289 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 356 us MUL_MAT_ID(ffn_moe_down-79): 244 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 352 us MUL_MAT_ID(ffn_moe_down-80): 538 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 382 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 383 us MUL_MAT_ID(ffn_moe_down-82): 191 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 423 us MUL_MAT_ID(ffn_moe_down-83): 204 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 358 us MUL_MAT_ID(ffn_moe_down-84): 280 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 341 us MUL_MAT_ID(ffn_moe_down-85): 249 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 359 us MUL_MAT_ID(ffn_moe_down-86): 471 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 400 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 483 us MUL_MAT_ID(ffn_moe_down-88): 203 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 417 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 380 us MUL_MAT_ID(ffn_moe_down-90): 226 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 459 us MUL_MAT_ID(ffn_moe_down-91): 295 us ggml_barrier(...): 66 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 420 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 500 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 389 us MUL_MAT_ID(ffn_moe_down-27): 218 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 385 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 391 us MUL_MAT_ID(ffn_moe_down-29): 292 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 404 us MUL_MAT_ID(ffn_moe_down-30): 201 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 383 us MUL_MAT_ID(ffn_moe_down-31): 396 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 368 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 387 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 397 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 363 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 417 us MUL_MAT_ID(ffn_moe_down-36): 217 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 367 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 380 us MUL_MAT_ID(ffn_moe_down-38): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 367 us MUL_MAT_ID(ffn_moe_down-39): 243 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 366 us MUL_MAT_ID(ffn_moe_down-40): 428 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 386 us MUL_MAT_ID(ffn_moe_down-41): 213 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 491 us MUL_MAT_ID(ffn_moe_down-42): 415 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 401 us MUL_MAT_ID(ffn_moe_down-43): 252 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 367 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 391 us MUL_MAT_ID(ffn_moe_down-45): 192 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 357 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 405 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 394 us MUL_MAT_ID(ffn_moe_down-48): 246 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 354 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 380 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 386 us MUL_MAT_ID(ffn_moe_down-51): 187 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 371 us MUL_MAT_ID(ffn_moe_down-52): 178 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 401 us MUL_MAT_ID(ffn_moe_down-53): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 392 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 337 us MUL_MAT_ID(ffn_moe_down-55): 180 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 394 us MUL_MAT_ID(ffn_moe_down-56): 209 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 387 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 376 us MUL_MAT_ID(ffn_moe_down-58): 188 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 363 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 346 us MUL_MAT_ID(ffn_moe_down-60): 193 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 373 us MUL_MAT_ID(ffn_moe_down-61): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 356 us MUL_MAT_ID(ffn_moe_down-62): 198 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 395 us MUL_MAT_ID(ffn_moe_down-63): 247 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 346 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 370 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 383 us MUL_MAT_ID(ffn_moe_down-66): 204 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 391 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 369 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 402 us MUL_MAT_ID(ffn_moe_down-69): 359 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 384 us MUL_MAT_ID(ffn_moe_down-70): 220 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 376 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 383 us MUL_MAT_ID(ffn_moe_down-72): 193 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 345 us MUL_MAT_ID(ffn_moe_down-73): 195 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 370 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 358 us MUL_MAT_ID(ffn_moe_down-75): 538 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 531 us MUL_MAT_ID(ffn_moe_down-76): 207 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 407 us MUL_MAT_ID(ffn_moe_down-77): 208 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 418 us MUL_MAT_ID(ffn_moe_down-78): 235 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 423 us MUL_MAT_ID(ffn_moe_down-79): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 478 us MUL_MAT_ID(ffn_moe_down-80): 301 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 366 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 421 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 363 us MUL_MAT_ID(ffn_moe_down-83): 397 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 374 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 373 us MUL_MAT_ID(ffn_moe_down-85): 314 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 357 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 426 us MUL_MAT_ID(ffn_moe_down-87): 262 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 360 us MUL_MAT_ID(ffn_moe_down-88): 221 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 377 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 402 us MUL_MAT_ID(ffn_moe_down-90): 229 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 400 us MUL_MAT_ID(ffn_moe_down-91): 248 us ggml_barrier(...): 35 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 480 us MUL_MAT_ID(ffn_moe_down-25): 169 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 409 us MUL_MAT_ID(ffn_moe_down-26): 220 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 396 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 415 us MUL_MAT_ID(ffn_moe_down-28): 176 us ggml_barrier(...): 334 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 408 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 225 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 411 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 379 us MUL_MAT_ID(ffn_moe_down-32): 276 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 442 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 389 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 485 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 405 us MUL_MAT_ID(ffn_moe_down-36): 212 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 478 us MUL_MAT_ID(ffn_moe_down-37): 266 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 398 us MUL_MAT_ID(ffn_moe_down-38): 218 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 359 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 389 us MUL_MAT_ID(ffn_moe_down-40): 439 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 372 us MUL_MAT_ID(ffn_moe_down-41): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 412 us MUL_MAT_ID(ffn_moe_down-42): 206 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 399 us MUL_MAT_ID(ffn_moe_down-43): 211 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 444 us MUL_MAT_ID(ffn_moe_down-44): 274 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 374 us MUL_MAT_ID(ffn_moe_down-45): 386 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 368 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 372 us MUL_MAT_ID(ffn_moe_down-47): 217 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 387 us MUL_MAT_ID(ffn_moe_down-48): 225 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 383 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 366 us MUL_MAT_ID(ffn_moe_down-50): 210 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 404 us MUL_MAT_ID(ffn_moe_down-51): 170 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 461 us MUL_MAT_ID(ffn_moe_down-52): 176 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 356 us MUL_MAT_ID(ffn_moe_down-53): 186 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 354 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 428 us MUL_MAT_ID(ffn_moe_down-55): 224 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 451 us MUL_MAT_ID(ffn_moe_down-56): 172 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 353 us MUL_MAT_ID(ffn_moe_down-57): 175 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 372 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 369 us MUL_MAT_ID(ffn_moe_down-59): 444 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 380 us MUL_MAT_ID(ffn_moe_down-60): 213 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 451 us MUL_MAT_ID(ffn_moe_down-61): 299 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 364 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 356 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 355 us MUL_MAT_ID(ffn_moe_down-64): 230 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 440 us MUL_MAT_ID(ffn_moe_down-65): 509 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 365 us MUL_MAT_ID(ffn_moe_down-66): 372 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 385 us MUL_MAT_ID(ffn_moe_down-67): 253 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 359 us MUL_MAT_ID(ffn_moe_down-68): 189 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 424 us MUL_MAT_ID(ffn_moe_down-69): 395 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 356 us MUL_MAT_ID(ffn_moe_down-70): 176 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 381 us MUL_MAT_ID(ffn_moe_down-71): 225 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 175 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 401 us MUL_MAT_ID(ffn_moe_down-73): 200 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 415 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 359 us MUL_MAT_ID(ffn_moe_down-75): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 380 us MUL_MAT_ID(ffn_moe_down-76): 187 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 364 us MUL_MAT_ID(ffn_moe_down-77): 190 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 351 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 381 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 375 us MUL_MAT_ID(ffn_moe_down-80): 206 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 364 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 361 us MUL_MAT_ID(ffn_moe_down-82): 432 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 449 us MUL_MAT_ID(ffn_moe_down-83): 210 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 460 us MUL_MAT_ID(ffn_moe_down-84): 213 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 405 us MUL_MAT_ID(ffn_moe_down-85): 630 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 388 us MUL_MAT_ID(ffn_moe_down-86): 247 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 368 us MUL_MAT_ID(ffn_moe_down-87): 207 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 396 us MUL_MAT_ID(ffn_moe_down-88): 189 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 389 us MUL_MAT_ID(ffn_moe_down-89): 415 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 413 us MUL_MAT_ID(ffn_moe_down-90): 213 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 403 us MUL_MAT_ID(ffn_moe_down-91): 310 us ggml_barrier(...): 134 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 464 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 474 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 405 us MUL_MAT_ID(ffn_moe_down-26): 210 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 401 us MUL_MAT_ID(ffn_moe_down-27): 200 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 399 us MUL_MAT_ID(ffn_moe_down-28): 195 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 427 us MUL_MAT_ID(ffn_moe_down-29): 203 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 378 us MUL_MAT_ID(ffn_moe_down-30): 389 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 394 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 483 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 365 us MUL_MAT_ID(ffn_moe_down-33): 293 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 402 us MUL_MAT_ID(ffn_moe_down-34): 295 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 369 us MUL_MAT_ID(ffn_moe_down-35): 311 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 457 us MUL_MAT_ID(ffn_moe_down-36): 226 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 407 us MUL_MAT_ID(ffn_moe_down-37): 186 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 396 us MUL_MAT_ID(ffn_moe_down-38): 202 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 377 us MUL_MAT_ID(ffn_moe_down-39): 200 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 388 us MUL_MAT_ID(ffn_moe_down-40): 530 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 362 us MUL_MAT_ID(ffn_moe_down-41): 322 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 384 us MUL_MAT_ID(ffn_moe_down-42): 396 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 384 us MUL_MAT_ID(ffn_moe_down-43): 178 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 407 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 369 us MUL_MAT_ID(ffn_moe_down-45): 178 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 345 us MUL_MAT_ID(ffn_moe_down-46): 211 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 390 us MUL_MAT_ID(ffn_moe_down-47): 355 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 371 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 414 us MUL_MAT_ID(ffn_moe_down-49): 251 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 401 us MUL_MAT_ID(ffn_moe_down-50): 424 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 465 us MUL_MAT_ID(ffn_moe_down-51): 181 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 421 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 322 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 639 us MUL_MAT_ID(ffn_moe_down-53): 211 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 389 us MUL_MAT_ID(ffn_moe_down-54): 219 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 480 us MUL_MAT_ID(ffn_moe_down-55): 169 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 492 us MUL_MAT_ID(ffn_moe_down-56): 231 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 450 us MUL_MAT_ID(ffn_moe_down-57): 173 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 457 us MUL_MAT_ID(ffn_moe_down-58): 287 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 372 us MUL_MAT_ID(ffn_moe_down-59): 208 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 375 us MUL_MAT_ID(ffn_moe_down-60): 281 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 382 us MUL_MAT_ID(ffn_moe_down-61): 183 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 348 us MUL_MAT_ID(ffn_moe_down-62): 246 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 553 us MUL_MAT_ID(ffn_moe_down-63): 340 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 368 us MUL_MAT_ID(ffn_moe_down-64): 443 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 375 us MUL_MAT_ID(ffn_moe_down-65): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 380 us MUL_MAT_ID(ffn_moe_down-66): 213 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 500 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 449 us MUL_MAT_ID(ffn_moe_down-68): 173 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 367 us MUL_MAT_ID(ffn_moe_down-69): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 396 us MUL_MAT_ID(ffn_moe_down-70): 231 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 386 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 403 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 448 us MUL_MAT_ID(ffn_moe_down-73): 293 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 389 us MUL_MAT_ID(ffn_moe_down-74): 228 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 428 us MUL_MAT_ID(ffn_moe_down-75): 206 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 383 us MUL_MAT_ID(ffn_moe_down-76): 183 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 513 us MUL_MAT_ID(ffn_moe_down-77): 448 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 390 us MUL_MAT_ID(ffn_moe_down-78): 223 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 384 us MUL_MAT_ID(ffn_moe_down-79): 203 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 401 us MUL_MAT_ID(ffn_moe_down-80): 275 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 387 us MUL_MAT_ID(ffn_moe_down-81): 256 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 382 us MUL_MAT_ID(ffn_moe_down-82): 349 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 385 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 393 us MUL_MAT_ID(ffn_moe_down-84): 625 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 456 us MUL_MAT_ID(ffn_moe_down-85): 213 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 396 us MUL_MAT_ID(ffn_moe_down-86): 236 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 465 us MUL_MAT_ID(ffn_moe_down-87): 383 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 383 us MUL_MAT_ID(ffn_moe_down-88): 557 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 379 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 400 us MUL_MAT_ID(ffn_moe_down-90): 230 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 509 us MUL_MAT_ID(ffn_moe_down-91): 197 us ggml_barrier(...): 128 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 388 us MUL_MAT_ID(ffn_moe_down-26): 200 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 402 us MUL_MAT_ID(ffn_moe_down-27): 208 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 371 us MUL_MAT_ID(ffn_moe_down-28): 238 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 398 us MUL_MAT_ID(ffn_moe_down-29): 222 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 414 us MUL_MAT_ID(ffn_moe_down-30): 218 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 405 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 412 us MUL_MAT_ID(ffn_moe_down-32): 278 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 487 us MUL_MAT_ID(ffn_moe_down-33): 191 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 365 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 421 us MUL_MAT_ID(ffn_moe_down-35): 182 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 376 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 379 us MUL_MAT_ID(ffn_moe_down-37): 395 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 379 us MUL_MAT_ID(ffn_moe_down-38): 357 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 417 us MUL_MAT_ID(ffn_moe_down-39): 524 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 378 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 397 us MUL_MAT_ID(ffn_moe_down-41): 208 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 387 us MUL_MAT_ID(ffn_moe_down-42): 192 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 378 us MUL_MAT_ID(ffn_moe_down-43): 192 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 392 us MUL_MAT_ID(ffn_moe_down-44): 366 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 393 us MUL_MAT_ID(ffn_moe_down-45): 221 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 357 us MUL_MAT_ID(ffn_moe_down-46): 225 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 486 us MUL_MAT_ID(ffn_moe_down-47): 270 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 398 us MUL_MAT_ID(ffn_moe_down-48): 183 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 366 us MUL_MAT_ID(ffn_moe_down-49): 205 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 389 us MUL_MAT_ID(ffn_moe_down-50): 531 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 344 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 348 us MUL_MAT_ID(ffn_moe_down-52): 177 us ggml_barrier(...): 360 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 387 us MUL_MAT_ID(ffn_moe_down-53): 386 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 437 us MUL_MAT_ID(ffn_moe_down-54): 662 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 419 us MUL_MAT_ID(ffn_moe_down-55): 251 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 374 us MUL_MAT_ID(ffn_moe_down-56): 379 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 405 us MUL_MAT_ID(ffn_moe_down-57): 176 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 379 us MUL_MAT_ID(ffn_moe_down-58): 202 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 363 us MUL_MAT_ID(ffn_moe_down-59): 207 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 391 us MUL_MAT_ID(ffn_moe_down-60): 175 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 396 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 375 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 387 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 376 us MUL_MAT_ID(ffn_moe_down-64): 221 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 351 us MUL_MAT_ID(ffn_moe_down-65): 306 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 381 us MUL_MAT_ID(ffn_moe_down-66): 362 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 398 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 352 us MUL_MAT_ID(ffn_moe_down-68): 224 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 370 us MUL_MAT_ID(ffn_moe_down-69): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 369 us MUL_MAT_ID(ffn_moe_down-70): 486 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 548 us MUL_MAT_ID(ffn_moe_down-71): 221 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 372 us MUL_MAT_ID(ffn_moe_down-72): 205 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 381 us MUL_MAT_ID(ffn_moe_down-73): 211 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 395 us MUL_MAT_ID(ffn_moe_down-74): 212 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 367 us MUL_MAT_ID(ffn_moe_down-75): 186 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 370 us MUL_MAT_ID(ffn_moe_down-76): 393 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 395 us MUL_MAT_ID(ffn_moe_down-77): 189 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 368 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 394 us MUL_MAT_ID(ffn_moe_down-79): 207 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 376 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 364 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 394 us MUL_MAT_ID(ffn_moe_down-82): 177 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 379 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 356 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 396 us MUL_MAT_ID(ffn_moe_down-85): 237 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 371 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 374 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 411 us MUL_MAT_ID(ffn_moe_down-88): 197 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 360 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 404 us MUL_MAT_ID(ffn_moe_down-90): 244 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 391 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 35 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 364 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 396 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 355 us MUL_MAT_ID(ffn_moe_down-27): 181 us ggml_barrier(...): 336 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 420 us MUL_MAT_ID(ffn_moe_down-28): 217 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 369 us MUL_MAT_ID(ffn_moe_down-29): 205 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 384 us MUL_MAT_ID(ffn_moe_down-30): 286 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 402 us MUL_MAT_ID(ffn_moe_down-31): 209 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 360 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 393 us MUL_MAT_ID(ffn_moe_down-33): 191 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 396 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 366 us MUL_MAT_ID(ffn_moe_down-35): 209 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 422 us MUL_MAT_ID(ffn_moe_down-36): 197 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 350 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 346 us MUL_MAT_ID(ffn_moe_down-38): 206 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 395 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 386 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 404 us MUL_MAT_ID(ffn_moe_down-41): 417 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 362 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 393 us MUL_MAT_ID(ffn_moe_down-43): 202 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 420 us MUL_MAT_ID(ffn_moe_down-44): 269 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 376 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 391 us MUL_MAT_ID(ffn_moe_down-46): 225 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 401 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 425 us MUL_MAT_ID(ffn_moe_down-48): 219 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 425 us MUL_MAT_ID(ffn_moe_down-49): 293 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 373 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 499 us MUL_MAT_ID(ffn_moe_down-51): 201 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 496 us MUL_MAT_ID(ffn_moe_down-52): 271 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 463 us MUL_MAT_ID(ffn_moe_down-53): 208 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 411 us MUL_MAT_ID(ffn_moe_down-54): 241 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 452 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 526 us MUL_MAT_ID(ffn_moe_down-56): 342 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 367 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 384 us MUL_MAT_ID(ffn_moe_down-58): 220 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 399 us MUL_MAT_ID(ffn_moe_down-59): 295 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 374 us MUL_MAT_ID(ffn_moe_down-60): 167 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 384 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 401 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 371 us MUL_MAT_ID(ffn_moe_down-63): 187 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 402 us MUL_MAT_ID(ffn_moe_down-64): 182 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 361 us MUL_MAT_ID(ffn_moe_down-65): 198 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 374 us MUL_MAT_ID(ffn_moe_down-66): 209 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 497 us MUL_MAT_ID(ffn_moe_down-67): 230 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 366 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 359 us MUL_MAT_ID(ffn_moe_down-69): 257 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 356 us MUL_MAT_ID(ffn_moe_down-70): 468 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 378 us MUL_MAT_ID(ffn_moe_down-71): 202 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 394 us MUL_MAT_ID(ffn_moe_down-72): 189 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 393 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 384 us MUL_MAT_ID(ffn_moe_down-74): 277 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 359 us MUL_MAT_ID(ffn_moe_down-75): 188 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 426 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 381 us MUL_MAT_ID(ffn_moe_down-77): 266 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 363 us MUL_MAT_ID(ffn_moe_down-78): 198 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 403 us MUL_MAT_ID(ffn_moe_down-79): 465 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 443 us MUL_MAT_ID(ffn_moe_down-80): 176 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 424 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 450 us MUL_MAT_ID(ffn_moe_down-82): 187 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 432 us MUL_MAT_ID(ffn_moe_down-83): 254 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 387 us MUL_MAT_ID(ffn_moe_down-84): 245 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 380 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 405 us MUL_MAT_ID(ffn_moe_down-86): 196 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 407 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 362 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 397 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 427 us MUL_MAT_ID(ffn_moe_down-90): 245 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 389 us MUL_MAT_ID(ffn_moe_down-91): 264 us ggml_barrier(...): 142 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 394 us MUL_MAT_ID(ffn_moe_down-25): 418 us ggml_barrier(...): 471 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 349 us MUL_MAT_ID(ffn_moe_down-26): 222 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 365 us MUL_MAT_ID(ffn_moe_down-27): 210 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 402 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 435 us MUL_MAT_ID(ffn_moe_down-29): 257 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 409 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 391 us MUL_MAT_ID(ffn_moe_down-31): 186 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 512 us MUL_MAT_ID(ffn_moe_down-32): 222 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 404 us MUL_MAT_ID(ffn_moe_down-33): 211 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 506 us MUL_MAT_ID(ffn_moe_down-34): 180 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 412 us MUL_MAT_ID(ffn_moe_down-35): 241 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 397 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 427 us MUL_MAT_ID(ffn_moe_down-37): 259 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 515 us MUL_MAT_ID(ffn_moe_down-38): 296 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 403 us MUL_MAT_ID(ffn_moe_down-39): 254 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 343 us MUL_MAT_ID(ffn_moe_down-40): 276 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 420 us MUL_MAT_ID(ffn_moe_down-41): 206 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 397 us MUL_MAT_ID(ffn_moe_down-42): 204 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 393 us MUL_MAT_ID(ffn_moe_down-43): 202 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 359 us MUL_MAT_ID(ffn_moe_down-44): 259 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 401 us MUL_MAT_ID(ffn_moe_down-45): 362 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 360 us MUL_MAT_ID(ffn_moe_down-46): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 482 us MUL_MAT_ID(ffn_moe_down-47): 201 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 385 us MUL_MAT_ID(ffn_moe_down-48): 230 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 420 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 357 us MUL_MAT_ID(ffn_moe_down-50): 166 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 396 us MUL_MAT_ID(ffn_moe_down-51): 199 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 489 us MUL_MAT_ID(ffn_moe_down-52): 270 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 396 us MUL_MAT_ID(ffn_moe_down-53): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 403 us MUL_MAT_ID(ffn_moe_down-54): 187 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 376 us MUL_MAT_ID(ffn_moe_down-55): 202 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 355 us MUL_MAT_ID(ffn_moe_down-56): 219 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 368 us MUL_MAT_ID(ffn_moe_down-57): 526 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 392 us MUL_MAT_ID(ffn_moe_down-58): 267 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 378 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 360 us MUL_MAT_ID(ffn_moe_down-60): 207 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 524 us MUL_MAT_ID(ffn_moe_down-61): 231 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 438 us MUL_MAT_ID(ffn_moe_down-62): 356 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 374 us MUL_MAT_ID(ffn_moe_down-63): 202 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 387 us MUL_MAT_ID(ffn_moe_down-64): 206 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 359 us MUL_MAT_ID(ffn_moe_down-65): 202 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 394 us MUL_MAT_ID(ffn_moe_down-66): 217 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 386 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 494 us MUL_MAT_ID(ffn_moe_down-68): 205 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 375 us MUL_MAT_ID(ffn_moe_down-69): 553 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 404 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 397 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 428 us MUL_MAT_ID(ffn_moe_down-72): 368 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 387 us MUL_MAT_ID(ffn_moe_down-73): 185 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 410 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 412 us MUL_MAT_ID(ffn_moe_down-75): 336 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 406 us MUL_MAT_ID(ffn_moe_down-76): 243 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 387 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 414 us MUL_MAT_ID(ffn_moe_down-78): 391 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 411 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 365 us MUL_MAT_ID(ffn_moe_down-80): 193 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 390 us MUL_MAT_ID(ffn_moe_down-81): 216 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 392 us MUL_MAT_ID(ffn_moe_down-82): 266 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 369 us MUL_MAT_ID(ffn_moe_down-83): 382 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 376 us MUL_MAT_ID(ffn_moe_down-84): 448 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 416 us MUL_MAT_ID(ffn_moe_down-85): 405 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 401 us MUL_MAT_ID(ffn_moe_down-86): 220 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 397 us MUL_MAT_ID(ffn_moe_down-87): 206 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 346 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 370 us MUL_MAT_ID(ffn_moe_down-89): 211 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 522 us MUL_MAT_ID(ffn_moe_down-90): 614 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 397 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 43 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 415 us MUL_MAT_ID(ffn_moe_down-25): 202 us ggml_barrier(...): 443 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 371 us MUL_MAT_ID(ffn_moe_down-26): 185 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 407 us MUL_MAT_ID(ffn_moe_down-27): 218 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 453 us MUL_MAT_ID(ffn_moe_down-28): 526 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 412 us MUL_MAT_ID(ffn_moe_down-29): 192 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 395 us MUL_MAT_ID(ffn_moe_down-30): 385 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 481 us MUL_MAT_ID(ffn_moe_down-31): 378 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 400 us MUL_MAT_ID(ffn_moe_down-32): 183 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 403 us MUL_MAT_ID(ffn_moe_down-33): 411 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 402 us MUL_MAT_ID(ffn_moe_down-34): 388 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 354 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 401 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 427 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 405 us MUL_MAT_ID(ffn_moe_down-38): 250 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 373 us MUL_MAT_ID(ffn_moe_down-39): 242 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 390 us MUL_MAT_ID(ffn_moe_down-40): 403 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 347 us MUL_MAT_ID(ffn_moe_down-41): 204 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 399 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 362 us MUL_MAT_ID(ffn_moe_down-43): 346 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 372 us MUL_MAT_ID(ffn_moe_down-44): 198 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 384 us MUL_MAT_ID(ffn_moe_down-45): 198 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 377 us MUL_MAT_ID(ffn_moe_down-46): 211 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 379 us MUL_MAT_ID(ffn_moe_down-47): 221 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 360 us MUL_MAT_ID(ffn_moe_down-48): 705 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 377 us MUL_MAT_ID(ffn_moe_down-49): 192 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 366 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 382 us MUL_MAT_ID(ffn_moe_down-51): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 403 us MUL_MAT_ID(ffn_moe_down-52): 258 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 350 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 373 us MUL_MAT_ID(ffn_moe_down-54): 415 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 376 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 344 us MUL_MAT_ID(ffn_moe_down-57): 266 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 389 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 384 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 388 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 477 us MUL_MAT_ID(ffn_moe_down-61): 161 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 368 us MUL_MAT_ID(ffn_moe_down-62): 261 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 366 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 368 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 359 us MUL_MAT_ID(ffn_moe_down-65): 210 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 392 us MUL_MAT_ID(ffn_moe_down-66): 434 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 397 us MUL_MAT_ID(ffn_moe_down-67): 377 us ggml_barrier(...): 335 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 381 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 371 us MUL_MAT_ID(ffn_moe_down-69): 189 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 390 us MUL_MAT_ID(ffn_moe_down-70): 182 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 399 us MUL_MAT_ID(ffn_moe_down-71): 201 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 456 us MUL_MAT_ID(ffn_moe_down-72): 216 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 393 us MUL_MAT_ID(ffn_moe_down-73): 213 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 382 us MUL_MAT_ID(ffn_moe_down-74): 213 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 420 us MUL_MAT_ID(ffn_moe_down-75): 205 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 390 us MUL_MAT_ID(ffn_moe_down-76): 448 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 393 us MUL_MAT_ID(ffn_moe_down-77): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 375 us MUL_MAT_ID(ffn_moe_down-78): 188 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 377 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 354 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 425 us MUL_MAT_ID(ffn_moe_down-81): 248 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 369 us MUL_MAT_ID(ffn_moe_down-82): 192 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 397 us MUL_MAT_ID(ffn_moe_down-83): 381 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 369 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 396 us MUL_MAT_ID(ffn_moe_down-85): 256 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 383 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 508 us MUL_MAT_ID(ffn_moe_down-87): 613 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 389 us MUL_MAT_ID(ffn_moe_down-88): 391 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 368 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 396 us MUL_MAT_ID(ffn_moe_down-90): 249 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 377 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 31 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 561 us MUL_MAT_ID(ffn_moe_down-25): 180 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 386 us MUL_MAT_ID(ffn_moe_down-26): 202 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 360 us MUL_MAT_ID(ffn_moe_down-27): 186 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 366 us MUL_MAT_ID(ffn_moe_down-28): 190 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 384 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 403 us MUL_MAT_ID(ffn_moe_down-30): 179 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 377 us MUL_MAT_ID(ffn_moe_down-31): 186 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 396 us MUL_MAT_ID(ffn_moe_down-32): 238 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 368 us MUL_MAT_ID(ffn_moe_down-33): 353 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 416 us MUL_MAT_ID(ffn_moe_down-34): 184 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 387 us MUL_MAT_ID(ffn_moe_down-35): 200 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 518 us MUL_MAT_ID(ffn_moe_down-36): 207 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 390 us MUL_MAT_ID(ffn_moe_down-37): 242 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 352 us MUL_MAT_ID(ffn_moe_down-38): 196 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 356 us MUL_MAT_ID(ffn_moe_down-39): 229 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 375 us MUL_MAT_ID(ffn_moe_down-40): 227 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 375 us MUL_MAT_ID(ffn_moe_down-41): 184 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 402 us MUL_MAT_ID(ffn_moe_down-42): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 380 us MUL_MAT_ID(ffn_moe_down-43): 512 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 408 us MUL_MAT_ID(ffn_moe_down-44): 212 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 390 us MUL_MAT_ID(ffn_moe_down-45): 369 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 350 us MUL_MAT_ID(ffn_moe_down-46): 181 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 385 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 395 us MUL_MAT_ID(ffn_moe_down-48): 190 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 388 us MUL_MAT_ID(ffn_moe_down-49): 203 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 451 us MUL_MAT_ID(ffn_moe_down-50): 281 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 383 us MUL_MAT_ID(ffn_moe_down-51): 204 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 372 us MUL_MAT_ID(ffn_moe_down-52): 213 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 375 us MUL_MAT_ID(ffn_moe_down-53): 179 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 421 us MUL_MAT_ID(ffn_moe_down-54): 263 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 395 us MUL_MAT_ID(ffn_moe_down-55): 534 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 400 us MUL_MAT_ID(ffn_moe_down-56): 187 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 419 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 407 us MUL_MAT_ID(ffn_moe_down-58): 188 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 405 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 353 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 431 us MUL_MAT_ID(ffn_moe_down-61): 383 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 420 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 366 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 406 us MUL_MAT_ID(ffn_moe_down-64): 202 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 367 us MUL_MAT_ID(ffn_moe_down-65): 210 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 365 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 409 us MUL_MAT_ID(ffn_moe_down-67): 530 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 394 us MUL_MAT_ID(ffn_moe_down-68): 206 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 363 us MUL_MAT_ID(ffn_moe_down-69): 204 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 429 us MUL_MAT_ID(ffn_moe_down-70): 282 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 375 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 379 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 392 us MUL_MAT_ID(ffn_moe_down-73): 195 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 357 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 372 us MUL_MAT_ID(ffn_moe_down-75): 172 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 530 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 344 us MUL_MAT_ID(ffn_moe_down-77): 332 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 403 us MUL_MAT_ID(ffn_moe_down-78): 202 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 377 us MUL_MAT_ID(ffn_moe_down-79): 207 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 372 us MUL_MAT_ID(ffn_moe_down-80): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 364 us MUL_MAT_ID(ffn_moe_down-81): 400 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 386 us MUL_MAT_ID(ffn_moe_down-82): 402 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 435 us MUL_MAT_ID(ffn_moe_down-83): 237 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 365 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 385 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 377 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 378 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 401 us MUL_MAT_ID(ffn_moe_down-88): 398 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 373 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 532 us MUL_MAT_ID(ffn_moe_down-90): 404 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 400 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 19 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 425 us MUL_MAT_ID(ffn_moe_down-25): 349 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 391 us MUL_MAT_ID(ffn_moe_down-26): 214 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 389 us MUL_MAT_ID(ffn_moe_down-27): 206 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 369 us MUL_MAT_ID(ffn_moe_down-28): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 400 us MUL_MAT_ID(ffn_moe_down-29): 203 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 375 us MUL_MAT_ID(ffn_moe_down-30): 256 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 392 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 351 us MUL_MAT_ID(ffn_moe_down-32): 217 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 395 us MUL_MAT_ID(ffn_moe_down-33): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 404 us MUL_MAT_ID(ffn_moe_down-34): 287 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 357 us MUL_MAT_ID(ffn_moe_down-35): 184 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 374 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 394 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 473 us MUL_MAT_ID(ffn_moe_down-38): 211 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 456 us MUL_MAT_ID(ffn_moe_down-39): 231 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 384 us MUL_MAT_ID(ffn_moe_down-40): 226 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 374 us MUL_MAT_ID(ffn_moe_down-41): 248 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 421 us MUL_MAT_ID(ffn_moe_down-42): 385 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 407 us MUL_MAT_ID(ffn_moe_down-43): 349 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 477 us MUL_MAT_ID(ffn_moe_down-44): 231 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 368 us MUL_MAT_ID(ffn_moe_down-45): 170 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 420 us MUL_MAT_ID(ffn_moe_down-46): 182 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 506 us MUL_MAT_ID(ffn_moe_down-47): 203 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 396 us MUL_MAT_ID(ffn_moe_down-48): 319 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 407 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 373 us MUL_MAT_ID(ffn_moe_down-51): 186 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 413 us MUL_MAT_ID(ffn_moe_down-52): 197 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 431 us MUL_MAT_ID(ffn_moe_down-53): 182 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 455 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 470 us MUL_MAT_ID(ffn_moe_down-55): 208 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 374 us MUL_MAT_ID(ffn_moe_down-56): 180 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 464 us MUL_MAT_ID(ffn_moe_down-57): 315 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 388 us MUL_MAT_ID(ffn_moe_down-58): 187 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 389 us MUL_MAT_ID(ffn_moe_down-59): 184 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 460 us MUL_MAT_ID(ffn_moe_down-60): 404 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 380 us MUL_MAT_ID(ffn_moe_down-61): 178 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 357 us MUL_MAT_ID(ffn_moe_down-62): 198 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 360 us MUL_MAT_ID(ffn_moe_down-63): 214 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 419 us MUL_MAT_ID(ffn_moe_down-64): 531 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 362 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 401 us MUL_MAT_ID(ffn_moe_down-66): 254 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 421 us MUL_MAT_ID(ffn_moe_down-67): 206 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 368 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 369 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 380 us MUL_MAT_ID(ffn_moe_down-70): 194 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 468 us MUL_MAT_ID(ffn_moe_down-71): 246 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 188 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 237 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 388 us MUL_MAT_ID(ffn_moe_down-74): 263 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 364 us MUL_MAT_ID(ffn_moe_down-75): 167 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 372 us MUL_MAT_ID(ffn_moe_down-76): 391 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 457 us MUL_MAT_ID(ffn_moe_down-77): 180 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 400 us MUL_MAT_ID(ffn_moe_down-78): 374 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 391 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 383 us MUL_MAT_ID(ffn_moe_down-80): 191 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 389 us MUL_MAT_ID(ffn_moe_down-81): 209 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 390 us MUL_MAT_ID(ffn_moe_down-82): 173 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 400 us MUL_MAT_ID(ffn_moe_down-83): 186 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 348 us MUL_MAT_ID(ffn_moe_down-84): 250 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 389 us MUL_MAT_ID(ffn_moe_down-85): 570 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 493 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 416 us MUL_MAT_ID(ffn_moe_down-87): 186 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 389 us MUL_MAT_ID(ffn_moe_down-88): 205 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 408 us MUL_MAT_ID(ffn_moe_down-89): 399 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 386 us MUL_MAT_ID(ffn_moe_down-90): 241 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 382 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 25 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 430 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 400 us MUL_MAT_ID(ffn_moe_down-26): 211 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 385 us MUL_MAT_ID(ffn_moe_down-27): 535 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 403 us MUL_MAT_ID(ffn_moe_down-28): 222 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 219 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 376 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 412 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 374 us MUL_MAT_ID(ffn_moe_down-32): 358 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 485 us MUL_MAT_ID(ffn_moe_down-33): 182 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 401 us MUL_MAT_ID(ffn_moe_down-34): 376 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 488 us MUL_MAT_ID(ffn_moe_down-35): 397 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 475 us MUL_MAT_ID(ffn_moe_down-36): 192 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 396 us MUL_MAT_ID(ffn_moe_down-37): 250 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 350 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 397 us MUL_MAT_ID(ffn_moe_down-39): 322 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 494 us MUL_MAT_ID(ffn_moe_down-40): 193 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 382 us MUL_MAT_ID(ffn_moe_down-41): 224 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 372 us MUL_MAT_ID(ffn_moe_down-42): 385 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 384 us MUL_MAT_ID(ffn_moe_down-43): 269 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 373 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 397 us MUL_MAT_ID(ffn_moe_down-45): 433 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 361 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 352 us MUL_MAT_ID(ffn_moe_down-47): 208 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 378 us MUL_MAT_ID(ffn_moe_down-48): 531 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 422 us MUL_MAT_ID(ffn_moe_down-49): 222 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 393 us MUL_MAT_ID(ffn_moe_down-50): 189 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 367 us MUL_MAT_ID(ffn_moe_down-51): 189 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 351 us MUL_MAT_ID(ffn_moe_down-52): 227 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 386 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 373 us MUL_MAT_ID(ffn_moe_down-54): 461 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 369 us MUL_MAT_ID(ffn_moe_down-55): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 378 us MUL_MAT_ID(ffn_moe_down-56): 185 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 357 us MUL_MAT_ID(ffn_moe_down-57): 188 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 379 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 417 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 365 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 358 us MUL_MAT_ID(ffn_moe_down-61): 374 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 375 us MUL_MAT_ID(ffn_moe_down-62): 291 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 398 us MUL_MAT_ID(ffn_moe_down-63): 186 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 378 us MUL_MAT_ID(ffn_moe_down-64): 288 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 371 us MUL_MAT_ID(ffn_moe_down-65): 220 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 369 us MUL_MAT_ID(ffn_moe_down-66): 298 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 400 us MUL_MAT_ID(ffn_moe_down-67): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 378 us MUL_MAT_ID(ffn_moe_down-68): 191 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 357 us MUL_MAT_ID(ffn_moe_down-69): 209 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 448 us MUL_MAT_ID(ffn_moe_down-70): 174 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 378 us MUL_MAT_ID(ffn_moe_down-71): 331 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 404 us MUL_MAT_ID(ffn_moe_down-72): 186 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 373 us MUL_MAT_ID(ffn_moe_down-73): 222 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 381 us MUL_MAT_ID(ffn_moe_down-74): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 410 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 406 us MUL_MAT_ID(ffn_moe_down-76): 392 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 384 us MUL_MAT_ID(ffn_moe_down-77): 544 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 502 us MUL_MAT_ID(ffn_moe_down-78): 338 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 405 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 400 us MUL_MAT_ID(ffn_moe_down-80): 189 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 393 us MUL_MAT_ID(ffn_moe_down-81): 209 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 378 us MUL_MAT_ID(ffn_moe_down-82): 208 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 388 us MUL_MAT_ID(ffn_moe_down-83): 180 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 392 us MUL_MAT_ID(ffn_moe_down-84): 243 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 371 us MUL_MAT_ID(ffn_moe_down-85): 249 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 420 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 376 us MUL_MAT_ID(ffn_moe_down-87): 208 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 388 us MUL_MAT_ID(ffn_moe_down-88): 210 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 367 us MUL_MAT_ID(ffn_moe_down-89): 203 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 370 us MUL_MAT_ID(ffn_moe_down-90): 227 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 409 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 14 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 526 us MUL_MAT_ID(ffn_moe_down-25): 575 us ggml_barrier(...): 342 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 377 us MUL_MAT_ID(ffn_moe_down-26): 246 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 411 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 366 us MUL_MAT_ID(ffn_moe_down-28): 184 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 429 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 377 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 374 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 361 us MUL_MAT_ID(ffn_moe_down-32): 184 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 399 us MUL_MAT_ID(ffn_moe_down-33): 162 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 411 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 391 us MUL_MAT_ID(ffn_moe_down-35): 195 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 375 us MUL_MAT_ID(ffn_moe_down-36): 203 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 395 us MUL_MAT_ID(ffn_moe_down-37): 246 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 380 us MUL_MAT_ID(ffn_moe_down-38): 197 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 388 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 364 us MUL_MAT_ID(ffn_moe_down-40): 246 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 363 us MUL_MAT_ID(ffn_moe_down-41): 190 us ggml_barrier(...): 386 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 386 us MUL_MAT_ID(ffn_moe_down-42): 190 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 555 us MUL_MAT_ID(ffn_moe_down-43): 163 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 374 us MUL_MAT_ID(ffn_moe_down-44): 242 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 442 us MUL_MAT_ID(ffn_moe_down-45): 193 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 415 us MUL_MAT_ID(ffn_moe_down-46): 181 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 404 us MUL_MAT_ID(ffn_moe_down-47): 203 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 465 us MUL_MAT_ID(ffn_moe_down-48): 204 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 371 us MUL_MAT_ID(ffn_moe_down-49): 226 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 510 us MUL_MAT_ID(ffn_moe_down-50): 208 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 418 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 389 us MUL_MAT_ID(ffn_moe_down-52): 213 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 402 us MUL_MAT_ID(ffn_moe_down-53): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 415 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 351 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 405 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 387 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 411 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 377 us MUL_MAT_ID(ffn_moe_down-59): 184 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 350 us MUL_MAT_ID(ffn_moe_down-60): 226 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 397 us MUL_MAT_ID(ffn_moe_down-61): 210 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 365 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 386 us MUL_MAT_ID(ffn_moe_down-63): 206 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 376 us MUL_MAT_ID(ffn_moe_down-64): 520 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 368 us MUL_MAT_ID(ffn_moe_down-65): 212 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 464 us MUL_MAT_ID(ffn_moe_down-66): 185 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 350 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 406 us MUL_MAT_ID(ffn_moe_down-68): 237 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 393 us MUL_MAT_ID(ffn_moe_down-69): 551 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 401 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 389 us MUL_MAT_ID(ffn_moe_down-71): 179 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 382 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 386 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 401 us MUL_MAT_ID(ffn_moe_down-75): 325 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 398 us MUL_MAT_ID(ffn_moe_down-76): 214 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 405 us MUL_MAT_ID(ffn_moe_down-77): 212 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 401 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 401 us MUL_MAT_ID(ffn_moe_down-79): 230 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 397 us MUL_MAT_ID(ffn_moe_down-80): 228 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 391 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 390 us MUL_MAT_ID(ffn_moe_down-82): 197 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 393 us MUL_MAT_ID(ffn_moe_down-83): 202 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 404 us MUL_MAT_ID(ffn_moe_down-84): 246 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 381 us MUL_MAT_ID(ffn_moe_down-85): 274 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 371 us MUL_MAT_ID(ffn_moe_down-86): 204 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 401 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 411 us MUL_MAT_ID(ffn_moe_down-88): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 397 us MUL_MAT_ID(ffn_moe_down-89): 222 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 389 us MUL_MAT_ID(ffn_moe_down-90): 255 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 380 us MUL_MAT_ID(ffn_moe_down-91): 227 us ggml_barrier(...): 41 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 378 us MUL_MAT_ID(ffn_moe_down-25): 188 us ggml_barrier(...): 522 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 369 us MUL_MAT_ID(ffn_moe_down-26): 318 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 347 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 408 us MUL_MAT_ID(ffn_moe_down-28): 271 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 389 us MUL_MAT_ID(ffn_moe_down-29): 183 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 455 us MUL_MAT_ID(ffn_moe_down-30): 175 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 386 us MUL_MAT_ID(ffn_moe_down-31): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 440 us MUL_MAT_ID(ffn_moe_down-32): 218 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 410 us MUL_MAT_ID(ffn_moe_down-33): 516 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 491 us MUL_MAT_ID(ffn_moe_down-34): 210 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 391 us MUL_MAT_ID(ffn_moe_down-35): 293 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 401 us MUL_MAT_ID(ffn_moe_down-36): 412 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 432 us MUL_MAT_ID(ffn_moe_down-37): 223 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 374 us MUL_MAT_ID(ffn_moe_down-38): 203 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 373 us MUL_MAT_ID(ffn_moe_down-39): 519 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 465 us MUL_MAT_ID(ffn_moe_down-40): 240 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 386 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 483 us MUL_MAT_ID(ffn_moe_down-42): 531 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 417 us MUL_MAT_ID(ffn_moe_down-43): 399 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 402 us MUL_MAT_ID(ffn_moe_down-44): 402 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 371 us MUL_MAT_ID(ffn_moe_down-45): 199 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 430 us MUL_MAT_ID(ffn_moe_down-46): 226 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 353 us MUL_MAT_ID(ffn_moe_down-47): 361 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 459 us MUL_MAT_ID(ffn_moe_down-48): 300 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 364 us MUL_MAT_ID(ffn_moe_down-49): 362 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 381 us MUL_MAT_ID(ffn_moe_down-50): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 373 us MUL_MAT_ID(ffn_moe_down-51): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 367 us MUL_MAT_ID(ffn_moe_down-52): 195 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 365 us MUL_MAT_ID(ffn_moe_down-53): 300 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 377 us MUL_MAT_ID(ffn_moe_down-54): 297 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 468 us MUL_MAT_ID(ffn_moe_down-55): 164 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 411 us MUL_MAT_ID(ffn_moe_down-56): 408 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 336 us MUL_MAT_ID(ffn_moe_down-57): 252 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 435 us MUL_MAT_ID(ffn_moe_down-58): 181 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 381 us MUL_MAT_ID(ffn_moe_down-59): 281 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 398 us MUL_MAT_ID(ffn_moe_down-60): 385 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 403 us MUL_MAT_ID(ffn_moe_down-61): 460 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 374 us MUL_MAT_ID(ffn_moe_down-62): 201 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 377 us MUL_MAT_ID(ffn_moe_down-63): 210 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 394 us MUL_MAT_ID(ffn_moe_down-64): 185 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 409 us MUL_MAT_ID(ffn_moe_down-65): 188 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 423 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 391 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 399 us MUL_MAT_ID(ffn_moe_down-68): 257 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 344 us MUL_MAT_ID(ffn_moe_down-69): 190 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 383 us MUL_MAT_ID(ffn_moe_down-70): 233 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 382 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 347 us MUL_MAT_ID(ffn_moe_down-72): 241 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 497 us MUL_MAT_ID(ffn_moe_down-73): 333 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 360 us MUL_MAT_ID(ffn_moe_down-74): 387 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 358 us MUL_MAT_ID(ffn_moe_down-75): 545 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 378 us MUL_MAT_ID(ffn_moe_down-76): 196 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 416 us MUL_MAT_ID(ffn_moe_down-77): 248 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 393 us MUL_MAT_ID(ffn_moe_down-78): 266 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 387 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 364 us MUL_MAT_ID(ffn_moe_down-80): 270 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 355 us MUL_MAT_ID(ffn_moe_down-81): 184 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 351 us MUL_MAT_ID(ffn_moe_down-82): 191 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 352 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 443 us MUL_MAT_ID(ffn_moe_down-84): 194 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 487 us MUL_MAT_ID(ffn_moe_down-85): 212 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 374 us MUL_MAT_ID(ffn_moe_down-86): 202 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 351 us MUL_MAT_ID(ffn_moe_down-87): 294 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 349 us MUL_MAT_ID(ffn_moe_down-88): 200 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 388 us MUL_MAT_ID(ffn_moe_down-89): 219 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 371 us MUL_MAT_ID(ffn_moe_down-90): 226 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 373 us MUL_MAT_ID(ffn_moe_down-91): 302 us ggml_barrier(...): 147 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 198 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 416 us MUL_MAT_ID(ffn_moe_down-26): 502 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 495 us MUL_MAT_ID(ffn_moe_down-27): 215 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 345 us MUL_MAT_ID(ffn_moe_down-28): 161 us ggml_barrier(...): 365 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 490 us MUL_MAT_ID(ffn_moe_down-29): 211 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 355 us MUL_MAT_ID(ffn_moe_down-30): 189 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 390 us MUL_MAT_ID(ffn_moe_down-31): 208 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 368 us MUL_MAT_ID(ffn_moe_down-32): 291 us ggml_barrier(...): 337 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 386 us MUL_MAT_ID(ffn_moe_down-33): 205 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 417 us MUL_MAT_ID(ffn_moe_down-34): 180 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 372 us MUL_MAT_ID(ffn_moe_down-35): 183 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 404 us MUL_MAT_ID(ffn_moe_down-36): 176 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 409 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 480 us MUL_MAT_ID(ffn_moe_down-38): 169 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 388 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 500 us MUL_MAT_ID(ffn_moe_down-40): 247 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 374 us MUL_MAT_ID(ffn_moe_down-41): 345 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 381 us MUL_MAT_ID(ffn_moe_down-42): 347 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 376 us MUL_MAT_ID(ffn_moe_down-43): 209 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 384 us MUL_MAT_ID(ffn_moe_down-44): 217 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 410 us MUL_MAT_ID(ffn_moe_down-45): 220 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 363 us MUL_MAT_ID(ffn_moe_down-46): 211 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 513 us MUL_MAT_ID(ffn_moe_down-47): 209 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 386 us MUL_MAT_ID(ffn_moe_down-48): 361 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 383 us MUL_MAT_ID(ffn_moe_down-49): 197 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 355 us MUL_MAT_ID(ffn_moe_down-50): 208 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 423 us MUL_MAT_ID(ffn_moe_down-51): 201 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 364 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 380 us MUL_MAT_ID(ffn_moe_down-53): 204 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 386 us MUL_MAT_ID(ffn_moe_down-54): 208 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 373 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 373 us MUL_MAT_ID(ffn_moe_down-56): 206 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 371 us MUL_MAT_ID(ffn_moe_down-57): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 397 us MUL_MAT_ID(ffn_moe_down-58): 342 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 383 us MUL_MAT_ID(ffn_moe_down-59): 187 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 365 us MUL_MAT_ID(ffn_moe_down-60): 189 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 404 us MUL_MAT_ID(ffn_moe_down-61): 179 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 366 us MUL_MAT_ID(ffn_moe_down-62): 179 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 470 us MUL_MAT_ID(ffn_moe_down-63): 395 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 399 us MUL_MAT_ID(ffn_moe_down-64): 402 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 357 us MUL_MAT_ID(ffn_moe_down-65): 187 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 396 us MUL_MAT_ID(ffn_moe_down-66): 310 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 478 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 371 us MUL_MAT_ID(ffn_moe_down-68): 380 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 393 us MUL_MAT_ID(ffn_moe_down-69): 186 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 391 us MUL_MAT_ID(ffn_moe_down-70): 206 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 420 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 189 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 381 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 395 us MUL_MAT_ID(ffn_moe_down-74): 182 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 376 us MUL_MAT_ID(ffn_moe_down-75): 553 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 371 us MUL_MAT_ID(ffn_moe_down-76): 312 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 470 us MUL_MAT_ID(ffn_moe_down-77): 206 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 398 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 363 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 393 us MUL_MAT_ID(ffn_moe_down-80): 214 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 371 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 488 us MUL_MAT_ID(ffn_moe_down-82): 274 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 364 us MUL_MAT_ID(ffn_moe_down-83): 188 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 389 us MUL_MAT_ID(ffn_moe_down-84): 303 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 364 us MUL_MAT_ID(ffn_moe_down-85): 358 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 380 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 385 us MUL_MAT_ID(ffn_moe_down-87): 533 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 380 us MUL_MAT_ID(ffn_moe_down-88): 202 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 376 us MUL_MAT_ID(ffn_moe_down-89): 281 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 392 us MUL_MAT_ID(ffn_moe_down-90): 571 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 395 us MUL_MAT_ID(ffn_moe_down-91): 246 us ggml_barrier(...): 129 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 395 us MUL_MAT_ID(ffn_moe_down-25): 198 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 391 us MUL_MAT_ID(ffn_moe_down-26): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 371 us MUL_MAT_ID(ffn_moe_down-27): 182 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 442 us MUL_MAT_ID(ffn_moe_down-28): 213 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 454 us MUL_MAT_ID(ffn_moe_down-29): 164 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 395 us MUL_MAT_ID(ffn_moe_down-30): 240 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 392 us MUL_MAT_ID(ffn_moe_down-31): 397 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 380 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 390 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 392 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 398 us MUL_MAT_ID(ffn_moe_down-35): 412 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 362 us MUL_MAT_ID(ffn_moe_down-36): 195 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 380 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 373 us MUL_MAT_ID(ffn_moe_down-38): 221 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 365 us MUL_MAT_ID(ffn_moe_down-39): 231 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 373 us MUL_MAT_ID(ffn_moe_down-40): 235 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 503 us MUL_MAT_ID(ffn_moe_down-41): 541 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 381 us MUL_MAT_ID(ffn_moe_down-42): 210 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 386 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 385 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 377 us MUL_MAT_ID(ffn_moe_down-45): 234 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 394 us MUL_MAT_ID(ffn_moe_down-46): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 365 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 362 us MUL_MAT_ID(ffn_moe_down-48): 208 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 385 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 395 us MUL_MAT_ID(ffn_moe_down-50): 199 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 370 us MUL_MAT_ID(ffn_moe_down-51): 186 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 409 us MUL_MAT_ID(ffn_moe_down-52): 218 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 367 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 386 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 423 us MUL_MAT_ID(ffn_moe_down-55): 226 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 375 us MUL_MAT_ID(ffn_moe_down-56): 310 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 385 us MUL_MAT_ID(ffn_moe_down-57): 292 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 419 us MUL_MAT_ID(ffn_moe_down-58): 163 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 387 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 359 us MUL_MAT_ID(ffn_moe_down-60): 158 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 394 us MUL_MAT_ID(ffn_moe_down-61): 370 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 356 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 522 us MUL_MAT_ID(ffn_moe_down-63): 209 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 372 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 390 us MUL_MAT_ID(ffn_moe_down-65): 238 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 368 us MUL_MAT_ID(ffn_moe_down-66): 294 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 384 us MUL_MAT_ID(ffn_moe_down-67): 217 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 375 us MUL_MAT_ID(ffn_moe_down-68): 192 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 377 us MUL_MAT_ID(ffn_moe_down-69): 327 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 399 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 383 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 322 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 517 us MUL_MAT_ID(ffn_moe_down-72): 216 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 376 us MUL_MAT_ID(ffn_moe_down-73): 258 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 367 us MUL_MAT_ID(ffn_moe_down-74): 223 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 460 us MUL_MAT_ID(ffn_moe_down-75): 200 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 398 us MUL_MAT_ID(ffn_moe_down-76): 208 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 368 us MUL_MAT_ID(ffn_moe_down-77): 241 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 371 us MUL_MAT_ID(ffn_moe_down-78): 221 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 368 us MUL_MAT_ID(ffn_moe_down-79): 371 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 398 us MUL_MAT_ID(ffn_moe_down-80): 210 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 372 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 366 us MUL_MAT_ID(ffn_moe_down-82): 190 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 391 us MUL_MAT_ID(ffn_moe_down-83): 298 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 426 us MUL_MAT_ID(ffn_moe_down-84): 271 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 412 us MUL_MAT_ID(ffn_moe_down-85): 261 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 351 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 391 us MUL_MAT_ID(ffn_moe_down-87): 197 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 408 us MUL_MAT_ID(ffn_moe_down-88): 236 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 357 us MUL_MAT_ID(ffn_moe_down-89): 195 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 351 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 371 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 67 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 595 us MUL_MAT_ID(ffn_moe_down-25): 209 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 408 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 376 us MUL_MAT_ID(ffn_moe_down-27): 232 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 405 us MUL_MAT_ID(ffn_moe_down-28): 269 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 375 us MUL_MAT_ID(ffn_moe_down-29): 223 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 367 us MUL_MAT_ID(ffn_moe_down-30): 168 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 371 us MUL_MAT_ID(ffn_moe_down-31): 391 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 403 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 344 us MUL_MAT_ID(ffn_moe_down-33): 451 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 366 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 391 us MUL_MAT_ID(ffn_moe_down-35): 209 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 363 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 460 us MUL_MAT_ID(ffn_moe_down-37): 225 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 367 us MUL_MAT_ID(ffn_moe_down-38): 249 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 407 us MUL_MAT_ID(ffn_moe_down-39): 244 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 496 us MUL_MAT_ID(ffn_moe_down-40): 299 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 364 us MUL_MAT_ID(ffn_moe_down-41): 367 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 362 us MUL_MAT_ID(ffn_moe_down-42): 185 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 391 us MUL_MAT_ID(ffn_moe_down-43): 197 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 359 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 378 us MUL_MAT_ID(ffn_moe_down-45): 200 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 383 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 353 us MUL_MAT_ID(ffn_moe_down-47): 236 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 383 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 405 us MUL_MAT_ID(ffn_moe_down-49): 196 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 418 us MUL_MAT_ID(ffn_moe_down-50): 527 us ggml_barrier(...): 382 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 400 us MUL_MAT_ID(ffn_moe_down-51): 211 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 461 us MUL_MAT_ID(ffn_moe_down-52): 206 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 389 us MUL_MAT_ID(ffn_moe_down-53): 212 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 374 us MUL_MAT_ID(ffn_moe_down-54): 197 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 359 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 406 us MUL_MAT_ID(ffn_moe_down-56): 336 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 421 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 400 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 377 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 379 us MUL_MAT_ID(ffn_moe_down-60): 406 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 391 us MUL_MAT_ID(ffn_moe_down-61): 203 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 357 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 357 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 405 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 382 us MUL_MAT_ID(ffn_moe_down-65): 190 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 365 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 395 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 346 us MUL_MAT_ID(ffn_moe_down-68): 298 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 380 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 393 us MUL_MAT_ID(ffn_moe_down-70): 233 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 346 us MUL_MAT_ID(ffn_moe_down-71): 397 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 395 us MUL_MAT_ID(ffn_moe_down-72): 209 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 393 us MUL_MAT_ID(ffn_moe_down-74): 299 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 368 us MUL_MAT_ID(ffn_moe_down-75): 206 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 391 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 387 us MUL_MAT_ID(ffn_moe_down-77): 216 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 398 us MUL_MAT_ID(ffn_moe_down-78): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 384 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 386 us MUL_MAT_ID(ffn_moe_down-80): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 389 us MUL_MAT_ID(ffn_moe_down-81): 211 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 404 us MUL_MAT_ID(ffn_moe_down-82): 362 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 379 us MUL_MAT_ID(ffn_moe_down-83): 212 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 379 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 362 us MUL_MAT_ID(ffn_moe_down-85): 240 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 351 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 366 us MUL_MAT_ID(ffn_moe_down-87): 379 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 351 us MUL_MAT_ID(ffn_moe_down-88): 202 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 338 us MUL_MAT_ID(ffn_moe_down-89): 185 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 348 us MUL_MAT_ID(ffn_moe_down-90): 223 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 372 us MUL_MAT_ID(ffn_moe_down-91): 209 us ggml_barrier(...): 272 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 197 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 419 us MUL_MAT_ID(ffn_moe_down-26): 252 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 362 us MUL_MAT_ID(ffn_moe_down-27): 262 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 408 us MUL_MAT_ID(ffn_moe_down-28): 214 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 410 us MUL_MAT_ID(ffn_moe_down-29): 250 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 405 us MUL_MAT_ID(ffn_moe_down-30): 267 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 352 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 372 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 377 us MUL_MAT_ID(ffn_moe_down-33): 205 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 443 us MUL_MAT_ID(ffn_moe_down-34): 215 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 380 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 414 us MUL_MAT_ID(ffn_moe_down-36): 381 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 389 us MUL_MAT_ID(ffn_moe_down-37): 224 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 447 us MUL_MAT_ID(ffn_moe_down-38): 226 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 483 us MUL_MAT_ID(ffn_moe_down-39): 253 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 491 us MUL_MAT_ID(ffn_moe_down-40): 709 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 408 us MUL_MAT_ID(ffn_moe_down-41): 188 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 401 us MUL_MAT_ID(ffn_moe_down-42): 483 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 349 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 453 us MUL_MAT_ID(ffn_moe_down-44): 273 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 380 us MUL_MAT_ID(ffn_moe_down-45): 216 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 409 us MUL_MAT_ID(ffn_moe_down-46): 219 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 393 us MUL_MAT_ID(ffn_moe_down-47): 180 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 335 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 364 us MUL_MAT_ID(ffn_moe_down-49): 209 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 366 us MUL_MAT_ID(ffn_moe_down-50): 396 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 376 us MUL_MAT_ID(ffn_moe_down-51): 235 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 372 us MUL_MAT_ID(ffn_moe_down-52): 210 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 361 us MUL_MAT_ID(ffn_moe_down-53): 206 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 391 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 218 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 379 us MUL_MAT_ID(ffn_moe_down-56): 220 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 402 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 402 us MUL_MAT_ID(ffn_moe_down-58): 198 us ggml_barrier(...): 448 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 376 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 418 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 348 us MUL_MAT_ID(ffn_moe_down-61): 268 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 379 us MUL_MAT_ID(ffn_moe_down-62): 414 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 399 us MUL_MAT_ID(ffn_moe_down-64): 189 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 393 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 393 us MUL_MAT_ID(ffn_moe_down-66): 372 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 345 us MUL_MAT_ID(ffn_moe_down-67): 208 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 383 us MUL_MAT_ID(ffn_moe_down-68): 288 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 373 us MUL_MAT_ID(ffn_moe_down-69): 164 us ggml_barrier(...): 505 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 406 us MUL_MAT_ID(ffn_moe_down-70): 221 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 405 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 368 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 350 us MUL_MAT_ID(ffn_moe_down-73): 218 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 365 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 373 us MUL_MAT_ID(ffn_moe_down-75): 190 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 361 us MUL_MAT_ID(ffn_moe_down-76): 178 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 367 us MUL_MAT_ID(ffn_moe_down-77): 229 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 387 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 347 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 404 us MUL_MAT_ID(ffn_moe_down-80): 218 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 404 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 363 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 368 us MUL_MAT_ID(ffn_moe_down-83): 348 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 405 us MUL_MAT_ID(ffn_moe_down-84): 240 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 410 us MUL_MAT_ID(ffn_moe_down-85): 437 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 387 us MUL_MAT_ID(ffn_moe_down-86): 545 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 402 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 375 us MUL_MAT_ID(ffn_moe_down-88): 367 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 395 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 352 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 407 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 19 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 407 us MUL_MAT_ID(ffn_moe_down-25): 185 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 412 us MUL_MAT_ID(ffn_moe_down-26): 222 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 372 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 381 us MUL_MAT_ID(ffn_moe_down-28): 232 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 395 us MUL_MAT_ID(ffn_moe_down-29): 207 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 403 us MUL_MAT_ID(ffn_moe_down-30): 182 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 395 us MUL_MAT_ID(ffn_moe_down-31): 222 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 412 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 397 us MUL_MAT_ID(ffn_moe_down-33): 349 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 373 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 393 us MUL_MAT_ID(ffn_moe_down-35): 349 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 396 us MUL_MAT_ID(ffn_moe_down-36): 394 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 381 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 462 us MUL_MAT_ID(ffn_moe_down-38): 373 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 410 us MUL_MAT_ID(ffn_moe_down-39): 457 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 414 us MUL_MAT_ID(ffn_moe_down-40): 240 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 370 us MUL_MAT_ID(ffn_moe_down-41): 186 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 357 us MUL_MAT_ID(ffn_moe_down-42): 204 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 389 us MUL_MAT_ID(ffn_moe_down-43): 220 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 397 us MUL_MAT_ID(ffn_moe_down-44): 418 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 417 us MUL_MAT_ID(ffn_moe_down-45): 181 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 386 us MUL_MAT_ID(ffn_moe_down-46): 227 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 446 us MUL_MAT_ID(ffn_moe_down-47): 179 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 344 us MUL_MAT_ID(ffn_moe_down-48): 199 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 403 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 383 us MUL_MAT_ID(ffn_moe_down-50): 214 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 513 us MUL_MAT_ID(ffn_moe_down-51): 186 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 386 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 414 us MUL_MAT_ID(ffn_moe_down-53): 216 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 382 us MUL_MAT_ID(ffn_moe_down-54): 615 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 385 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 464 us MUL_MAT_ID(ffn_moe_down-56): 282 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 399 us MUL_MAT_ID(ffn_moe_down-57): 219 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 375 us MUL_MAT_ID(ffn_moe_down-58): 176 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 395 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 402 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 435 us MUL_MAT_ID(ffn_moe_down-62): 263 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 408 us MUL_MAT_ID(ffn_moe_down-63): 209 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 380 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 390 us MUL_MAT_ID(ffn_moe_down-65): 187 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 363 us MUL_MAT_ID(ffn_moe_down-66): 411 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 358 us MUL_MAT_ID(ffn_moe_down-67): 345 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 393 us MUL_MAT_ID(ffn_moe_down-68): 211 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 354 us MUL_MAT_ID(ffn_moe_down-69): 478 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 372 us MUL_MAT_ID(ffn_moe_down-70): 397 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 382 us MUL_MAT_ID(ffn_moe_down-71): 223 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 359 us MUL_MAT_ID(ffn_moe_down-72): 327 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 413 us MUL_MAT_ID(ffn_moe_down-73): 180 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 360 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 388 us MUL_MAT_ID(ffn_moe_down-75): 255 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 390 us MUL_MAT_ID(ffn_moe_down-76): 200 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 351 us MUL_MAT_ID(ffn_moe_down-77): 214 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 357 us MUL_MAT_ID(ffn_moe_down-78): 482 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 398 us MUL_MAT_ID(ffn_moe_down-79): 251 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 496 us MUL_MAT_ID(ffn_moe_down-80): 224 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 385 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 489 us MUL_MAT_ID(ffn_moe_down-82): 202 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 409 us MUL_MAT_ID(ffn_moe_down-83): 224 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 405 us MUL_MAT_ID(ffn_moe_down-84): 249 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 401 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 374 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 378 us MUL_MAT_ID(ffn_moe_down-87): 201 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 419 us MUL_MAT_ID(ffn_moe_down-88): 529 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 383 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 355 us MUL_MAT_ID(ffn_moe_down-90): 627 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 389 us MUL_MAT_ID(ffn_moe_down-91): 604 us ggml_barrier(...): 18 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 285 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 394 us MUL_MAT_ID(ffn_moe_down-26): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 395 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 369 us MUL_MAT_ID(ffn_moe_down-28): 377 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 358 us MUL_MAT_ID(ffn_moe_down-29): 207 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 400 us MUL_MAT_ID(ffn_moe_down-30): 203 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 368 us MUL_MAT_ID(ffn_moe_down-31): 218 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 408 us MUL_MAT_ID(ffn_moe_down-32): 412 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 426 us MUL_MAT_ID(ffn_moe_down-33): 204 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 383 us MUL_MAT_ID(ffn_moe_down-34): 186 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 368 us MUL_MAT_ID(ffn_moe_down-35): 395 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 465 us MUL_MAT_ID(ffn_moe_down-36): 297 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 462 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 351 us MUL_MAT_ID(ffn_moe_down-38): 161 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 399 us MUL_MAT_ID(ffn_moe_down-39): 643 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 360 us MUL_MAT_ID(ffn_moe_down-40): 248 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 520 us MUL_MAT_ID(ffn_moe_down-41): 238 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 389 us MUL_MAT_ID(ffn_moe_down-42): 184 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 359 us MUL_MAT_ID(ffn_moe_down-43): 307 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 495 us MUL_MAT_ID(ffn_moe_down-44): 555 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 408 us MUL_MAT_ID(ffn_moe_down-45): 285 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 414 us MUL_MAT_ID(ffn_moe_down-46): 185 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 376 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 418 us MUL_MAT_ID(ffn_moe_down-48): 217 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 400 us MUL_MAT_ID(ffn_moe_down-49): 230 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 390 us MUL_MAT_ID(ffn_moe_down-50): 405 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 370 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 403 us MUL_MAT_ID(ffn_moe_down-52): 212 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 441 us MUL_MAT_ID(ffn_moe_down-53): 380 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 365 us MUL_MAT_ID(ffn_moe_down-54): 169 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 399 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 427 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 395 us MUL_MAT_ID(ffn_moe_down-57): 218 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 385 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 380 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 402 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 397 us MUL_MAT_ID(ffn_moe_down-61): 393 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 395 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 470 us MUL_MAT_ID(ffn_moe_down-63): 185 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 412 us MUL_MAT_ID(ffn_moe_down-64): 199 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 400 us MUL_MAT_ID(ffn_moe_down-65): 317 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 430 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 396 us MUL_MAT_ID(ffn_moe_down-67): 265 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 368 us MUL_MAT_ID(ffn_moe_down-68): 285 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 432 us MUL_MAT_ID(ffn_moe_down-69): 279 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 374 us MUL_MAT_ID(ffn_moe_down-70): 547 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 389 us MUL_MAT_ID(ffn_moe_down-71): 186 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 376 us MUL_MAT_ID(ffn_moe_down-72): 207 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 391 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 360 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 384 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 409 us MUL_MAT_ID(ffn_moe_down-76): 223 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 376 us MUL_MAT_ID(ffn_moe_down-77): 369 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 416 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 376 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 373 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 374 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 368 us MUL_MAT_ID(ffn_moe_down-82): 552 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 408 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 392 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 395 us MUL_MAT_ID(ffn_moe_down-85): 318 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 381 us MUL_MAT_ID(ffn_moe_down-86): 272 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 392 us MUL_MAT_ID(ffn_moe_down-87): 416 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 373 us MUL_MAT_ID(ffn_moe_down-88): 186 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 362 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 390 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 387 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 31 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 388 us MUL_MAT_ID(ffn_moe_down-25): 193 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 394 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 393 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 369 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 522 us MUL_MAT_ID(ffn_moe_down-29): 229 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 188 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 401 us MUL_MAT_ID(ffn_moe_down-31): 186 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 447 us MUL_MAT_ID(ffn_moe_down-32): 284 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 357 us MUL_MAT_ID(ffn_moe_down-33): 291 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 410 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 416 us MUL_MAT_ID(ffn_moe_down-35): 296 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 404 us MUL_MAT_ID(ffn_moe_down-36): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 388 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 387 us MUL_MAT_ID(ffn_moe_down-38): 241 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 365 us MUL_MAT_ID(ffn_moe_down-39): 243 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 484 us MUL_MAT_ID(ffn_moe_down-40): 368 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 355 us MUL_MAT_ID(ffn_moe_down-41): 207 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 369 us MUL_MAT_ID(ffn_moe_down-42): 284 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 359 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 361 us MUL_MAT_ID(ffn_moe_down-44): 274 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 383 us MUL_MAT_ID(ffn_moe_down-45): 326 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 365 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 353 us MUL_MAT_ID(ffn_moe_down-47): 172 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 359 us MUL_MAT_ID(ffn_moe_down-48): 194 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 354 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 362 us MUL_MAT_ID(ffn_moe_down-50): 210 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 396 us MUL_MAT_ID(ffn_moe_down-51): 236 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 342 us MUL_MAT_ID(ffn_moe_down-52): 233 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 401 us MUL_MAT_ID(ffn_moe_down-53): 332 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 424 us MUL_MAT_ID(ffn_moe_down-54): 257 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 376 us MUL_MAT_ID(ffn_moe_down-55): 245 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 384 us MUL_MAT_ID(ffn_moe_down-56): 189 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 417 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 397 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 362 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 398 us MUL_MAT_ID(ffn_moe_down-60): 187 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 402 us MUL_MAT_ID(ffn_moe_down-61): 188 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 380 us MUL_MAT_ID(ffn_moe_down-62): 201 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 425 us MUL_MAT_ID(ffn_moe_down-63): 260 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 385 us MUL_MAT_ID(ffn_moe_down-64): 204 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 395 us MUL_MAT_ID(ffn_moe_down-65): 204 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 387 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 403 us MUL_MAT_ID(ffn_moe_down-67): 314 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 383 us MUL_MAT_ID(ffn_moe_down-68): 189 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 368 us MUL_MAT_ID(ffn_moe_down-69): 333 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 386 us MUL_MAT_ID(ffn_moe_down-70): 264 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 399 us MUL_MAT_ID(ffn_moe_down-71): 536 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 368 us MUL_MAT_ID(ffn_moe_down-72): 222 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 354 us MUL_MAT_ID(ffn_moe_down-73): 302 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 341 us MUL_MAT_ID(ffn_moe_down-74): 380 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 389 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 490 us MUL_MAT_ID(ffn_moe_down-76): 201 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 361 us MUL_MAT_ID(ffn_moe_down-77): 220 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 406 us MUL_MAT_ID(ffn_moe_down-78): 204 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 426 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 380 us MUL_MAT_ID(ffn_moe_down-80): 211 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 381 us MUL_MAT_ID(ffn_moe_down-81): 206 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 380 us MUL_MAT_ID(ffn_moe_down-82): 230 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 387 us MUL_MAT_ID(ffn_moe_down-83): 190 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 395 us MUL_MAT_ID(ffn_moe_down-84): 247 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 377 us MUL_MAT_ID(ffn_moe_down-85): 595 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 393 us MUL_MAT_ID(ffn_moe_down-86): 360 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 374 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 411 us MUL_MAT_ID(ffn_moe_down-88): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 380 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 381 us MUL_MAT_ID(ffn_moe_down-90): 249 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 397 us MUL_MAT_ID(ffn_moe_down-91): 251 us ggml_barrier(...): 28 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 378 us MUL_MAT_ID(ffn_moe_down-25): 196 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 389 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 387 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 350 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 365 us MUL_MAT_ID(ffn_moe_down-29): 289 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 391 us MUL_MAT_ID(ffn_moe_down-30): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 432 us MUL_MAT_ID(ffn_moe_down-31): 207 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 406 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 385 us MUL_MAT_ID(ffn_moe_down-33): 218 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 348 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 381 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 364 us MUL_MAT_ID(ffn_moe_down-36): 209 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 387 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 350 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 379 us MUL_MAT_ID(ffn_moe_down-39): 241 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 356 us MUL_MAT_ID(ffn_moe_down-40): 245 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 406 us MUL_MAT_ID(ffn_moe_down-41): 378 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 420 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 423 us MUL_MAT_ID(ffn_moe_down-43): 188 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 369 us MUL_MAT_ID(ffn_moe_down-44): 341 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 370 us MUL_MAT_ID(ffn_moe_down-45): 213 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 353 us MUL_MAT_ID(ffn_moe_down-46): 188 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 372 us MUL_MAT_ID(ffn_moe_down-47): 373 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 366 us MUL_MAT_ID(ffn_moe_down-48): 194 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 378 us MUL_MAT_ID(ffn_moe_down-49): 213 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 465 us MUL_MAT_ID(ffn_moe_down-50): 177 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 369 us MUL_MAT_ID(ffn_moe_down-51): 384 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 378 us MUL_MAT_ID(ffn_moe_down-52): 221 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 409 us MUL_MAT_ID(ffn_moe_down-53): 177 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 371 us MUL_MAT_ID(ffn_moe_down-54): 189 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 399 us MUL_MAT_ID(ffn_moe_down-55): 341 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 360 us MUL_MAT_ID(ffn_moe_down-56): 225 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 398 us MUL_MAT_ID(ffn_moe_down-57): 231 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 363 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 387 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 377 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 438 us MUL_MAT_ID(ffn_moe_down-61): 205 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 342 us MUL_MAT_ID(ffn_moe_down-62): 198 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 371 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 360 us MUL_MAT_ID(ffn_moe_down-64): 248 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 393 us MUL_MAT_ID(ffn_moe_down-65): 179 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 364 us MUL_MAT_ID(ffn_moe_down-66): 207 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 534 us MUL_MAT_ID(ffn_moe_down-67): 164 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 267 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 377 us MUL_MAT_ID(ffn_moe_down-69): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 427 us MUL_MAT_ID(ffn_moe_down-70): 193 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 407 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 374 us MUL_MAT_ID(ffn_moe_down-72): 192 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 372 us MUL_MAT_ID(ffn_moe_down-73): 200 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 379 us MUL_MAT_ID(ffn_moe_down-74): 208 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 478 us MUL_MAT_ID(ffn_moe_down-75): 216 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 409 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 397 us MUL_MAT_ID(ffn_moe_down-77): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 378 us MUL_MAT_ID(ffn_moe_down-78): 190 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 370 us MUL_MAT_ID(ffn_moe_down-79): 223 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 370 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 379 us MUL_MAT_ID(ffn_moe_down-81): 214 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 403 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 420 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 368 us MUL_MAT_ID(ffn_moe_down-84): 242 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 375 us MUL_MAT_ID(ffn_moe_down-85): 241 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 507 us MUL_MAT_ID(ffn_moe_down-86): 253 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 398 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 372 us MUL_MAT_ID(ffn_moe_down-88): 193 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 388 us MUL_MAT_ID(ffn_moe_down-89): 209 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 352 us MUL_MAT_ID(ffn_moe_down-90): 251 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 370 us MUL_MAT_ID(ffn_moe_down-91): 241 us ggml_barrier(...): 54 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 430 us MUL_MAT_ID(ffn_moe_down-25): 197 us ggml_barrier(...): 464 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 407 us MUL_MAT_ID(ffn_moe_down-26): 277 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 379 us MUL_MAT_ID(ffn_moe_down-27): 631 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 412 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 399 us MUL_MAT_ID(ffn_moe_down-29): 204 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 559 us MUL_MAT_ID(ffn_moe_down-30): 211 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 405 us MUL_MAT_ID(ffn_moe_down-31): 373 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 467 us MUL_MAT_ID(ffn_moe_down-32): 282 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 449 us MUL_MAT_ID(ffn_moe_down-33): 245 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 423 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 506 us MUL_MAT_ID(ffn_moe_down-35): 182 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 421 us MUL_MAT_ID(ffn_moe_down-36): 359 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 397 us MUL_MAT_ID(ffn_moe_down-37): 245 us ggml_barrier(...): 343 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 379 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 484 us MUL_MAT_ID(ffn_moe_down-40): 260 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 357 us MUL_MAT_ID(ffn_moe_down-41): 374 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 470 us MUL_MAT_ID(ffn_moe_down-42): 158 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 370 us MUL_MAT_ID(ffn_moe_down-43): 184 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 458 us MUL_MAT_ID(ffn_moe_down-44): 256 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 367 us MUL_MAT_ID(ffn_moe_down-45): 204 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 437 us MUL_MAT_ID(ffn_moe_down-46): 369 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 396 us MUL_MAT_ID(ffn_moe_down-47): 387 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 372 us MUL_MAT_ID(ffn_moe_down-48): 203 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 503 us MUL_MAT_ID(ffn_moe_down-49): 299 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 378 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 363 us MUL_MAT_ID(ffn_moe_down-51): 201 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 406 us MUL_MAT_ID(ffn_moe_down-52): 188 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 403 us MUL_MAT_ID(ffn_moe_down-53): 228 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 419 us MUL_MAT_ID(ffn_moe_down-54): 206 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 352 us MUL_MAT_ID(ffn_moe_down-55): 236 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 400 us MUL_MAT_ID(ffn_moe_down-56): 379 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 470 us MUL_MAT_ID(ffn_moe_down-57): 195 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 379 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 442 us MUL_MAT_ID(ffn_moe_down-59): 219 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 364 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 363 us MUL_MAT_ID(ffn_moe_down-61): 243 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 400 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 374 us MUL_MAT_ID(ffn_moe_down-63): 184 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 360 us MUL_MAT_ID(ffn_moe_down-64): 165 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 412 us MUL_MAT_ID(ffn_moe_down-65): 518 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 400 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 425 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 398 us MUL_MAT_ID(ffn_moe_down-68): 196 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 401 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 493 us MUL_MAT_ID(ffn_moe_down-70): 171 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 386 us MUL_MAT_ID(ffn_moe_down-71): 255 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 386 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 363 us MUL_MAT_ID(ffn_moe_down-73): 196 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 377 us MUL_MAT_ID(ffn_moe_down-74): 178 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 465 us MUL_MAT_ID(ffn_moe_down-75): 331 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 373 us MUL_MAT_ID(ffn_moe_down-76): 197 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 373 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 363 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 377 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 380 us MUL_MAT_ID(ffn_moe_down-80): 240 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 384 us MUL_MAT_ID(ffn_moe_down-81): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 408 us MUL_MAT_ID(ffn_moe_down-82): 361 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 418 us MUL_MAT_ID(ffn_moe_down-83): 247 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 367 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 448 us MUL_MAT_ID(ffn_moe_down-85): 241 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 380 us MUL_MAT_ID(ffn_moe_down-86): 173 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 410 us MUL_MAT_ID(ffn_moe_down-87): 183 us ggml_barrier(...): 335 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 491 us MUL_MAT_ID(ffn_moe_down-88): 174 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 411 us MUL_MAT_ID(ffn_moe_down-89): 182 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 428 us MUL_MAT_ID(ffn_moe_down-90): 226 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 397 us MUL_MAT_ID(ffn_moe_down-91): 293 us ggml_barrier(...): 181 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 420 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 388 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 363 us MUL_MAT_ID(ffn_moe_down-27): 211 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 410 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 373 us MUL_MAT_ID(ffn_moe_down-30): 213 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 392 us MUL_MAT_ID(ffn_moe_down-31): 190 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 382 us MUL_MAT_ID(ffn_moe_down-32): 211 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 364 us MUL_MAT_ID(ffn_moe_down-33): 221 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 383 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 372 us MUL_MAT_ID(ffn_moe_down-35): 197 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 403 us MUL_MAT_ID(ffn_moe_down-36): 225 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 403 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 368 us MUL_MAT_ID(ffn_moe_down-38): 253 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 387 us MUL_MAT_ID(ffn_moe_down-39): 321 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 377 us MUL_MAT_ID(ffn_moe_down-40): 245 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 379 us MUL_MAT_ID(ffn_moe_down-41): 319 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 498 us MUL_MAT_ID(ffn_moe_down-42): 345 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 387 us MUL_MAT_ID(ffn_moe_down-43): 203 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 389 us MUL_MAT_ID(ffn_moe_down-44): 312 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 406 us MUL_MAT_ID(ffn_moe_down-45): 281 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 383 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 349 us MUL_MAT_ID(ffn_moe_down-47): 210 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 413 us MUL_MAT_ID(ffn_moe_down-48): 214 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 337 us MUL_MAT_ID(ffn_moe_down-49): 314 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 493 us MUL_MAT_ID(ffn_moe_down-50): 183 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 344 us MUL_MAT_ID(ffn_moe_down-51): 188 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 378 us MUL_MAT_ID(ffn_moe_down-52): 351 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 370 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 431 us MUL_MAT_ID(ffn_moe_down-54): 166 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 390 us MUL_MAT_ID(ffn_moe_down-55): 220 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 366 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 420 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 448 us MUL_MAT_ID(ffn_moe_down-58): 385 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 376 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 361 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 397 us MUL_MAT_ID(ffn_moe_down-61): 173 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 397 us MUL_MAT_ID(ffn_moe_down-62): 333 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 533 us MUL_MAT_ID(ffn_moe_down-63): 177 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 478 us MUL_MAT_ID(ffn_moe_down-64): 208 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 421 us MUL_MAT_ID(ffn_moe_down-65): 195 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 503 us MUL_MAT_ID(ffn_moe_down-66): 481 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 458 us MUL_MAT_ID(ffn_moe_down-67): 205 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 439 us MUL_MAT_ID(ffn_moe_down-68): 162 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 361 us MUL_MAT_ID(ffn_moe_down-69): 187 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 518 us MUL_MAT_ID(ffn_moe_down-70): 214 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 466 us MUL_MAT_ID(ffn_moe_down-71): 189 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 455 us MUL_MAT_ID(ffn_moe_down-72): 309 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 504 us MUL_MAT_ID(ffn_moe_down-73): 225 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 411 us MUL_MAT_ID(ffn_moe_down-74): 220 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 406 us MUL_MAT_ID(ffn_moe_down-75): 187 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 377 us MUL_MAT_ID(ffn_moe_down-76): 181 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 364 us MUL_MAT_ID(ffn_moe_down-77): 231 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 478 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 481 us MUL_MAT_ID(ffn_moe_down-79): 208 us ggml_barrier(...): 383 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 502 us MUL_MAT_ID(ffn_moe_down-80): 226 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 432 us MUL_MAT_ID(ffn_moe_down-81): 286 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 428 us MUL_MAT_ID(ffn_moe_down-82): 230 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 472 us MUL_MAT_ID(ffn_moe_down-83): 174 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 387 us MUL_MAT_ID(ffn_moe_down-84): 198 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 400 us MUL_MAT_ID(ffn_moe_down-85): 222 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 369 us MUL_MAT_ID(ffn_moe_down-86): 557 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 412 us MUL_MAT_ID(ffn_moe_down-87): 180 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 423 us MUL_MAT_ID(ffn_moe_down-88): 454 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 458 us MUL_MAT_ID(ffn_moe_down-89): 303 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 434 us MUL_MAT_ID(ffn_moe_down-90): 287 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 477 us MUL_MAT_ID(ffn_moe_down-91): 197 us ggml_barrier(...): 156 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 404 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 356 us MUL_MAT_ID(ffn_moe_down-26): 200 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 433 us MUL_MAT_ID(ffn_moe_down-27): 229 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 350 us MUL_MAT_ID(ffn_moe_down-28): 182 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 390 us MUL_MAT_ID(ffn_moe_down-29): 185 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 365 us MUL_MAT_ID(ffn_moe_down-30): 244 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 370 us MUL_MAT_ID(ffn_moe_down-31): 181 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 462 us MUL_MAT_ID(ffn_moe_down-32): 290 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 440 us MUL_MAT_ID(ffn_moe_down-33): 393 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 433 us MUL_MAT_ID(ffn_moe_down-34): 238 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 395 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 363 us MUL_MAT_ID(ffn_moe_down-36): 190 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 412 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 412 us MUL_MAT_ID(ffn_moe_down-38): 239 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 373 us MUL_MAT_ID(ffn_moe_down-39): 202 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 412 us MUL_MAT_ID(ffn_moe_down-40): 761 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 370 us MUL_MAT_ID(ffn_moe_down-41): 237 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 451 us MUL_MAT_ID(ffn_moe_down-42): 208 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 401 us MUL_MAT_ID(ffn_moe_down-43): 291 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 380 us MUL_MAT_ID(ffn_moe_down-44): 266 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 428 us MUL_MAT_ID(ffn_moe_down-45): 168 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 463 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 358 us MUL_MAT_ID(ffn_moe_down-47): 220 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 362 us MUL_MAT_ID(ffn_moe_down-48): 273 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 378 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 410 us MUL_MAT_ID(ffn_moe_down-50): 180 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 408 us MUL_MAT_ID(ffn_moe_down-51): 207 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 399 us MUL_MAT_ID(ffn_moe_down-52): 535 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 403 us MUL_MAT_ID(ffn_moe_down-53): 169 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 399 us MUL_MAT_ID(ffn_moe_down-54): 295 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 427 us MUL_MAT_ID(ffn_moe_down-55): 470 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 467 us MUL_MAT_ID(ffn_moe_down-56): 165 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 363 us MUL_MAT_ID(ffn_moe_down-57): 195 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 336 us MUL_MAT_ID(ffn_moe_down-58): 259 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 361 us MUL_MAT_ID(ffn_moe_down-59): 438 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 366 us MUL_MAT_ID(ffn_moe_down-60): 166 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 449 us MUL_MAT_ID(ffn_moe_down-61): 383 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 413 us MUL_MAT_ID(ffn_moe_down-62): 219 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 393 us MUL_MAT_ID(ffn_moe_down-63): 210 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 473 us MUL_MAT_ID(ffn_moe_down-64): 199 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 361 us MUL_MAT_ID(ffn_moe_down-65): 226 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 383 us MUL_MAT_ID(ffn_moe_down-66): 209 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 346 us MUL_MAT_ID(ffn_moe_down-67): 166 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 412 us MUL_MAT_ID(ffn_moe_down-68): 317 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 450 us MUL_MAT_ID(ffn_moe_down-69): 256 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 366 us MUL_MAT_ID(ffn_moe_down-70): 357 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 479 us MUL_MAT_ID(ffn_moe_down-71): 173 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 401 us MUL_MAT_ID(ffn_moe_down-72): 278 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 398 us MUL_MAT_ID(ffn_moe_down-73): 540 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 401 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 449 us MUL_MAT_ID(ffn_moe_down-75): 300 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 412 us MUL_MAT_ID(ffn_moe_down-76): 198 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 404 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 436 us MUL_MAT_ID(ffn_moe_down-78): 201 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 499 us MUL_MAT_ID(ffn_moe_down-79): 214 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 426 us MUL_MAT_ID(ffn_moe_down-80): 251 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 355 us MUL_MAT_ID(ffn_moe_down-81): 236 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 417 us MUL_MAT_ID(ffn_moe_down-82): 267 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 421 us MUL_MAT_ID(ffn_moe_down-83): 249 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 478 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 378 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 447 us MUL_MAT_ID(ffn_moe_down-86): 211 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 430 us MUL_MAT_ID(ffn_moe_down-87): 253 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 478 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 395 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 405 us MUL_MAT_ID(ffn_moe_down-90): 244 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 567 us MUL_MAT_ID(ffn_moe_down-91): 728 us ggml_barrier(...): 130 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 390 us MUL_MAT_ID(ffn_moe_down-25): 183 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 361 us MUL_MAT_ID(ffn_moe_down-26): 220 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 378 us MUL_MAT_ID(ffn_moe_down-27): 186 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 516 us MUL_MAT_ID(ffn_moe_down-28): 167 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 369 us MUL_MAT_ID(ffn_moe_down-29): 160 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 366 us MUL_MAT_ID(ffn_moe_down-30): 192 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 431 us MUL_MAT_ID(ffn_moe_down-31): 169 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 401 us MUL_MAT_ID(ffn_moe_down-32): 207 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 375 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 384 us MUL_MAT_ID(ffn_moe_down-34): 203 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 392 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 361 us MUL_MAT_ID(ffn_moe_down-36): 375 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 377 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 373 us MUL_MAT_ID(ffn_moe_down-38): 200 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 506 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 388 us MUL_MAT_ID(ffn_moe_down-40): 243 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 415 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 376 us MUL_MAT_ID(ffn_moe_down-42): 408 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 518 us MUL_MAT_ID(ffn_moe_down-43): 177 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 392 us MUL_MAT_ID(ffn_moe_down-44): 182 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 469 us MUL_MAT_ID(ffn_moe_down-45): 181 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 351 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 394 us MUL_MAT_ID(ffn_moe_down-47): 197 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 507 us MUL_MAT_ID(ffn_moe_down-48): 346 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 368 us MUL_MAT_ID(ffn_moe_down-49): 219 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 425 us MUL_MAT_ID(ffn_moe_down-50): 170 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 372 us MUL_MAT_ID(ffn_moe_down-51): 209 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 381 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 505 us MUL_MAT_ID(ffn_moe_down-53): 365 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 345 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 332 us MUL_MAT_ID(ffn_moe_down-55): 211 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 385 us MUL_MAT_ID(ffn_moe_down-56): 511 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 391 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 379 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 396 us MUL_MAT_ID(ffn_moe_down-59): 189 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 388 us MUL_MAT_ID(ffn_moe_down-60): 178 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 406 us MUL_MAT_ID(ffn_moe_down-61): 177 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 356 us MUL_MAT_ID(ffn_moe_down-62): 187 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 376 us MUL_MAT_ID(ffn_moe_down-63): 327 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 416 us MUL_MAT_ID(ffn_moe_down-64): 363 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 374 us MUL_MAT_ID(ffn_moe_down-65): 184 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 362 us MUL_MAT_ID(ffn_moe_down-66): 174 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 402 us MUL_MAT_ID(ffn_moe_down-67): 188 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 380 us MUL_MAT_ID(ffn_moe_down-68): 219 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 398 us MUL_MAT_ID(ffn_moe_down-69): 219 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 433 us MUL_MAT_ID(ffn_moe_down-70): 183 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 377 us MUL_MAT_ID(ffn_moe_down-71): 359 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 397 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 398 us MUL_MAT_ID(ffn_moe_down-73): 521 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 374 us MUL_MAT_ID(ffn_moe_down-74): 566 us ggml_barrier(...): 379 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 358 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 386 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 354 us MUL_MAT_ID(ffn_moe_down-77): 209 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 380 us MUL_MAT_ID(ffn_moe_down-78): 213 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 378 us MUL_MAT_ID(ffn_moe_down-79): 287 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 345 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 350 us MUL_MAT_ID(ffn_moe_down-81): 179 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 475 us MUL_MAT_ID(ffn_moe_down-82): 179 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 376 us MUL_MAT_ID(ffn_moe_down-83): 179 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 375 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 384 us MUL_MAT_ID(ffn_moe_down-85): 269 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 397 us MUL_MAT_ID(ffn_moe_down-86): 213 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 357 us MUL_MAT_ID(ffn_moe_down-87): 210 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 355 us MUL_MAT_ID(ffn_moe_down-88): 388 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 355 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 372 us MUL_MAT_ID(ffn_moe_down-90): 226 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 390 us MUL_MAT_ID(ffn_moe_down-91): 453 us ggml_barrier(...): 31 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 385 us MUL_MAT_ID(ffn_moe_down-25): 453 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 376 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 544 us MUL_MAT_ID(ffn_moe_down-27): 329 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 392 us MUL_MAT_ID(ffn_moe_down-28): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 420 us MUL_MAT_ID(ffn_moe_down-29): 183 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 376 us MUL_MAT_ID(ffn_moe_down-30): 218 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 487 us MUL_MAT_ID(ffn_moe_down-31): 193 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 356 us MUL_MAT_ID(ffn_moe_down-32): 212 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 376 us MUL_MAT_ID(ffn_moe_down-33): 211 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 363 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 417 us MUL_MAT_ID(ffn_moe_down-35): 234 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 394 us MUL_MAT_ID(ffn_moe_down-36): 216 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 384 us MUL_MAT_ID(ffn_moe_down-37): 239 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 379 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 388 us MUL_MAT_ID(ffn_moe_down-39): 289 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 421 us MUL_MAT_ID(ffn_moe_down-40): 253 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 395 us MUL_MAT_ID(ffn_moe_down-41): 276 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 410 us MUL_MAT_ID(ffn_moe_down-42): 203 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 382 us MUL_MAT_ID(ffn_moe_down-43): 206 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 397 us MUL_MAT_ID(ffn_moe_down-44): 329 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 412 us MUL_MAT_ID(ffn_moe_down-45): 224 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 432 us MUL_MAT_ID(ffn_moe_down-46): 173 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 361 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 357 us MUL_MAT_ID(ffn_moe_down-48): 201 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 422 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 383 us MUL_MAT_ID(ffn_moe_down-50): 266 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 365 us MUL_MAT_ID(ffn_moe_down-51): 205 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 364 us MUL_MAT_ID(ffn_moe_down-52): 203 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 396 us MUL_MAT_ID(ffn_moe_down-53): 214 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 392 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 404 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 359 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 356 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 382 us MUL_MAT_ID(ffn_moe_down-58): 213 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 385 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 427 us MUL_MAT_ID(ffn_moe_down-60): 282 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 366 us MUL_MAT_ID(ffn_moe_down-61): 184 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 381 us MUL_MAT_ID(ffn_moe_down-62): 201 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 397 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 338 us MUL_MAT_ID(ffn_moe_down-64): 196 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 401 us MUL_MAT_ID(ffn_moe_down-65): 243 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 384 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 392 us MUL_MAT_ID(ffn_moe_down-67): 202 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 361 us MUL_MAT_ID(ffn_moe_down-68): 203 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 406 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 350 us MUL_MAT_ID(ffn_moe_down-70): 514 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 438 us MUL_MAT_ID(ffn_moe_down-71): 298 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 358 us MUL_MAT_ID(ffn_moe_down-72): 337 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 394 us MUL_MAT_ID(ffn_moe_down-73): 395 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 387 us MUL_MAT_ID(ffn_moe_down-74): 216 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 358 us MUL_MAT_ID(ffn_moe_down-75): 194 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 385 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 389 us MUL_MAT_ID(ffn_moe_down-77): 382 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 385 us MUL_MAT_ID(ffn_moe_down-78): 206 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 378 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 380 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 400 us MUL_MAT_ID(ffn_moe_down-81): 403 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 384 us MUL_MAT_ID(ffn_moe_down-82): 507 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 357 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 380 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 370 us MUL_MAT_ID(ffn_moe_down-85): 692 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 405 us MUL_MAT_ID(ffn_moe_down-86): 258 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 402 us MUL_MAT_ID(ffn_moe_down-88): 348 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 397 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 382 us MUL_MAT_ID(ffn_moe_down-90): 245 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 382 us MUL_MAT_ID(ffn_moe_down-91): 486 us ggml_barrier(...): 29 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 536 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 400 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 394 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 415 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 385 us MUL_MAT_ID(ffn_moe_down-29): 191 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 405 us MUL_MAT_ID(ffn_moe_down-30): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 389 us MUL_MAT_ID(ffn_moe_down-31): 246 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 405 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 427 us MUL_MAT_ID(ffn_moe_down-33): 208 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 354 us MUL_MAT_ID(ffn_moe_down-34): 182 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 400 us MUL_MAT_ID(ffn_moe_down-35): 370 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 397 us MUL_MAT_ID(ffn_moe_down-36): 210 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 411 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 182 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 356 us MUL_MAT_ID(ffn_moe_down-39): 229 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 358 us MUL_MAT_ID(ffn_moe_down-40): 245 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 370 us MUL_MAT_ID(ffn_moe_down-41): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 361 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 404 us MUL_MAT_ID(ffn_moe_down-43): 219 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 361 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 383 us MUL_MAT_ID(ffn_moe_down-45): 488 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 541 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 399 us MUL_MAT_ID(ffn_moe_down-47): 402 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 397 us MUL_MAT_ID(ffn_moe_down-48): 319 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 420 us MUL_MAT_ID(ffn_moe_down-50): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 375 us MUL_MAT_ID(ffn_moe_down-51): 220 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 494 us MUL_MAT_ID(ffn_moe_down-52): 270 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 415 us MUL_MAT_ID(ffn_moe_down-53): 205 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 497 us MUL_MAT_ID(ffn_moe_down-54): 204 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 367 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 331 us MUL_MAT_ID(ffn_moe_down-56): 357 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 366 us MUL_MAT_ID(ffn_moe_down-57): 284 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 402 us MUL_MAT_ID(ffn_moe_down-58): 208 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 390 us MUL_MAT_ID(ffn_moe_down-59): 204 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 375 us MUL_MAT_ID(ffn_moe_down-60): 421 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 366 us MUL_MAT_ID(ffn_moe_down-61): 252 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 399 us MUL_MAT_ID(ffn_moe_down-62): 185 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 348 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 326 us MUL_MAT_ID(ffn_moe_down-64): 165 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 366 us MUL_MAT_ID(ffn_moe_down-65): 373 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 401 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 402 us MUL_MAT_ID(ffn_moe_down-67): 259 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 395 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 388 us MUL_MAT_ID(ffn_moe_down-69): 462 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 404 us MUL_MAT_ID(ffn_moe_down-70): 425 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 355 us MUL_MAT_ID(ffn_moe_down-71): 175 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 392 us MUL_MAT_ID(ffn_moe_down-72): 341 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 373 us MUL_MAT_ID(ffn_moe_down-73): 208 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 386 us MUL_MAT_ID(ffn_moe_down-74): 223 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 382 us MUL_MAT_ID(ffn_moe_down-75): 248 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 503 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 383 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 354 us MUL_MAT_ID(ffn_moe_down-78): 207 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 359 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 358 us MUL_MAT_ID(ffn_moe_down-80): 224 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 426 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 352 us MUL_MAT_ID(ffn_moe_down-82): 204 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 404 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 342 us MUL_MAT_ID(ffn_moe_down-84): 242 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 554 us MUL_MAT_ID(ffn_moe_down-85): 225 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 398 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 396 us MUL_MAT_ID(ffn_moe_down-87): 207 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 385 us MUL_MAT_ID(ffn_moe_down-88): 344 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 395 us MUL_MAT_ID(ffn_moe_down-89): 179 us ggml_barrier(...): 294 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 417 us MUL_MAT_ID(ffn_moe_down-90): 815 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 402 us MUL_MAT_ID(ffn_moe_down-91): 784 us ggml_barrier(...): 19 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 417 us MUL_MAT_ID(ffn_moe_down-25): 184 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 411 us MUL_MAT_ID(ffn_moe_down-26): 191 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 376 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 481 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 398 us MUL_MAT_ID(ffn_moe_down-29): 221 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 380 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 406 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 416 us MUL_MAT_ID(ffn_moe_down-32): 203 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 354 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 397 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 382 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 380 us MUL_MAT_ID(ffn_moe_down-36): 208 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 384 us MUL_MAT_ID(ffn_moe_down-37): 612 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 405 us MUL_MAT_ID(ffn_moe_down-38): 213 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 366 us MUL_MAT_ID(ffn_moe_down-39): 265 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 388 us MUL_MAT_ID(ffn_moe_down-40): 238 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 359 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 380 us MUL_MAT_ID(ffn_moe_down-42): 201 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 400 us MUL_MAT_ID(ffn_moe_down-43): 199 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 378 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 405 us MUL_MAT_ID(ffn_moe_down-45): 426 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 358 us MUL_MAT_ID(ffn_moe_down-46): 507 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 378 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 387 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 387 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 395 us MUL_MAT_ID(ffn_moe_down-50): 225 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 393 us MUL_MAT_ID(ffn_moe_down-51): 376 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 411 us MUL_MAT_ID(ffn_moe_down-52): 215 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 395 us MUL_MAT_ID(ffn_moe_down-53): 186 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 368 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 417 us MUL_MAT_ID(ffn_moe_down-55): 211 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 391 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 410 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 473 us MUL_MAT_ID(ffn_moe_down-58): 212 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 379 us MUL_MAT_ID(ffn_moe_down-59): 273 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 368 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 362 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 419 us MUL_MAT_ID(ffn_moe_down-62): 390 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 399 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 397 us MUL_MAT_ID(ffn_moe_down-64): 201 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 432 us MUL_MAT_ID(ffn_moe_down-65): 162 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 401 us MUL_MAT_ID(ffn_moe_down-66): 226 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 380 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 398 us MUL_MAT_ID(ffn_moe_down-68): 207 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 537 us MUL_MAT_ID(ffn_moe_down-69): 225 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 367 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 376 us MUL_MAT_ID(ffn_moe_down-71): 341 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 374 us MUL_MAT_ID(ffn_moe_down-72): 183 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 392 us MUL_MAT_ID(ffn_moe_down-73): 210 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 381 us MUL_MAT_ID(ffn_moe_down-74): 271 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 377 us MUL_MAT_ID(ffn_moe_down-75): 238 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 396 us MUL_MAT_ID(ffn_moe_down-76): 292 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 380 us MUL_MAT_ID(ffn_moe_down-77): 212 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 360 us MUL_MAT_ID(ffn_moe_down-78): 222 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 396 us MUL_MAT_ID(ffn_moe_down-79): 203 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 387 us MUL_MAT_ID(ffn_moe_down-80): 253 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 365 us MUL_MAT_ID(ffn_moe_down-81): 170 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 397 us MUL_MAT_ID(ffn_moe_down-82): 336 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 375 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 375 us MUL_MAT_ID(ffn_moe_down-84): 568 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 388 us MUL_MAT_ID(ffn_moe_down-85): 287 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 385 us MUL_MAT_ID(ffn_moe_down-86): 242 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 351 us MUL_MAT_ID(ffn_moe_down-88): 189 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 395 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 400 us MUL_MAT_ID(ffn_moe_down-90): 247 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 389 us MUL_MAT_ID(ffn_moe_down-91): 246 us ggml_barrier(...): 85 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 415 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 364 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 363 us MUL_MAT_ID(ffn_moe_down-27): 213 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 357 us MUL_MAT_ID(ffn_moe_down-28): 213 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 394 us MUL_MAT_ID(ffn_moe_down-29): 207 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 381 us MUL_MAT_ID(ffn_moe_down-30): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 375 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 386 us MUL_MAT_ID(ffn_moe_down-32): 225 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 363 us MUL_MAT_ID(ffn_moe_down-33): 213 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 373 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 370 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 385 us MUL_MAT_ID(ffn_moe_down-36): 224 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 370 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 397 us MUL_MAT_ID(ffn_moe_down-38): 202 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 403 us MUL_MAT_ID(ffn_moe_down-39): 252 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 395 us MUL_MAT_ID(ffn_moe_down-40): 271 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 406 us MUL_MAT_ID(ffn_moe_down-41): 206 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 412 us MUL_MAT_ID(ffn_moe_down-42): 233 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 399 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 399 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 390 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 349 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 372 us MUL_MAT_ID(ffn_moe_down-47): 208 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 402 us MUL_MAT_ID(ffn_moe_down-48): 205 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 381 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 364 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 381 us MUL_MAT_ID(ffn_moe_down-52): 351 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 392 us MUL_MAT_ID(ffn_moe_down-53): 224 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 383 us MUL_MAT_ID(ffn_moe_down-54): 209 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 383 us MUL_MAT_ID(ffn_moe_down-55): 387 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 395 us MUL_MAT_ID(ffn_moe_down-56): 205 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 392 us MUL_MAT_ID(ffn_moe_down-57): 172 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 378 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 410 us MUL_MAT_ID(ffn_moe_down-59): 226 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 353 us MUL_MAT_ID(ffn_moe_down-60): 179 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 421 us MUL_MAT_ID(ffn_moe_down-61): 294 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 377 us MUL_MAT_ID(ffn_moe_down-62): 203 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 352 us MUL_MAT_ID(ffn_moe_down-63): 274 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 384 us MUL_MAT_ID(ffn_moe_down-64): 202 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 382 us MUL_MAT_ID(ffn_moe_down-65): 188 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 385 us MUL_MAT_ID(ffn_moe_down-66): 254 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 400 us MUL_MAT_ID(ffn_moe_down-67): 317 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 352 us MUL_MAT_ID(ffn_moe_down-68): 381 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 364 us MUL_MAT_ID(ffn_moe_down-69): 186 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 350 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 370 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 402 us MUL_MAT_ID(ffn_moe_down-72): 188 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 359 us MUL_MAT_ID(ffn_moe_down-73): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 396 us MUL_MAT_ID(ffn_moe_down-74): 208 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 354 us MUL_MAT_ID(ffn_moe_down-75): 209 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 395 us MUL_MAT_ID(ffn_moe_down-76): 184 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 372 us MUL_MAT_ID(ffn_moe_down-77): 204 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 396 us MUL_MAT_ID(ffn_moe_down-78): 411 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 395 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 389 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 378 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 358 us MUL_MAT_ID(ffn_moe_down-82): 224 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 353 us MUL_MAT_ID(ffn_moe_down-83): 221 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 373 us MUL_MAT_ID(ffn_moe_down-84): 225 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 428 us MUL_MAT_ID(ffn_moe_down-85): 240 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 406 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 367 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 402 us MUL_MAT_ID(ffn_moe_down-88): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 366 us MUL_MAT_ID(ffn_moe_down-89): 363 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 383 us MUL_MAT_ID(ffn_moe_down-90): 219 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 360 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 43 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 380 us MUL_MAT_ID(ffn_moe_down-25): 186 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 396 us MUL_MAT_ID(ffn_moe_down-26): 212 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 383 us MUL_MAT_ID(ffn_moe_down-27): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 409 us MUL_MAT_ID(ffn_moe_down-28): 213 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 420 us MUL_MAT_ID(ffn_moe_down-29): 416 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 394 us MUL_MAT_ID(ffn_moe_down-30): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 372 us MUL_MAT_ID(ffn_moe_down-31): 534 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 365 us MUL_MAT_ID(ffn_moe_down-32): 312 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 365 us MUL_MAT_ID(ffn_moe_down-33): 281 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 377 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 407 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 385 us MUL_MAT_ID(ffn_moe_down-36): 242 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 372 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 364 us MUL_MAT_ID(ffn_moe_down-38): 186 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 363 us MUL_MAT_ID(ffn_moe_down-39): 259 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 399 us MUL_MAT_ID(ffn_moe_down-40): 393 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 403 us MUL_MAT_ID(ffn_moe_down-41): 207 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 389 us MUL_MAT_ID(ffn_moe_down-42): 244 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 381 us MUL_MAT_ID(ffn_moe_down-43): 207 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 474 us MUL_MAT_ID(ffn_moe_down-44): 279 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 417 us MUL_MAT_ID(ffn_moe_down-45): 203 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 376 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 483 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 362 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 400 us MUL_MAT_ID(ffn_moe_down-49): 422 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 373 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 401 us MUL_MAT_ID(ffn_moe_down-51): 215 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 403 us MUL_MAT_ID(ffn_moe_down-52): 220 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 368 us MUL_MAT_ID(ffn_moe_down-53): 206 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 357 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 422 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 389 us MUL_MAT_ID(ffn_moe_down-56): 266 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 404 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 393 us MUL_MAT_ID(ffn_moe_down-58): 350 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 365 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 432 us MUL_MAT_ID(ffn_moe_down-60): 259 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 378 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 407 us MUL_MAT_ID(ffn_moe_down-62): 343 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 398 us MUL_MAT_ID(ffn_moe_down-63): 194 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 399 us MUL_MAT_ID(ffn_moe_down-64): 477 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 387 us MUL_MAT_ID(ffn_moe_down-65): 428 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 403 us MUL_MAT_ID(ffn_moe_down-66): 203 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 424 us MUL_MAT_ID(ffn_moe_down-67): 202 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 393 us MUL_MAT_ID(ffn_moe_down-68): 197 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 406 us MUL_MAT_ID(ffn_moe_down-69): 304 us ggml_barrier(...): 372 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 414 us MUL_MAT_ID(ffn_moe_down-70): 215 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 362 us MUL_MAT_ID(ffn_moe_down-71): 202 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 454 us MUL_MAT_ID(ffn_moe_down-72): 192 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 184 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 372 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 388 us MUL_MAT_ID(ffn_moe_down-75): 199 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 365 us MUL_MAT_ID(ffn_moe_down-76): 474 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 392 us MUL_MAT_ID(ffn_moe_down-77): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 362 us MUL_MAT_ID(ffn_moe_down-78): 406 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 363 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 390 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 396 us MUL_MAT_ID(ffn_moe_down-81): 181 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 362 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 409 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 371 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 408 us MUL_MAT_ID(ffn_moe_down-85): 256 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 371 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 386 us MUL_MAT_ID(ffn_moe_down-87): 207 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 421 us MUL_MAT_ID(ffn_moe_down-88): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 363 us MUL_MAT_ID(ffn_moe_down-89): 278 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 354 us MUL_MAT_ID(ffn_moe_down-90): 250 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 382 us MUL_MAT_ID(ffn_moe_down-91): 667 us ggml_barrier(...): 29 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 402 us MUL_MAT_ID(ffn_moe_down-25): 333 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 377 us MUL_MAT_ID(ffn_moe_down-26): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 405 us MUL_MAT_ID(ffn_moe_down-27): 529 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 366 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 385 us MUL_MAT_ID(ffn_moe_down-29): 383 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 384 us MUL_MAT_ID(ffn_moe_down-30): 202 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 379 us MUL_MAT_ID(ffn_moe_down-31): 226 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 393 us MUL_MAT_ID(ffn_moe_down-32): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 423 us MUL_MAT_ID(ffn_moe_down-33): 401 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 402 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 349 us MUL_MAT_ID(ffn_moe_down-35): 184 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 358 us MUL_MAT_ID(ffn_moe_down-36): 179 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 355 us MUL_MAT_ID(ffn_moe_down-37): 244 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 401 us MUL_MAT_ID(ffn_moe_down-38): 275 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 365 us MUL_MAT_ID(ffn_moe_down-39): 231 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 379 us MUL_MAT_ID(ffn_moe_down-40): 235 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 378 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 410 us MUL_MAT_ID(ffn_moe_down-42): 183 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 358 us MUL_MAT_ID(ffn_moe_down-43): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 385 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 383 us MUL_MAT_ID(ffn_moe_down-45): 215 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 398 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 377 us MUL_MAT_ID(ffn_moe_down-47): 207 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 353 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 366 us MUL_MAT_ID(ffn_moe_down-49): 222 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 407 us MUL_MAT_ID(ffn_moe_down-50): 223 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 374 us MUL_MAT_ID(ffn_moe_down-51): 206 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 363 us MUL_MAT_ID(ffn_moe_down-52): 513 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 389 us MUL_MAT_ID(ffn_moe_down-53): 196 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 422 us MUL_MAT_ID(ffn_moe_down-54): 205 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 376 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 385 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 539 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 410 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 362 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 381 us MUL_MAT_ID(ffn_moe_down-60): 213 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 378 us MUL_MAT_ID(ffn_moe_down-61): 352 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 363 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 368 us MUL_MAT_ID(ffn_moe_down-63): 163 us ggml_barrier(...): 336 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 373 us MUL_MAT_ID(ffn_moe_down-64): 205 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 352 us MUL_MAT_ID(ffn_moe_down-65): 185 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 360 us MUL_MAT_ID(ffn_moe_down-66): 216 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 374 us MUL_MAT_ID(ffn_moe_down-67): 205 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 368 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 374 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 367 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 356 us MUL_MAT_ID(ffn_moe_down-71): 208 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 397 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 370 us MUL_MAT_ID(ffn_moe_down-73): 193 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 409 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 362 us MUL_MAT_ID(ffn_moe_down-75): 187 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 383 us MUL_MAT_ID(ffn_moe_down-76): 220 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 377 us MUL_MAT_ID(ffn_moe_down-77): 212 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 354 us MUL_MAT_ID(ffn_moe_down-78): 191 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 372 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 374 us MUL_MAT_ID(ffn_moe_down-80): 536 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 371 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 361 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 387 us MUL_MAT_ID(ffn_moe_down-83): 313 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 363 us MUL_MAT_ID(ffn_moe_down-84): 244 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 375 us MUL_MAT_ID(ffn_moe_down-85): 234 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 362 us MUL_MAT_ID(ffn_moe_down-86): 235 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 421 us MUL_MAT_ID(ffn_moe_down-87): 186 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 371 us MUL_MAT_ID(ffn_moe_down-88): 213 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 384 us MUL_MAT_ID(ffn_moe_down-89): 204 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 372 us MUL_MAT_ID(ffn_moe_down-90): 243 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 369 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 49 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 372 us MUL_MAT_ID(ffn_moe_down-25): 219 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 399 us MUL_MAT_ID(ffn_moe_down-26): 374 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 394 us MUL_MAT_ID(ffn_moe_down-27): 505 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 393 us MUL_MAT_ID(ffn_moe_down-28): 203 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 398 us MUL_MAT_ID(ffn_moe_down-29): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 389 us MUL_MAT_ID(ffn_moe_down-30): 428 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 382 us MUL_MAT_ID(ffn_moe_down-31): 359 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 486 us MUL_MAT_ID(ffn_moe_down-32): 212 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 437 us MUL_MAT_ID(ffn_moe_down-33): 254 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 442 us MUL_MAT_ID(ffn_moe_down-34): 248 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 454 us MUL_MAT_ID(ffn_moe_down-35): 245 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 358 us MUL_MAT_ID(ffn_moe_down-36): 220 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 412 us MUL_MAT_ID(ffn_moe_down-37): 328 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 396 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 390 us MUL_MAT_ID(ffn_moe_down-39): 738 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 453 us MUL_MAT_ID(ffn_moe_down-40): 185 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 412 us MUL_MAT_ID(ffn_moe_down-41): 536 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 404 us MUL_MAT_ID(ffn_moe_down-42): 202 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 396 us MUL_MAT_ID(ffn_moe_down-43): 187 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 342 us MUL_MAT_ID(ffn_moe_down-44): 211 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 428 us MUL_MAT_ID(ffn_moe_down-45): 194 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 458 us MUL_MAT_ID(ffn_moe_down-46): 221 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 412 us MUL_MAT_ID(ffn_moe_down-47): 220 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 454 us MUL_MAT_ID(ffn_moe_down-48): 300 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 372 us MUL_MAT_ID(ffn_moe_down-49): 169 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 563 us MUL_MAT_ID(ffn_moe_down-50): 216 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 458 us MUL_MAT_ID(ffn_moe_down-51): 304 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 346 us MUL_MAT_ID(ffn_moe_down-52): 197 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 396 us MUL_MAT_ID(ffn_moe_down-53): 367 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 374 us MUL_MAT_ID(ffn_moe_down-54): 250 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 390 us MUL_MAT_ID(ffn_moe_down-55): 176 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 399 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 413 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 413 us MUL_MAT_ID(ffn_moe_down-58): 204 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 388 us MUL_MAT_ID(ffn_moe_down-59): 213 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 257 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 378 us MUL_MAT_ID(ffn_moe_down-61): 247 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 412 us MUL_MAT_ID(ffn_moe_down-62): 242 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 378 us MUL_MAT_ID(ffn_moe_down-63): 205 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 389 us MUL_MAT_ID(ffn_moe_down-64): 366 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 379 us MUL_MAT_ID(ffn_moe_down-65): 300 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 398 us MUL_MAT_ID(ffn_moe_down-66): 198 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 380 us MUL_MAT_ID(ffn_moe_down-67): 227 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 409 us MUL_MAT_ID(ffn_moe_down-68): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 426 us MUL_MAT_ID(ffn_moe_down-69): 207 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 443 us MUL_MAT_ID(ffn_moe_down-70): 272 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 406 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 442 us MUL_MAT_ID(ffn_moe_down-72): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 360 us MUL_MAT_ID(ffn_moe_down-73): 321 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 369 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 451 us MUL_MAT_ID(ffn_moe_down-75): 253 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 401 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 381 us MUL_MAT_ID(ffn_moe_down-77): 260 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 425 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 397 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 405 us MUL_MAT_ID(ffn_moe_down-80): 328 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 403 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 400 us MUL_MAT_ID(ffn_moe_down-82): 184 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 377 us MUL_MAT_ID(ffn_moe_down-83): 186 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 366 us MUL_MAT_ID(ffn_moe_down-84): 248 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 416 us MUL_MAT_ID(ffn_moe_down-85): 253 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 382 us MUL_MAT_ID(ffn_moe_down-86): 204 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 379 us MUL_MAT_ID(ffn_moe_down-87): 402 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 370 us MUL_MAT_ID(ffn_moe_down-88): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 381 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 350 us MUL_MAT_ID(ffn_moe_down-90): 449 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 386 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 31 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 402 us MUL_MAT_ID(ffn_moe_down-25): 175 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 399 us MUL_MAT_ID(ffn_moe_down-26): 209 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 392 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 351 us MUL_MAT_ID(ffn_moe_down-28): 183 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 394 us MUL_MAT_ID(ffn_moe_down-29): 341 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 357 us MUL_MAT_ID(ffn_moe_down-30): 288 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 350 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 406 us MUL_MAT_ID(ffn_moe_down-32): 181 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 188 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 376 us MUL_MAT_ID(ffn_moe_down-34): 205 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 394 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 412 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 468 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 364 us MUL_MAT_ID(ffn_moe_down-38): 177 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 358 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 371 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 399 us MUL_MAT_ID(ffn_moe_down-41): 211 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 397 us MUL_MAT_ID(ffn_moe_down-42): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 391 us MUL_MAT_ID(ffn_moe_down-43): 203 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 399 us MUL_MAT_ID(ffn_moe_down-44): 229 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 387 us MUL_MAT_ID(ffn_moe_down-45): 333 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 378 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 407 us MUL_MAT_ID(ffn_moe_down-47): 210 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 359 us MUL_MAT_ID(ffn_moe_down-48): 187 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 379 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 372 us MUL_MAT_ID(ffn_moe_down-50): 206 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 381 us MUL_MAT_ID(ffn_moe_down-51): 405 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 393 us MUL_MAT_ID(ffn_moe_down-52): 219 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 391 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 393 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 395 us MUL_MAT_ID(ffn_moe_down-55): 193 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 381 us MUL_MAT_ID(ffn_moe_down-56): 189 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 385 us MUL_MAT_ID(ffn_moe_down-57): 208 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 349 us MUL_MAT_ID(ffn_moe_down-58): 374 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 384 us MUL_MAT_ID(ffn_moe_down-59): 205 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 367 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 392 us MUL_MAT_ID(ffn_moe_down-61): 185 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 342 us MUL_MAT_ID(ffn_moe_down-62): 198 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 333 us MUL_MAT_ID(ffn_moe_down-63): 518 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 391 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 392 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 354 us MUL_MAT_ID(ffn_moe_down-66): 212 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 380 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 355 us MUL_MAT_ID(ffn_moe_down-68): 521 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 373 us MUL_MAT_ID(ffn_moe_down-69): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 406 us MUL_MAT_ID(ffn_moe_down-70): 174 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 389 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 380 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 364 us MUL_MAT_ID(ffn_moe_down-73): 200 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 389 us MUL_MAT_ID(ffn_moe_down-74): 216 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 374 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 359 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 350 us MUL_MAT_ID(ffn_moe_down-77): 215 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 388 us MUL_MAT_ID(ffn_moe_down-78): 248 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 378 us MUL_MAT_ID(ffn_moe_down-79): 189 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 346 us MUL_MAT_ID(ffn_moe_down-80): 268 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 474 us MUL_MAT_ID(ffn_moe_down-81): 265 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 356 us MUL_MAT_ID(ffn_moe_down-82): 181 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 387 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 380 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 474 us MUL_MAT_ID(ffn_moe_down-85): 401 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 402 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 398 us MUL_MAT_ID(ffn_moe_down-87): 223 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 389 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 405 us MUL_MAT_ID(ffn_moe_down-89): 186 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 413 us MUL_MAT_ID(ffn_moe_down-90): 225 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 391 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 132 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 368 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 399 us MUL_MAT_ID(ffn_moe_down-26): 188 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 370 us MUL_MAT_ID(ffn_moe_down-27): 319 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 373 us MUL_MAT_ID(ffn_moe_down-28): 361 us ggml_barrier(...): 337 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 377 us MUL_MAT_ID(ffn_moe_down-29): 181 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 464 us MUL_MAT_ID(ffn_moe_down-30): 293 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 378 us MUL_MAT_ID(ffn_moe_down-31): 228 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 386 us MUL_MAT_ID(ffn_moe_down-32): 188 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 522 us MUL_MAT_ID(ffn_moe_down-33): 162 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 450 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 369 us MUL_MAT_ID(ffn_moe_down-35): 184 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 438 us MUL_MAT_ID(ffn_moe_down-36): 155 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 374 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 355 us MUL_MAT_ID(ffn_moe_down-39): 698 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 376 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 404 us MUL_MAT_ID(ffn_moe_down-41): 209 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 379 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 362 us MUL_MAT_ID(ffn_moe_down-43): 259 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 380 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 428 us MUL_MAT_ID(ffn_moe_down-45): 222 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 469 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 390 us MUL_MAT_ID(ffn_moe_down-47): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 374 us MUL_MAT_ID(ffn_moe_down-48): 323 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 392 us MUL_MAT_ID(ffn_moe_down-49): 438 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 346 us MUL_MAT_ID(ffn_moe_down-50): 239 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 362 us MUL_MAT_ID(ffn_moe_down-51): 233 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 410 us MUL_MAT_ID(ffn_moe_down-52): 241 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 392 us MUL_MAT_ID(ffn_moe_down-53): 271 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 485 us MUL_MAT_ID(ffn_moe_down-54): 207 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 430 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 382 us MUL_MAT_ID(ffn_moe_down-56): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 400 us MUL_MAT_ID(ffn_moe_down-57): 185 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 386 us MUL_MAT_ID(ffn_moe_down-58): 180 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 366 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 362 us MUL_MAT_ID(ffn_moe_down-60): 371 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 369 us MUL_MAT_ID(ffn_moe_down-61): 188 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 427 us MUL_MAT_ID(ffn_moe_down-62): 173 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 384 us MUL_MAT_ID(ffn_moe_down-63): 340 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 377 us MUL_MAT_ID(ffn_moe_down-64): 380 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 393 us MUL_MAT_ID(ffn_moe_down-65): 254 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 230 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 378 us MUL_MAT_ID(ffn_moe_down-67): 384 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 389 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 471 us MUL_MAT_ID(ffn_moe_down-69): 376 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 407 us MUL_MAT_ID(ffn_moe_down-70): 183 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 402 us MUL_MAT_ID(ffn_moe_down-71): 189 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 365 us MUL_MAT_ID(ffn_moe_down-72): 250 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 374 us MUL_MAT_ID(ffn_moe_down-73): 345 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 384 us MUL_MAT_ID(ffn_moe_down-74): 304 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 371 us MUL_MAT_ID(ffn_moe_down-75): 246 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 381 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 334 us MUL_MAT_ID(ffn_moe_down-77): 222 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 523 us MUL_MAT_ID(ffn_moe_down-78): 276 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 398 us MUL_MAT_ID(ffn_moe_down-79): 223 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 509 us MUL_MAT_ID(ffn_moe_down-80): 165 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 398 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 414 us MUL_MAT_ID(ffn_moe_down-82): 354 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 404 us MUL_MAT_ID(ffn_moe_down-83): 209 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 370 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 391 us MUL_MAT_ID(ffn_moe_down-85): 217 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 419 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 407 us MUL_MAT_ID(ffn_moe_down-87): 314 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 369 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 409 us MUL_MAT_ID(ffn_moe_down-89): 218 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 386 us MUL_MAT_ID(ffn_moe_down-90): 236 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 408 us MUL_MAT_ID(ffn_moe_down-91): 297 us ggml_barrier(...): 49 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 432 us MUL_MAT_ID(ffn_moe_down-25): 168 us ggml_barrier(...): 448 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 403 us MUL_MAT_ID(ffn_moe_down-26): 204 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 428 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 485 us MUL_MAT_ID(ffn_moe_down-28): 304 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 404 us MUL_MAT_ID(ffn_moe_down-29): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 387 us MUL_MAT_ID(ffn_moe_down-30): 208 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 358 us MUL_MAT_ID(ffn_moe_down-31): 384 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 377 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 367 us MUL_MAT_ID(ffn_moe_down-33): 245 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 393 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 405 us MUL_MAT_ID(ffn_moe_down-35): 202 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 374 us MUL_MAT_ID(ffn_moe_down-36): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 423 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 370 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 397 us MUL_MAT_ID(ffn_moe_down-39): 247 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 420 us MUL_MAT_ID(ffn_moe_down-40): 268 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 392 us MUL_MAT_ID(ffn_moe_down-41): 204 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 357 us MUL_MAT_ID(ffn_moe_down-42): 221 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 376 us MUL_MAT_ID(ffn_moe_down-43): 184 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 395 us MUL_MAT_ID(ffn_moe_down-44): 202 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 368 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 404 us MUL_MAT_ID(ffn_moe_down-46): 212 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 383 us MUL_MAT_ID(ffn_moe_down-47): 355 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 369 us MUL_MAT_ID(ffn_moe_down-48): 205 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 507 us MUL_MAT_ID(ffn_moe_down-49): 460 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 371 us MUL_MAT_ID(ffn_moe_down-50): 357 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 413 us MUL_MAT_ID(ffn_moe_down-51): 202 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 368 us MUL_MAT_ID(ffn_moe_down-52): 227 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 380 us MUL_MAT_ID(ffn_moe_down-53): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 374 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 399 us MUL_MAT_ID(ffn_moe_down-55): 280 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 388 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 393 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 369 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 345 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 351 us MUL_MAT_ID(ffn_moe_down-60): 374 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 363 us MUL_MAT_ID(ffn_moe_down-61): 202 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 373 us MUL_MAT_ID(ffn_moe_down-62): 373 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 392 us MUL_MAT_ID(ffn_moe_down-63): 229 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 353 us MUL_MAT_ID(ffn_moe_down-64): 208 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 373 us MUL_MAT_ID(ffn_moe_down-65): 207 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 377 us MUL_MAT_ID(ffn_moe_down-66): 237 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 381 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 366 us MUL_MAT_ID(ffn_moe_down-68): 208 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 404 us MUL_MAT_ID(ffn_moe_down-69): 258 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 402 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 385 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 371 us MUL_MAT_ID(ffn_moe_down-72): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 403 us MUL_MAT_ID(ffn_moe_down-73): 184 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 381 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 349 us MUL_MAT_ID(ffn_moe_down-75): 248 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 371 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 361 us MUL_MAT_ID(ffn_moe_down-77): 207 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 348 us MUL_MAT_ID(ffn_moe_down-78): 206 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 386 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 352 us MUL_MAT_ID(ffn_moe_down-80): 299 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 380 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 350 us MUL_MAT_ID(ffn_moe_down-82): 222 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 384 us MUL_MAT_ID(ffn_moe_down-83): 201 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 374 us MUL_MAT_ID(ffn_moe_down-84): 512 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 375 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 423 us MUL_MAT_ID(ffn_moe_down-86): 218 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 410 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 405 us MUL_MAT_ID(ffn_moe_down-88): 518 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 385 us MUL_MAT_ID(ffn_moe_down-89): 392 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 407 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 402 us MUL_MAT_ID(ffn_moe_down-91): 257 us ggml_barrier(...): 23 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 414 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 420 us MUL_MAT_ID(ffn_moe_down-26): 206 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 401 us MUL_MAT_ID(ffn_moe_down-27): 215 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 378 us MUL_MAT_ID(ffn_moe_down-28): 326 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 399 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 391 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 357 us MUL_MAT_ID(ffn_moe_down-31): 202 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 508 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 421 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 397 us MUL_MAT_ID(ffn_moe_down-34): 210 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 372 us MUL_MAT_ID(ffn_moe_down-35): 201 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 481 us MUL_MAT_ID(ffn_moe_down-36): 226 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 393 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 385 us MUL_MAT_ID(ffn_moe_down-38): 208 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 397 us MUL_MAT_ID(ffn_moe_down-39): 317 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 358 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 375 us MUL_MAT_ID(ffn_moe_down-41): 199 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 415 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 383 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 357 us MUL_MAT_ID(ffn_moe_down-44): 433 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 359 us MUL_MAT_ID(ffn_moe_down-45): 372 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 361 us MUL_MAT_ID(ffn_moe_down-46): 188 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 362 us MUL_MAT_ID(ffn_moe_down-47): 185 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 387 us MUL_MAT_ID(ffn_moe_down-48): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 343 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 371 us MUL_MAT_ID(ffn_moe_down-50): 193 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 352 us MUL_MAT_ID(ffn_moe_down-51): 212 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 364 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 380 us MUL_MAT_ID(ffn_moe_down-53): 362 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 387 us MUL_MAT_ID(ffn_moe_down-54): 208 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 390 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 385 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 469 us MUL_MAT_ID(ffn_moe_down-57): 488 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 358 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 359 us MUL_MAT_ID(ffn_moe_down-59): 210 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 399 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 402 us MUL_MAT_ID(ffn_moe_down-61): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 379 us MUL_MAT_ID(ffn_moe_down-62): 177 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 379 us MUL_MAT_ID(ffn_moe_down-63): 205 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 469 us MUL_MAT_ID(ffn_moe_down-64): 251 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 395 us MUL_MAT_ID(ffn_moe_down-65): 217 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 482 us MUL_MAT_ID(ffn_moe_down-66): 163 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 357 us MUL_MAT_ID(ffn_moe_down-67): 364 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 429 us MUL_MAT_ID(ffn_moe_down-68): 291 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 423 us MUL_MAT_ID(ffn_moe_down-69): 401 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 402 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 451 us MUL_MAT_ID(ffn_moe_down-71): 187 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 362 us MUL_MAT_ID(ffn_moe_down-72): 186 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 494 us MUL_MAT_ID(ffn_moe_down-73): 171 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 385 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 414 us MUL_MAT_ID(ffn_moe_down-75): 276 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 417 us MUL_MAT_ID(ffn_moe_down-76): 360 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 369 us MUL_MAT_ID(ffn_moe_down-77): 196 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 376 us MUL_MAT_ID(ffn_moe_down-78): 186 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 389 us MUL_MAT_ID(ffn_moe_down-79): 336 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 369 us MUL_MAT_ID(ffn_moe_down-80): 201 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 414 us MUL_MAT_ID(ffn_moe_down-81): 181 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 416 us MUL_MAT_ID(ffn_moe_down-82): 268 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 454 us MUL_MAT_ID(ffn_moe_down-83): 431 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 501 us MUL_MAT_ID(ffn_moe_down-84): 618 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 447 us MUL_MAT_ID(ffn_moe_down-85): 265 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 444 us MUL_MAT_ID(ffn_moe_down-86): 211 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 404 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 396 us MUL_MAT_ID(ffn_moe_down-88): 232 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 382 us MUL_MAT_ID(ffn_moe_down-89): 249 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 386 us MUL_MAT_ID(ffn_moe_down-90): 249 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 468 us MUL_MAT_ID(ffn_moe_down-91): 631 us ggml_barrier(...): 152 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 426 us MUL_MAT_ID(ffn_moe_down-25): 193 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 382 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 399 us MUL_MAT_ID(ffn_moe_down-27): 217 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 352 us MUL_MAT_ID(ffn_moe_down-28): 236 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 402 us MUL_MAT_ID(ffn_moe_down-29): 216 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 391 us MUL_MAT_ID(ffn_moe_down-30): 419 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 366 us MUL_MAT_ID(ffn_moe_down-31): 208 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 374 us MUL_MAT_ID(ffn_moe_down-32): 188 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 420 us MUL_MAT_ID(ffn_moe_down-33): 238 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 399 us MUL_MAT_ID(ffn_moe_down-34): 207 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 376 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 380 us MUL_MAT_ID(ffn_moe_down-36): 201 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 360 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 403 us MUL_MAT_ID(ffn_moe_down-38): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 378 us MUL_MAT_ID(ffn_moe_down-39): 227 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 366 us MUL_MAT_ID(ffn_moe_down-40): 248 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 433 us MUL_MAT_ID(ffn_moe_down-41): 200 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 396 us MUL_MAT_ID(ffn_moe_down-42): 193 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 386 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 396 us MUL_MAT_ID(ffn_moe_down-44): 205 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 356 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 395 us MUL_MAT_ID(ffn_moe_down-46): 272 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 380 us MUL_MAT_ID(ffn_moe_down-47): 198 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 350 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 394 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 377 us MUL_MAT_ID(ffn_moe_down-50): 189 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 372 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 392 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 344 us MUL_MAT_ID(ffn_moe_down-53): 176 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 398 us MUL_MAT_ID(ffn_moe_down-54): 211 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 392 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 407 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 406 us MUL_MAT_ID(ffn_moe_down-57): 308 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 388 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 373 us MUL_MAT_ID(ffn_moe_down-59): 261 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 416 us MUL_MAT_ID(ffn_moe_down-60): 204 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 371 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 399 us MUL_MAT_ID(ffn_moe_down-62): 185 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 361 us MUL_MAT_ID(ffn_moe_down-63): 228 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 379 us MUL_MAT_ID(ffn_moe_down-64): 371 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 340 us MUL_MAT_ID(ffn_moe_down-65): 202 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 359 us MUL_MAT_ID(ffn_moe_down-66): 377 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 414 us MUL_MAT_ID(ffn_moe_down-67): 403 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 382 us MUL_MAT_ID(ffn_moe_down-68): 372 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 401 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 388 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 373 us MUL_MAT_ID(ffn_moe_down-71): 422 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 401 us MUL_MAT_ID(ffn_moe_down-72): 182 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 351 us MUL_MAT_ID(ffn_moe_down-73): 328 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 388 us MUL_MAT_ID(ffn_moe_down-74): 211 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 402 us MUL_MAT_ID(ffn_moe_down-75): 223 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 373 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 372 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 403 us MUL_MAT_ID(ffn_moe_down-78): 212 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 385 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 440 us MUL_MAT_ID(ffn_moe_down-80): 206 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 377 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 362 us MUL_MAT_ID(ffn_moe_down-82): 196 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 395 us MUL_MAT_ID(ffn_moe_down-83): 360 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 397 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 406 us MUL_MAT_ID(ffn_moe_down-85): 237 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 486 us MUL_MAT_ID(ffn_moe_down-86): 241 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 404 us MUL_MAT_ID(ffn_moe_down-87): 227 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 385 us MUL_MAT_ID(ffn_moe_down-88): 297 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 392 us MUL_MAT_ID(ffn_moe_down-89): 209 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 395 us MUL_MAT_ID(ffn_moe_down-90): 227 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 380 us MUL_MAT_ID(ffn_moe_down-91): 300 us ggml_barrier(...): 148 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 385 us MUL_MAT_ID(ffn_moe_down-25): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 378 us MUL_MAT_ID(ffn_moe_down-26): 183 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 385 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 385 us MUL_MAT_ID(ffn_moe_down-28): 310 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 494 us MUL_MAT_ID(ffn_moe_down-29): 174 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 393 us MUL_MAT_ID(ffn_moe_down-30): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 372 us MUL_MAT_ID(ffn_moe_down-31): 204 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 393 us MUL_MAT_ID(ffn_moe_down-32): 215 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 422 us MUL_MAT_ID(ffn_moe_down-33): 240 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 420 us MUL_MAT_ID(ffn_moe_down-34): 189 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 382 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 353 us MUL_MAT_ID(ffn_moe_down-36): 206 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 400 us MUL_MAT_ID(ffn_moe_down-37): 219 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 375 us MUL_MAT_ID(ffn_moe_down-38): 239 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 373 us MUL_MAT_ID(ffn_moe_down-39): 271 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 374 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 401 us MUL_MAT_ID(ffn_moe_down-41): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 394 us MUL_MAT_ID(ffn_moe_down-42): 211 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 369 us MUL_MAT_ID(ffn_moe_down-43): 533 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 391 us MUL_MAT_ID(ffn_moe_down-44): 257 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 370 us MUL_MAT_ID(ffn_moe_down-45): 525 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 381 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 385 us MUL_MAT_ID(ffn_moe_down-47): 271 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 363 us MUL_MAT_ID(ffn_moe_down-48): 185 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 408 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 378 us MUL_MAT_ID(ffn_moe_down-50): 209 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 395 us MUL_MAT_ID(ffn_moe_down-51): 353 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 425 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 433 us MUL_MAT_ID(ffn_moe_down-53): 162 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 404 us MUL_MAT_ID(ffn_moe_down-54): 368 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 202 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 400 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 374 us MUL_MAT_ID(ffn_moe_down-57): 214 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 396 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 398 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 400 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 396 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 394 us MUL_MAT_ID(ffn_moe_down-62): 210 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 412 us MUL_MAT_ID(ffn_moe_down-63): 201 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 376 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 372 us MUL_MAT_ID(ffn_moe_down-65): 161 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 372 us MUL_MAT_ID(ffn_moe_down-66): 185 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 394 us MUL_MAT_ID(ffn_moe_down-67): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 363 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 402 us MUL_MAT_ID(ffn_moe_down-69): 384 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 376 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 409 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 511 us MUL_MAT_ID(ffn_moe_down-72): 180 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 432 us MUL_MAT_ID(ffn_moe_down-73): 235 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 358 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 374 us MUL_MAT_ID(ffn_moe_down-75): 205 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 398 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 367 us MUL_MAT_ID(ffn_moe_down-77): 209 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 400 us MUL_MAT_ID(ffn_moe_down-78): 169 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 354 us MUL_MAT_ID(ffn_moe_down-79): 202 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 355 us MUL_MAT_ID(ffn_moe_down-80): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 339 us MUL_MAT_ID(ffn_moe_down-81): 196 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 390 us MUL_MAT_ID(ffn_moe_down-82): 200 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 402 us MUL_MAT_ID(ffn_moe_down-83): 209 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 379 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 358 us MUL_MAT_ID(ffn_moe_down-85): 290 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 385 us MUL_MAT_ID(ffn_moe_down-86): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 393 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 389 us MUL_MAT_ID(ffn_moe_down-88): 221 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 394 us MUL_MAT_ID(ffn_moe_down-89): 186 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 401 us MUL_MAT_ID(ffn_moe_down-90): 326 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 418 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 33 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 394 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 385 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 388 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 368 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 405 us MUL_MAT_ID(ffn_moe_down-29): 204 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 398 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 389 us MUL_MAT_ID(ffn_moe_down-31): 402 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 382 us MUL_MAT_ID(ffn_moe_down-32): 338 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 411 us MUL_MAT_ID(ffn_moe_down-33): 227 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 394 us MUL_MAT_ID(ffn_moe_down-34): 202 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 370 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 353 us MUL_MAT_ID(ffn_moe_down-36): 214 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 391 us MUL_MAT_ID(ffn_moe_down-37): 406 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 365 us MUL_MAT_ID(ffn_moe_down-39): 242 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 358 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 390 us MUL_MAT_ID(ffn_moe_down-41): 514 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 366 us MUL_MAT_ID(ffn_moe_down-42): 216 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 397 us MUL_MAT_ID(ffn_moe_down-43): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 360 us MUL_MAT_ID(ffn_moe_down-44): 388 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 353 us MUL_MAT_ID(ffn_moe_down-45): 498 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 368 us MUL_MAT_ID(ffn_moe_down-46): 203 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 369 us MUL_MAT_ID(ffn_moe_down-47): 207 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 381 us MUL_MAT_ID(ffn_moe_down-48): 205 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 346 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 392 us MUL_MAT_ID(ffn_moe_down-50): 221 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 386 us MUL_MAT_ID(ffn_moe_down-51): 204 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 393 us MUL_MAT_ID(ffn_moe_down-52): 212 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 355 us MUL_MAT_ID(ffn_moe_down-53): 186 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 388 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 359 us MUL_MAT_ID(ffn_moe_down-55): 193 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 366 us MUL_MAT_ID(ffn_moe_down-56): 206 us ggml_barrier(...): 401 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 367 us MUL_MAT_ID(ffn_moe_down-57): 384 us ggml_barrier(...): 346 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 400 us MUL_MAT_ID(ffn_moe_down-58): 211 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 373 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 356 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 377 us MUL_MAT_ID(ffn_moe_down-61): 260 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 380 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 388 us MUL_MAT_ID(ffn_moe_down-63): 376 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 400 us MUL_MAT_ID(ffn_moe_down-64): 335 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 350 us MUL_MAT_ID(ffn_moe_down-65): 210 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 365 us MUL_MAT_ID(ffn_moe_down-66): 182 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 376 us MUL_MAT_ID(ffn_moe_down-67): 222 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 391 us MUL_MAT_ID(ffn_moe_down-68): 334 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 373 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 398 us MUL_MAT_ID(ffn_moe_down-70): 185 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 342 us MUL_MAT_ID(ffn_moe_down-71): 186 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 369 us MUL_MAT_ID(ffn_moe_down-72): 188 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 393 us MUL_MAT_ID(ffn_moe_down-73): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 356 us MUL_MAT_ID(ffn_moe_down-74): 227 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 381 us MUL_MAT_ID(ffn_moe_down-75): 191 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 398 us MUL_MAT_ID(ffn_moe_down-76): 214 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 463 us MUL_MAT_ID(ffn_moe_down-77): 508 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 381 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 407 us MUL_MAT_ID(ffn_moe_down-79): 207 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 412 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 385 us MUL_MAT_ID(ffn_moe_down-81): 313 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 388 us MUL_MAT_ID(ffn_moe_down-82): 206 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 379 us MUL_MAT_ID(ffn_moe_down-83): 371 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 378 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 351 us MUL_MAT_ID(ffn_moe_down-85): 226 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 421 us MUL_MAT_ID(ffn_moe_down-86): 221 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 401 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 420 us MUL_MAT_ID(ffn_moe_down-88): 221 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 408 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 385 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 403 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 10 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 432 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 374 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 396 us MUL_MAT_ID(ffn_moe_down-27): 184 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 399 us MUL_MAT_ID(ffn_moe_down-28): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 370 us MUL_MAT_ID(ffn_moe_down-29): 192 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 389 us MUL_MAT_ID(ffn_moe_down-30): 238 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 365 us MUL_MAT_ID(ffn_moe_down-31): 373 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 388 us MUL_MAT_ID(ffn_moe_down-32): 339 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 415 us MUL_MAT_ID(ffn_moe_down-33): 211 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 415 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 384 us MUL_MAT_ID(ffn_moe_down-35): 205 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 400 us MUL_MAT_ID(ffn_moe_down-36): 191 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 390 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 419 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 369 us MUL_MAT_ID(ffn_moe_down-39): 598 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 413 us MUL_MAT_ID(ffn_moe_down-40): 227 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 396 us MUL_MAT_ID(ffn_moe_down-41): 216 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 406 us MUL_MAT_ID(ffn_moe_down-42): 210 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 379 us MUL_MAT_ID(ffn_moe_down-43): 197 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 399 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 373 us MUL_MAT_ID(ffn_moe_down-45): 201 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 363 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 409 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 373 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 402 us MUL_MAT_ID(ffn_moe_down-49): 196 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 383 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 360 us MUL_MAT_ID(ffn_moe_down-51): 338 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 380 us MUL_MAT_ID(ffn_moe_down-52): 208 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 370 us MUL_MAT_ID(ffn_moe_down-53): 397 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 356 us MUL_MAT_ID(ffn_moe_down-54): 250 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 395 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 373 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 382 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 354 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 341 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 387 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 373 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 373 us MUL_MAT_ID(ffn_moe_down-62): 430 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 376 us MUL_MAT_ID(ffn_moe_down-63): 262 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 377 us MUL_MAT_ID(ffn_moe_down-64): 212 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 403 us MUL_MAT_ID(ffn_moe_down-65): 193 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 384 us MUL_MAT_ID(ffn_moe_down-66): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 354 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 375 us MUL_MAT_ID(ffn_moe_down-68): 212 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 394 us MUL_MAT_ID(ffn_moe_down-69): 190 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 377 us MUL_MAT_ID(ffn_moe_down-70): 211 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 383 us MUL_MAT_ID(ffn_moe_down-71): 317 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 360 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 404 us MUL_MAT_ID(ffn_moe_down-73): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 383 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 400 us MUL_MAT_ID(ffn_moe_down-75): 211 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 405 us MUL_MAT_ID(ffn_moe_down-76): 184 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 392 us MUL_MAT_ID(ffn_moe_down-77): 265 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 406 us MUL_MAT_ID(ffn_moe_down-78): 277 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 396 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 363 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 392 us MUL_MAT_ID(ffn_moe_down-81): 400 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 370 us MUL_MAT_ID(ffn_moe_down-82): 194 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 370 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 357 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 381 us MUL_MAT_ID(ffn_moe_down-85): 227 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 367 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 402 us MUL_MAT_ID(ffn_moe_down-87): 207 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 317 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 393 us MUL_MAT_ID(ffn_moe_down-89): 394 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 392 us MUL_MAT_ID(ffn_moe_down-90): 705 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 386 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 379 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 399 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 385 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 362 us MUL_MAT_ID(ffn_moe_down-28): 212 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 378 us MUL_MAT_ID(ffn_moe_down-29): 347 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 418 us MUL_MAT_ID(ffn_moe_down-30): 312 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 395 us MUL_MAT_ID(ffn_moe_down-31): 252 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 405 us MUL_MAT_ID(ffn_moe_down-32): 203 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 453 us MUL_MAT_ID(ffn_moe_down-33): 239 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 393 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 416 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 403 us MUL_MAT_ID(ffn_moe_down-36): 208 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 414 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 393 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 391 us MUL_MAT_ID(ffn_moe_down-39): 247 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 427 us MUL_MAT_ID(ffn_moe_down-40): 378 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 413 us MUL_MAT_ID(ffn_moe_down-41): 205 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 371 us MUL_MAT_ID(ffn_moe_down-42): 315 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 388 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 356 us MUL_MAT_ID(ffn_moe_down-44): 198 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 366 us MUL_MAT_ID(ffn_moe_down-45): 193 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 367 us MUL_MAT_ID(ffn_moe_down-46): 431 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 403 us MUL_MAT_ID(ffn_moe_down-47): 208 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 398 us MUL_MAT_ID(ffn_moe_down-48): 374 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 343 us MUL_MAT_ID(ffn_moe_down-49): 212 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 371 us MUL_MAT_ID(ffn_moe_down-50): 178 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 358 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 348 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 388 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 379 us MUL_MAT_ID(ffn_moe_down-54): 210 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 414 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 373 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 483 us MUL_MAT_ID(ffn_moe_down-57): 288 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 406 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 372 us MUL_MAT_ID(ffn_moe_down-59): 189 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 386 us MUL_MAT_ID(ffn_moe_down-60): 197 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 420 us MUL_MAT_ID(ffn_moe_down-61): 384 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 476 us MUL_MAT_ID(ffn_moe_down-62): 183 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 393 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 392 us MUL_MAT_ID(ffn_moe_down-64): 196 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 395 us MUL_MAT_ID(ffn_moe_down-65): 205 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 531 us MUL_MAT_ID(ffn_moe_down-66): 167 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 361 us MUL_MAT_ID(ffn_moe_down-67): 384 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 377 us MUL_MAT_ID(ffn_moe_down-68): 444 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 358 us MUL_MAT_ID(ffn_moe_down-69): 221 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 394 us MUL_MAT_ID(ffn_moe_down-70): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 522 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 385 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 396 us MUL_MAT_ID(ffn_moe_down-73): 209 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 501 us MUL_MAT_ID(ffn_moe_down-74): 245 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 437 us MUL_MAT_ID(ffn_moe_down-75): 172 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 400 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 363 us MUL_MAT_ID(ffn_moe_down-77): 280 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 393 us MUL_MAT_ID(ffn_moe_down-78): 286 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 430 us MUL_MAT_ID(ffn_moe_down-79): 298 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 363 us MUL_MAT_ID(ffn_moe_down-80): 415 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 367 us MUL_MAT_ID(ffn_moe_down-81): 254 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 368 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 383 us MUL_MAT_ID(ffn_moe_down-83): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 353 us MUL_MAT_ID(ffn_moe_down-84): 221 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 384 us MUL_MAT_ID(ffn_moe_down-85): 693 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 392 us MUL_MAT_ID(ffn_moe_down-86): 183 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 389 us MUL_MAT_ID(ffn_moe_down-87): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 337 us MUL_MAT_ID(ffn_moe_down-88): 188 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 357 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 411 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 352 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 76 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 176 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 358 us MUL_MAT_ID(ffn_moe_down-26): 210 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 457 us MUL_MAT_ID(ffn_moe_down-27): 301 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 387 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 207 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 495 us MUL_MAT_ID(ffn_moe_down-30): 184 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 364 us MUL_MAT_ID(ffn_moe_down-31): 183 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 370 us MUL_MAT_ID(ffn_moe_down-32): 208 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 354 us MUL_MAT_ID(ffn_moe_down-33): 185 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 387 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 374 us MUL_MAT_ID(ffn_moe_down-35): 377 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 212 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 373 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 378 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 396 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 414 us MUL_MAT_ID(ffn_moe_down-40): 245 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 359 us MUL_MAT_ID(ffn_moe_down-41): 361 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 366 us MUL_MAT_ID(ffn_moe_down-42): 202 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 404 us MUL_MAT_ID(ffn_moe_down-43): 189 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 396 us MUL_MAT_ID(ffn_moe_down-44): 347 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 369 us MUL_MAT_ID(ffn_moe_down-45): 237 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 386 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 377 us MUL_MAT_ID(ffn_moe_down-47): 187 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 387 us MUL_MAT_ID(ffn_moe_down-48): 386 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 369 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 379 us MUL_MAT_ID(ffn_moe_down-50): 214 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 413 us MUL_MAT_ID(ffn_moe_down-51): 360 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 373 us MUL_MAT_ID(ffn_moe_down-52): 205 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 368 us MUL_MAT_ID(ffn_moe_down-53): 362 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 475 us MUL_MAT_ID(ffn_moe_down-54): 430 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 490 us MUL_MAT_ID(ffn_moe_down-55): 227 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 406 us MUL_MAT_ID(ffn_moe_down-56): 197 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 367 us MUL_MAT_ID(ffn_moe_down-57): 169 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 402 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 372 us MUL_MAT_ID(ffn_moe_down-59): 188 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 201 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 475 us MUL_MAT_ID(ffn_moe_down-61): 246 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 373 us MUL_MAT_ID(ffn_moe_down-62): 205 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 404 us MUL_MAT_ID(ffn_moe_down-63): 214 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 400 us MUL_MAT_ID(ffn_moe_down-64): 183 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 376 us MUL_MAT_ID(ffn_moe_down-65): 213 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 395 us MUL_MAT_ID(ffn_moe_down-66): 181 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 403 us MUL_MAT_ID(ffn_moe_down-67): 203 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 380 us MUL_MAT_ID(ffn_moe_down-68): 282 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 429 us MUL_MAT_ID(ffn_moe_down-69): 212 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 360 us MUL_MAT_ID(ffn_moe_down-70): 223 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 385 us MUL_MAT_ID(ffn_moe_down-71): 191 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 372 us MUL_MAT_ID(ffn_moe_down-72): 187 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 402 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 340 us MUL_MAT_ID(ffn_moe_down-74): 233 us ggml_barrier(...): 335 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 358 us MUL_MAT_ID(ffn_moe_down-75): 344 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 400 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 408 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 374 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 392 us MUL_MAT_ID(ffn_moe_down-79): 394 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 529 us MUL_MAT_ID(ffn_moe_down-80): 208 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 386 us MUL_MAT_ID(ffn_moe_down-81): 196 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 354 us MUL_MAT_ID(ffn_moe_down-82): 184 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 401 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 470 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 398 us MUL_MAT_ID(ffn_moe_down-85): 252 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 428 us MUL_MAT_ID(ffn_moe_down-86): 189 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 410 us MUL_MAT_ID(ffn_moe_down-88): 323 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 385 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 508 us MUL_MAT_ID(ffn_moe_down-90): 734 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 393 us MUL_MAT_ID(ffn_moe_down-91): 233 us ggml_barrier(...): 46 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 408 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 353 us MUL_MAT_ID(ffn_moe_down-26): 203 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 381 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 398 us MUL_MAT_ID(ffn_moe_down-28): 184 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 414 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 403 us MUL_MAT_ID(ffn_moe_down-30): 188 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 356 us MUL_MAT_ID(ffn_moe_down-31): 478 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 370 us MUL_MAT_ID(ffn_moe_down-32): 292 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 393 us MUL_MAT_ID(ffn_moe_down-33): 220 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 392 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 382 us MUL_MAT_ID(ffn_moe_down-35): 185 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 347 us MUL_MAT_ID(ffn_moe_down-36): 181 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 451 us MUL_MAT_ID(ffn_moe_down-37): 252 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 396 us MUL_MAT_ID(ffn_moe_down-38): 187 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 423 us MUL_MAT_ID(ffn_moe_down-39): 638 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 367 us MUL_MAT_ID(ffn_moe_down-40): 248 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 388 us MUL_MAT_ID(ffn_moe_down-41): 185 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 399 us MUL_MAT_ID(ffn_moe_down-42): 185 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 387 us MUL_MAT_ID(ffn_moe_down-43): 181 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 404 us MUL_MAT_ID(ffn_moe_down-44): 194 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 362 us MUL_MAT_ID(ffn_moe_down-45): 194 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 362 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 394 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 406 us MUL_MAT_ID(ffn_moe_down-48): 209 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 379 us MUL_MAT_ID(ffn_moe_down-49): 228 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 358 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 362 us MUL_MAT_ID(ffn_moe_down-51): 221 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 415 us MUL_MAT_ID(ffn_moe_down-52): 186 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 357 us MUL_MAT_ID(ffn_moe_down-53): 187 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 378 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 395 us MUL_MAT_ID(ffn_moe_down-55): 204 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 391 us MUL_MAT_ID(ffn_moe_down-56): 191 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 391 us MUL_MAT_ID(ffn_moe_down-57): 214 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 413 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 351 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 356 us MUL_MAT_ID(ffn_moe_down-60): 212 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 362 us MUL_MAT_ID(ffn_moe_down-61): 181 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 350 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 400 us MUL_MAT_ID(ffn_moe_down-63): 205 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 400 us MUL_MAT_ID(ffn_moe_down-64): 514 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 345 us MUL_MAT_ID(ffn_moe_down-65): 187 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 344 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 400 us MUL_MAT_ID(ffn_moe_down-67): 195 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 209 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 369 us MUL_MAT_ID(ffn_moe_down-69): 259 us ggml_barrier(...): 438 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 362 us MUL_MAT_ID(ffn_moe_down-70): 211 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 387 us MUL_MAT_ID(ffn_moe_down-71): 212 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 374 us MUL_MAT_ID(ffn_moe_down-72): 352 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 392 us MUL_MAT_ID(ffn_moe_down-73): 210 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 410 us MUL_MAT_ID(ffn_moe_down-74): 236 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 370 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 405 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 353 us MUL_MAT_ID(ffn_moe_down-77): 204 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 370 us MUL_MAT_ID(ffn_moe_down-78): 356 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 372 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 356 us MUL_MAT_ID(ffn_moe_down-80): 246 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 357 us MUL_MAT_ID(ffn_moe_down-81): 214 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 380 us MUL_MAT_ID(ffn_moe_down-82): 208 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 372 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 387 us MUL_MAT_ID(ffn_moe_down-84): 561 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 402 us MUL_MAT_ID(ffn_moe_down-85): 225 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 396 us MUL_MAT_ID(ffn_moe_down-86): 230 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 412 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 367 us MUL_MAT_ID(ffn_moe_down-88): 221 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 410 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 349 us MUL_MAT_ID(ffn_moe_down-90): 224 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 379 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 54 us GET_ROWS(inp_embd): 26 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 414 us MUL_MAT_ID(ffn_moe_down-25): 459 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 386 us MUL_MAT_ID(ffn_moe_down-26): 301 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 392 us MUL_MAT_ID(ffn_moe_down-27): 222 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 367 us MUL_MAT_ID(ffn_moe_down-28): 337 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 373 us MUL_MAT_ID(ffn_moe_down-29): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 373 us MUL_MAT_ID(ffn_moe_down-30): 218 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 355 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 397 us MUL_MAT_ID(ffn_moe_down-32): 216 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 374 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 381 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 364 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 398 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 383 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 395 us MUL_MAT_ID(ffn_moe_down-38): 516 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 353 us MUL_MAT_ID(ffn_moe_down-39): 247 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 366 us MUL_MAT_ID(ffn_moe_down-40): 229 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 402 us MUL_MAT_ID(ffn_moe_down-41): 194 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 364 us MUL_MAT_ID(ffn_moe_down-42): 204 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 387 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 374 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 380 us MUL_MAT_ID(ffn_moe_down-45): 380 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 401 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 404 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 339 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 381 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 479 us MUL_MAT_ID(ffn_moe_down-50): 174 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 342 us MUL_MAT_ID(ffn_moe_down-51): 419 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 375 us MUL_MAT_ID(ffn_moe_down-52): 210 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 340 us MUL_MAT_ID(ffn_moe_down-53): 199 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 387 us MUL_MAT_ID(ffn_moe_down-54): 333 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 389 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 398 us MUL_MAT_ID(ffn_moe_down-56): 292 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 407 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 400 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 372 us MUL_MAT_ID(ffn_moe_down-59): 188 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 408 us MUL_MAT_ID(ffn_moe_down-60): 209 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 361 us MUL_MAT_ID(ffn_moe_down-61): 208 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 353 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 351 us MUL_MAT_ID(ffn_moe_down-63): 204 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 401 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 350 us MUL_MAT_ID(ffn_moe_down-65): 190 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 363 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 357 us MUL_MAT_ID(ffn_moe_down-67): 359 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 379 us MUL_MAT_ID(ffn_moe_down-68): 204 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 363 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 382 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 370 us MUL_MAT_ID(ffn_moe_down-71): 240 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 402 us MUL_MAT_ID(ffn_moe_down-72): 203 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 363 us MUL_MAT_ID(ffn_moe_down-73): 290 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 358 us MUL_MAT_ID(ffn_moe_down-74): 356 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 409 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 401 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 372 us MUL_MAT_ID(ffn_moe_down-77): 227 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 409 us MUL_MAT_ID(ffn_moe_down-78): 275 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 400 us MUL_MAT_ID(ffn_moe_down-79): 188 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 378 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 408 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 401 us MUL_MAT_ID(ffn_moe_down-82): 457 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 360 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 350 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 374 us MUL_MAT_ID(ffn_moe_down-85): 447 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 375 us MUL_MAT_ID(ffn_moe_down-86): 218 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 374 us MUL_MAT_ID(ffn_moe_down-87): 188 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 385 us MUL_MAT_ID(ffn_moe_down-88): 439 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 396 us MUL_MAT_ID(ffn_moe_down-89): 377 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 417 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 371 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 34 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 379 us MUL_MAT_ID(ffn_moe_down-25): 293 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 394 us MUL_MAT_ID(ffn_moe_down-26): 428 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 389 us MUL_MAT_ID(ffn_moe_down-27): 180 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 361 us MUL_MAT_ID(ffn_moe_down-28): 186 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 395 us MUL_MAT_ID(ffn_moe_down-29): 188 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 400 us MUL_MAT_ID(ffn_moe_down-30): 363 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 440 us MUL_MAT_ID(ffn_moe_down-31): 181 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 380 us MUL_MAT_ID(ffn_moe_down-32): 276 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 384 us MUL_MAT_ID(ffn_moe_down-33): 224 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 392 us MUL_MAT_ID(ffn_moe_down-34): 469 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 383 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 394 us MUL_MAT_ID(ffn_moe_down-36): 433 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 368 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 380 us MUL_MAT_ID(ffn_moe_down-38): 176 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 377 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 411 us MUL_MAT_ID(ffn_moe_down-40): 242 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 440 us MUL_MAT_ID(ffn_moe_down-41): 200 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 390 us MUL_MAT_ID(ffn_moe_down-42): 343 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 373 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 385 us MUL_MAT_ID(ffn_moe_down-44): 288 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 366 us MUL_MAT_ID(ffn_moe_down-45): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 382 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 402 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 392 us MUL_MAT_ID(ffn_moe_down-48): 524 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 391 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 410 us MUL_MAT_ID(ffn_moe_down-50): 265 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 393 us MUL_MAT_ID(ffn_moe_down-51): 348 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 408 us MUL_MAT_ID(ffn_moe_down-52): 220 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 410 us MUL_MAT_ID(ffn_moe_down-53): 302 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 358 us MUL_MAT_ID(ffn_moe_down-54): 217 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 443 us MUL_MAT_ID(ffn_moe_down-55): 265 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 393 us MUL_MAT_ID(ffn_moe_down-56): 278 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 391 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 428 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 379 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 419 us MUL_MAT_ID(ffn_moe_down-60): 236 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 374 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 388 us MUL_MAT_ID(ffn_moe_down-62): 207 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 379 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 428 us MUL_MAT_ID(ffn_moe_down-64): 263 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 394 us MUL_MAT_ID(ffn_moe_down-65): 203 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 408 us MUL_MAT_ID(ffn_moe_down-66): 184 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 554 us MUL_MAT_ID(ffn_moe_down-67): 439 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 375 us MUL_MAT_ID(ffn_moe_down-68): 202 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 389 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 360 us MUL_MAT_ID(ffn_moe_down-70): 396 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 406 us MUL_MAT_ID(ffn_moe_down-71): 224 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 364 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 370 us MUL_MAT_ID(ffn_moe_down-73): 361 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 385 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 345 us MUL_MAT_ID(ffn_moe_down-75): 208 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 369 us MUL_MAT_ID(ffn_moe_down-76): 210 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 369 us MUL_MAT_ID(ffn_moe_down-77): 244 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 375 us MUL_MAT_ID(ffn_moe_down-78): 189 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 374 us MUL_MAT_ID(ffn_moe_down-79): 216 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 503 us MUL_MAT_ID(ffn_moe_down-80): 167 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 375 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 368 us MUL_MAT_ID(ffn_moe_down-82): 246 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 506 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 399 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 387 us MUL_MAT_ID(ffn_moe_down-85): 710 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 383 us MUL_MAT_ID(ffn_moe_down-86): 367 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 365 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 500 us MUL_MAT_ID(ffn_moe_down-88): 163 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 390 us MUL_MAT_ID(ffn_moe_down-89): 311 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 358 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 365 us MUL_MAT_ID(ffn_moe_down-91): 246 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 376 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 374 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 419 us MUL_MAT_ID(ffn_moe_down-27): 223 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 424 us MUL_MAT_ID(ffn_moe_down-28): 187 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 396 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 375 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 402 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 391 us MUL_MAT_ID(ffn_moe_down-32): 348 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 395 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 387 us MUL_MAT_ID(ffn_moe_down-34): 520 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 387 us MUL_MAT_ID(ffn_moe_down-35): 183 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 377 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 407 us MUL_MAT_ID(ffn_moe_down-37): 241 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 408 us MUL_MAT_ID(ffn_moe_down-38): 208 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 406 us MUL_MAT_ID(ffn_moe_down-39): 310 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 399 us MUL_MAT_ID(ffn_moe_down-40): 234 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 408 us MUL_MAT_ID(ffn_moe_down-41): 200 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 373 us MUL_MAT_ID(ffn_moe_down-42): 375 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 383 us MUL_MAT_ID(ffn_moe_down-43): 206 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 379 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 383 us MUL_MAT_ID(ffn_moe_down-45): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 360 us MUL_MAT_ID(ffn_moe_down-46): 184 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 418 us MUL_MAT_ID(ffn_moe_down-47): 242 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 467 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 405 us MUL_MAT_ID(ffn_moe_down-49): 214 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 394 us MUL_MAT_ID(ffn_moe_down-50): 206 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 454 us MUL_MAT_ID(ffn_moe_down-51): 465 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 385 us MUL_MAT_ID(ffn_moe_down-52): 185 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 357 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 366 us MUL_MAT_ID(ffn_moe_down-54): 209 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 358 us MUL_MAT_ID(ffn_moe_down-55): 214 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 393 us MUL_MAT_ID(ffn_moe_down-56): 261 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 400 us MUL_MAT_ID(ffn_moe_down-57): 211 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 362 us MUL_MAT_ID(ffn_moe_down-58): 354 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 373 us MUL_MAT_ID(ffn_moe_down-59): 206 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 375 us MUL_MAT_ID(ffn_moe_down-60): 294 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 354 us MUL_MAT_ID(ffn_moe_down-61): 172 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 360 us MUL_MAT_ID(ffn_moe_down-62): 313 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 355 us MUL_MAT_ID(ffn_moe_down-63): 221 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 393 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 326 us MUL_MAT_ID(ffn_moe_down-65): 394 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 373 us MUL_MAT_ID(ffn_moe_down-66): 170 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 368 us MUL_MAT_ID(ffn_moe_down-67): 215 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 401 us MUL_MAT_ID(ffn_moe_down-68): 191 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 390 us MUL_MAT_ID(ffn_moe_down-69): 221 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 366 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 365 us MUL_MAT_ID(ffn_moe_down-71): 191 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 393 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 384 us MUL_MAT_ID(ffn_moe_down-73): 206 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 337 us MUL_MAT_ID(ffn_moe_down-74): 214 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 370 us MUL_MAT_ID(ffn_moe_down-75): 199 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 364 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 375 us MUL_MAT_ID(ffn_moe_down-77): 187 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 401 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 378 us MUL_MAT_ID(ffn_moe_down-79): 230 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 375 us MUL_MAT_ID(ffn_moe_down-80): 187 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 371 us MUL_MAT_ID(ffn_moe_down-81): 365 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 214 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 388 us MUL_MAT_ID(ffn_moe_down-83): 206 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 416 us MUL_MAT_ID(ffn_moe_down-84): 212 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 408 us MUL_MAT_ID(ffn_moe_down-85): 258 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 466 us MUL_MAT_ID(ffn_moe_down-86): 219 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 506 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 372 us MUL_MAT_ID(ffn_moe_down-88): 367 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 400 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 428 us MUL_MAT_ID(ffn_moe_down-90): 288 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 390 us MUL_MAT_ID(ffn_moe_down-91): 259 us ggml_barrier(...): 42 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 422 us MUL_MAT_ID(ffn_moe_down-25): 334 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 388 us MUL_MAT_ID(ffn_moe_down-26): 202 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 383 us MUL_MAT_ID(ffn_moe_down-27): 212 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 422 us MUL_MAT_ID(ffn_moe_down-28): 206 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 363 us MUL_MAT_ID(ffn_moe_down-29): 276 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 470 us MUL_MAT_ID(ffn_moe_down-30): 493 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 373 us MUL_MAT_ID(ffn_moe_down-31): 374 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 366 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 401 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 391 us MUL_MAT_ID(ffn_moe_down-34): 258 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 381 us MUL_MAT_ID(ffn_moe_down-35): 217 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 420 us MUL_MAT_ID(ffn_moe_down-36): 282 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 377 us MUL_MAT_ID(ffn_moe_down-37): 628 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 366 us MUL_MAT_ID(ffn_moe_down-38): 202 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 460 us MUL_MAT_ID(ffn_moe_down-39): 275 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 378 us MUL_MAT_ID(ffn_moe_down-40): 238 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 362 us MUL_MAT_ID(ffn_moe_down-41): 520 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 379 us MUL_MAT_ID(ffn_moe_down-42): 183 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 408 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 401 us MUL_MAT_ID(ffn_moe_down-44): 185 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 383 us MUL_MAT_ID(ffn_moe_down-45): 215 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 390 us MUL_MAT_ID(ffn_moe_down-46): 242 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 436 us MUL_MAT_ID(ffn_moe_down-47): 161 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 377 us MUL_MAT_ID(ffn_moe_down-48): 185 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 384 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 356 us MUL_MAT_ID(ffn_moe_down-50): 188 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 471 us MUL_MAT_ID(ffn_moe_down-51): 168 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 373 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 398 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 347 us MUL_MAT_ID(ffn_moe_down-54): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 395 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 491 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 352 us MUL_MAT_ID(ffn_moe_down-57): 535 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 383 us MUL_MAT_ID(ffn_moe_down-58): 224 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 351 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 524 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 372 us MUL_MAT_ID(ffn_moe_down-61): 220 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 390 us MUL_MAT_ID(ffn_moe_down-62): 269 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 408 us MUL_MAT_ID(ffn_moe_down-63): 235 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 381 us MUL_MAT_ID(ffn_moe_down-64): 208 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 395 us MUL_MAT_ID(ffn_moe_down-65): 368 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 367 us ggml_barrier(...): 343 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 400 us MUL_MAT_ID(ffn_moe_down-67): 214 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 363 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 391 us MUL_MAT_ID(ffn_moe_down-69): 173 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 383 us MUL_MAT_ID(ffn_moe_down-70): 193 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 359 us MUL_MAT_ID(ffn_moe_down-71): 370 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 363 us MUL_MAT_ID(ffn_moe_down-72): 210 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 345 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 395 us MUL_MAT_ID(ffn_moe_down-74): 228 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 356 us MUL_MAT_ID(ffn_moe_down-75): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 396 us MUL_MAT_ID(ffn_moe_down-76): 334 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 398 us MUL_MAT_ID(ffn_moe_down-77): 304 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 352 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 399 us MUL_MAT_ID(ffn_moe_down-79): 185 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 356 us MUL_MAT_ID(ffn_moe_down-80): 194 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 397 us MUL_MAT_ID(ffn_moe_down-81): 175 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 395 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 311 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 396 us MUL_MAT_ID(ffn_moe_down-83): 184 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 488 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 402 us MUL_MAT_ID(ffn_moe_down-85): 228 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 402 us MUL_MAT_ID(ffn_moe_down-86): 246 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 403 us MUL_MAT_ID(ffn_moe_down-87): 183 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 398 us MUL_MAT_ID(ffn_moe_down-89): 246 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 383 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 408 us MUL_MAT_ID(ffn_moe_down-91): 247 us ggml_barrier(...): 26 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 394 us MUL_MAT_ID(ffn_moe_down-25): 185 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 398 us MUL_MAT_ID(ffn_moe_down-26): 273 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 401 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 390 us MUL_MAT_ID(ffn_moe_down-28): 190 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 365 us MUL_MAT_ID(ffn_moe_down-29): 185 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 344 us MUL_MAT_ID(ffn_moe_down-30): 217 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 434 us MUL_MAT_ID(ffn_moe_down-31): 216 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 396 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 367 us MUL_MAT_ID(ffn_moe_down-33): 224 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 375 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 382 us MUL_MAT_ID(ffn_moe_down-35): 216 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 418 us MUL_MAT_ID(ffn_moe_down-36): 187 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 399 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 202 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 393 us MUL_MAT_ID(ffn_moe_down-39): 312 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 420 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 420 us MUL_MAT_ID(ffn_moe_down-41): 190 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 406 us MUL_MAT_ID(ffn_moe_down-42): 199 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 409 us MUL_MAT_ID(ffn_moe_down-43): 214 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 387 us MUL_MAT_ID(ffn_moe_down-44): 184 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 384 us MUL_MAT_ID(ffn_moe_down-45): 506 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 395 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 419 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 408 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 414 us MUL_MAT_ID(ffn_moe_down-49): 231 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 378 us MUL_MAT_ID(ffn_moe_down-50): 209 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 379 us MUL_MAT_ID(ffn_moe_down-51): 354 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 386 us MUL_MAT_ID(ffn_moe_down-52): 201 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 372 us MUL_MAT_ID(ffn_moe_down-53): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 389 us MUL_MAT_ID(ffn_moe_down-54): 197 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 364 us MUL_MAT_ID(ffn_moe_down-55): 208 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 197 us ggml_barrier(...): 336 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 382 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 358 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 388 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 394 us MUL_MAT_ID(ffn_moe_down-60): 430 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 393 us MUL_MAT_ID(ffn_moe_down-61): 177 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 375 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 484 us MUL_MAT_ID(ffn_moe_down-63): 481 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 419 us MUL_MAT_ID(ffn_moe_down-64): 365 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 414 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 392 us MUL_MAT_ID(ffn_moe_down-66): 334 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 382 us MUL_MAT_ID(ffn_moe_down-67): 191 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 393 us MUL_MAT_ID(ffn_moe_down-68): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 494 us MUL_MAT_ID(ffn_moe_down-69): 184 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 391 us MUL_MAT_ID(ffn_moe_down-70): 434 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 349 us MUL_MAT_ID(ffn_moe_down-71): 202 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 407 us MUL_MAT_ID(ffn_moe_down-72): 378 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 369 us MUL_MAT_ID(ffn_moe_down-73): 384 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 384 us MUL_MAT_ID(ffn_moe_down-74): 191 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 347 us MUL_MAT_ID(ffn_moe_down-75): 200 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 393 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 370 us MUL_MAT_ID(ffn_moe_down-77): 211 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 368 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 392 us MUL_MAT_ID(ffn_moe_down-79): 210 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 363 us MUL_MAT_ID(ffn_moe_down-80): 206 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 386 us MUL_MAT_ID(ffn_moe_down-81): 205 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 382 us MUL_MAT_ID(ffn_moe_down-82): 395 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 355 us MUL_MAT_ID(ffn_moe_down-83): 215 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 356 us MUL_MAT_ID(ffn_moe_down-84): 245 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 348 us MUL_MAT_ID(ffn_moe_down-85): 227 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 374 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 388 us MUL_MAT_ID(ffn_moe_down-87): 214 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 387 us MUL_MAT_ID(ffn_moe_down-88): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 384 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 388 us MUL_MAT_ID(ffn_moe_down-90): 240 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 376 us MUL_MAT_ID(ffn_moe_down-91): 255 us ggml_barrier(...): 37 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 416 us MUL_MAT_ID(ffn_moe_down-25): 203 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 397 us MUL_MAT_ID(ffn_moe_down-26): 400 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 395 us MUL_MAT_ID(ffn_moe_down-27): 206 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 383 us MUL_MAT_ID(ffn_moe_down-28): 257 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 370 us MUL_MAT_ID(ffn_moe_down-29): 218 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 381 us MUL_MAT_ID(ffn_moe_down-30): 207 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 363 us MUL_MAT_ID(ffn_moe_down-31): 181 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 371 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 359 us MUL_MAT_ID(ffn_moe_down-33): 202 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 398 us MUL_MAT_ID(ffn_moe_down-34): 206 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 495 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 453 us MUL_MAT_ID(ffn_moe_down-36): 170 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 509 us MUL_MAT_ID(ffn_moe_down-37): 266 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 373 us MUL_MAT_ID(ffn_moe_down-38): 209 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 383 us MUL_MAT_ID(ffn_moe_down-39): 241 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 379 us MUL_MAT_ID(ffn_moe_down-40): 234 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 402 us MUL_MAT_ID(ffn_moe_down-41): 205 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 393 us MUL_MAT_ID(ffn_moe_down-42): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 388 us MUL_MAT_ID(ffn_moe_down-43): 363 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 396 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 351 us MUL_MAT_ID(ffn_moe_down-45): 204 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 396 us MUL_MAT_ID(ffn_moe_down-46): 220 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 394 us MUL_MAT_ID(ffn_moe_down-47): 236 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 404 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 396 us MUL_MAT_ID(ffn_moe_down-49): 197 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 391 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 374 us MUL_MAT_ID(ffn_moe_down-52): 202 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 389 us MUL_MAT_ID(ffn_moe_down-53): 205 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 395 us MUL_MAT_ID(ffn_moe_down-54): 233 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 204 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 373 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 360 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 388 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 381 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 374 us MUL_MAT_ID(ffn_moe_down-60): 202 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 377 us MUL_MAT_ID(ffn_moe_down-61): 188 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 364 us MUL_MAT_ID(ffn_moe_down-62): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 367 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 370 us MUL_MAT_ID(ffn_moe_down-64): 206 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 402 us MUL_MAT_ID(ffn_moe_down-65): 210 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 415 us MUL_MAT_ID(ffn_moe_down-66): 259 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 380 us MUL_MAT_ID(ffn_moe_down-67): 212 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 457 us MUL_MAT_ID(ffn_moe_down-68): 190 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 366 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 491 us MUL_MAT_ID(ffn_moe_down-70): 176 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 375 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 357 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 393 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 364 us MUL_MAT_ID(ffn_moe_down-74): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 375 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 354 us MUL_MAT_ID(ffn_moe_down-76): 188 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 384 us MUL_MAT_ID(ffn_moe_down-77): 203 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 564 us MUL_MAT_ID(ffn_moe_down-78): 209 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 377 us MUL_MAT_ID(ffn_moe_down-79): 197 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 371 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 422 us MUL_MAT_ID(ffn_moe_down-81): 188 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 365 us MUL_MAT_ID(ffn_moe_down-82): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 409 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 468 us MUL_MAT_ID(ffn_moe_down-84): 382 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 391 us MUL_MAT_ID(ffn_moe_down-85): 642 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 358 us MUL_MAT_ID(ffn_moe_down-86): 173 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 365 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 379 us MUL_MAT_ID(ffn_moe_down-88): 271 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 398 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 368 us MUL_MAT_ID(ffn_moe_down-90): 226 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 388 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 33 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 396 us MUL_MAT_ID(ffn_moe_down-26): 205 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 389 us MUL_MAT_ID(ffn_moe_down-27): 220 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 352 us MUL_MAT_ID(ffn_moe_down-28): 202 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 371 us MUL_MAT_ID(ffn_moe_down-29): 187 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 401 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 360 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 398 us MUL_MAT_ID(ffn_moe_down-32): 222 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 380 us MUL_MAT_ID(ffn_moe_down-33): 214 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 396 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 388 us MUL_MAT_ID(ffn_moe_down-35): 227 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 368 us MUL_MAT_ID(ffn_moe_down-36): 251 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 421 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 402 us MUL_MAT_ID(ffn_moe_down-38): 295 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 343 us MUL_MAT_ID(ffn_moe_down-39): 396 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 394 us MUL_MAT_ID(ffn_moe_down-40): 610 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 363 us MUL_MAT_ID(ffn_moe_down-41): 210 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 378 us MUL_MAT_ID(ffn_moe_down-42): 205 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 376 us MUL_MAT_ID(ffn_moe_down-43): 197 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 460 us MUL_MAT_ID(ffn_moe_down-44): 450 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 353 us MUL_MAT_ID(ffn_moe_down-45): 446 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 357 us MUL_MAT_ID(ffn_moe_down-46): 315 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 351 us MUL_MAT_ID(ffn_moe_down-47): 243 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 345 us MUL_MAT_ID(ffn_moe_down-48): 194 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 403 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 372 us MUL_MAT_ID(ffn_moe_down-50): 204 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 384 us MUL_MAT_ID(ffn_moe_down-51): 322 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 352 us MUL_MAT_ID(ffn_moe_down-52): 325 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 406 us MUL_MAT_ID(ffn_moe_down-53): 346 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 417 us MUL_MAT_ID(ffn_moe_down-54): 190 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 402 us MUL_MAT_ID(ffn_moe_down-55): 259 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 353 us MUL_MAT_ID(ffn_moe_down-56): 204 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 405 us MUL_MAT_ID(ffn_moe_down-57): 182 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 387 us MUL_MAT_ID(ffn_moe_down-58): 183 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 403 us MUL_MAT_ID(ffn_moe_down-59): 187 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 393 us MUL_MAT_ID(ffn_moe_down-60): 532 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 357 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 205 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 379 us MUL_MAT_ID(ffn_moe_down-63): 294 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 403 us MUL_MAT_ID(ffn_moe_down-64): 350 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 359 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 408 us MUL_MAT_ID(ffn_moe_down-66): 201 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 398 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 357 us MUL_MAT_ID(ffn_moe_down-68): 196 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 425 us MUL_MAT_ID(ffn_moe_down-69): 285 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 409 us MUL_MAT_ID(ffn_moe_down-70): 512 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 384 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 408 us MUL_MAT_ID(ffn_moe_down-72): 338 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 487 us MUL_MAT_ID(ffn_moe_down-73): 223 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 367 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 396 us MUL_MAT_ID(ffn_moe_down-75): 204 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 374 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 367 us MUL_MAT_ID(ffn_moe_down-77): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 367 us MUL_MAT_ID(ffn_moe_down-78): 528 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 369 us MUL_MAT_ID(ffn_moe_down-79): 187 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 398 us MUL_MAT_ID(ffn_moe_down-80): 202 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 437 us MUL_MAT_ID(ffn_moe_down-81): 272 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 390 us MUL_MAT_ID(ffn_moe_down-82): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 441 us MUL_MAT_ID(ffn_moe_down-83): 191 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 373 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 478 us MUL_MAT_ID(ffn_moe_down-85): 206 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 411 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 362 us MUL_MAT_ID(ffn_moe_down-87): 200 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 374 us MUL_MAT_ID(ffn_moe_down-88): 188 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 456 us MUL_MAT_ID(ffn_moe_down-89): 164 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 369 us MUL_MAT_ID(ffn_moe_down-90): 542 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 425 us MUL_MAT_ID(ffn_moe_down-91): 225 us ggml_barrier(...): 31 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 428 us MUL_MAT_ID(ffn_moe_down-25): 429 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 384 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 369 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 402 us MUL_MAT_ID(ffn_moe_down-28): 392 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 395 us MUL_MAT_ID(ffn_moe_down-29): 186 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 388 us MUL_MAT_ID(ffn_moe_down-30): 223 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 385 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 415 us MUL_MAT_ID(ffn_moe_down-32): 181 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 394 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 413 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 407 us MUL_MAT_ID(ffn_moe_down-36): 202 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 403 us MUL_MAT_ID(ffn_moe_down-37): 239 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 377 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 381 us MUL_MAT_ID(ffn_moe_down-39): 269 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 412 us MUL_MAT_ID(ffn_moe_down-40): 228 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 399 us MUL_MAT_ID(ffn_moe_down-41): 187 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 357 us MUL_MAT_ID(ffn_moe_down-42): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 405 us MUL_MAT_ID(ffn_moe_down-43): 203 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 392 us MUL_MAT_ID(ffn_moe_down-44): 216 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 368 us MUL_MAT_ID(ffn_moe_down-45): 222 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 348 us MUL_MAT_ID(ffn_moe_down-46): 203 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 365 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 398 us MUL_MAT_ID(ffn_moe_down-48): 221 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 369 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 395 us MUL_MAT_ID(ffn_moe_down-50): 211 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 388 us MUL_MAT_ID(ffn_moe_down-51): 187 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 520 us MUL_MAT_ID(ffn_moe_down-52): 186 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 365 us MUL_MAT_ID(ffn_moe_down-53): 195 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 358 us MUL_MAT_ID(ffn_moe_down-54): 251 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 358 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 387 us MUL_MAT_ID(ffn_moe_down-57): 234 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 379 us MUL_MAT_ID(ffn_moe_down-58): 206 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 364 us MUL_MAT_ID(ffn_moe_down-59): 176 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 473 us MUL_MAT_ID(ffn_moe_down-60): 207 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 367 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 382 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 356 us MUL_MAT_ID(ffn_moe_down-63): 282 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 356 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 458 us MUL_MAT_ID(ffn_moe_down-65): 158 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 388 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 367 us MUL_MAT_ID(ffn_moe_down-67): 214 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 370 us MUL_MAT_ID(ffn_moe_down-68): 208 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 393 us MUL_MAT_ID(ffn_moe_down-69): 188 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 390 us MUL_MAT_ID(ffn_moe_down-70): 209 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 370 us MUL_MAT_ID(ffn_moe_down-71): 205 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 352 us MUL_MAT_ID(ffn_moe_down-72): 199 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 378 us MUL_MAT_ID(ffn_moe_down-73): 242 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 361 us MUL_MAT_ID(ffn_moe_down-74): 221 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 356 us MUL_MAT_ID(ffn_moe_down-75): 188 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 353 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 382 us MUL_MAT_ID(ffn_moe_down-77): 219 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 369 us MUL_MAT_ID(ffn_moe_down-78): 248 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 356 us MUL_MAT_ID(ffn_moe_down-79): 190 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 353 us MUL_MAT_ID(ffn_moe_down-80): 191 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 407 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 439 us MUL_MAT_ID(ffn_moe_down-82): 210 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 400 us MUL_MAT_ID(ffn_moe_down-83): 218 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 412 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 413 us MUL_MAT_ID(ffn_moe_down-85): 224 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 418 us MUL_MAT_ID(ffn_moe_down-86): 157 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 376 us MUL_MAT_ID(ffn_moe_down-87): 225 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 402 us MUL_MAT_ID(ffn_moe_down-88): 178 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 395 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 382 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 383 us MUL_MAT_ID(ffn_moe_down-91): 275 us ggml_barrier(...): 27 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 381 us MUL_MAT_ID(ffn_moe_down-25): 185 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 380 us MUL_MAT_ID(ffn_moe_down-26): 231 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 378 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 410 us MUL_MAT_ID(ffn_moe_down-28): 186 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 379 us MUL_MAT_ID(ffn_moe_down-29): 184 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 384 us MUL_MAT_ID(ffn_moe_down-30): 217 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 382 us MUL_MAT_ID(ffn_moe_down-31): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 410 us MUL_MAT_ID(ffn_moe_down-32): 185 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 425 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 383 us MUL_MAT_ID(ffn_moe_down-34): 453 us ggml_barrier(...): 341 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 396 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 387 us MUL_MAT_ID(ffn_moe_down-36): 189 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 395 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 391 us MUL_MAT_ID(ffn_moe_down-38): 199 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 382 us MUL_MAT_ID(ffn_moe_down-39): 243 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 367 us MUL_MAT_ID(ffn_moe_down-40): 624 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 366 us MUL_MAT_ID(ffn_moe_down-41): 212 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 361 us MUL_MAT_ID(ffn_moe_down-42): 190 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 386 us MUL_MAT_ID(ffn_moe_down-43): 192 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 347 us MUL_MAT_ID(ffn_moe_down-44): 202 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 363 us MUL_MAT_ID(ffn_moe_down-45): 216 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 400 us MUL_MAT_ID(ffn_moe_down-46): 185 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 410 us MUL_MAT_ID(ffn_moe_down-47): 186 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 383 us MUL_MAT_ID(ffn_moe_down-48): 197 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 390 us MUL_MAT_ID(ffn_moe_down-49): 222 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 369 us MUL_MAT_ID(ffn_moe_down-50): 219 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 402 us MUL_MAT_ID(ffn_moe_down-51): 532 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 470 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 411 us MUL_MAT_ID(ffn_moe_down-53): 368 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 369 us MUL_MAT_ID(ffn_moe_down-54): 359 us ggml_barrier(...): 336 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 377 us MUL_MAT_ID(ffn_moe_down-55): 189 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 378 us MUL_MAT_ID(ffn_moe_down-56): 377 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 427 us MUL_MAT_ID(ffn_moe_down-57): 273 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 399 us MUL_MAT_ID(ffn_moe_down-58): 188 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 396 us MUL_MAT_ID(ffn_moe_down-59): 208 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 382 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 390 us MUL_MAT_ID(ffn_moe_down-61): 222 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 354 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 393 us MUL_MAT_ID(ffn_moe_down-63): 298 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 362 us MUL_MAT_ID(ffn_moe_down-64): 215 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 356 us MUL_MAT_ID(ffn_moe_down-65): 308 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 362 us MUL_MAT_ID(ffn_moe_down-66): 258 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 418 us MUL_MAT_ID(ffn_moe_down-67): 195 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 400 us MUL_MAT_ID(ffn_moe_down-68): 261 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 377 us MUL_MAT_ID(ffn_moe_down-69): 213 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 396 us MUL_MAT_ID(ffn_moe_down-70): 198 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 442 us MUL_MAT_ID(ffn_moe_down-71): 314 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 393 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 373 us MUL_MAT_ID(ffn_moe_down-74): 194 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 503 us MUL_MAT_ID(ffn_moe_down-75): 272 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 491 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 411 us MUL_MAT_ID(ffn_moe_down-77): 212 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 350 us MUL_MAT_ID(ffn_moe_down-78): 162 us ggml_barrier(...): 339 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 357 us MUL_MAT_ID(ffn_moe_down-79): 209 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 394 us MUL_MAT_ID(ffn_moe_down-80): 184 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 353 us MUL_MAT_ID(ffn_moe_down-81): 162 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 366 us MUL_MAT_ID(ffn_moe_down-82): 209 us ggml_barrier(...): 328 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 394 us MUL_MAT_ID(ffn_moe_down-83): 285 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 361 us MUL_MAT_ID(ffn_moe_down-84): 292 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 385 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 405 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 414 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 395 us MUL_MAT_ID(ffn_moe_down-88): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 360 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 357 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 387 us MUL_MAT_ID(ffn_moe_down-91): 275 us ggml_barrier(...): 28 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 395 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 374 us MUL_MAT_ID(ffn_moe_down-26): 219 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 397 us MUL_MAT_ID(ffn_moe_down-27): 203 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 368 us MUL_MAT_ID(ffn_moe_down-28): 372 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 208 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 376 us MUL_MAT_ID(ffn_moe_down-30): 181 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 440 us MUL_MAT_ID(ffn_moe_down-31): 287 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 391 us MUL_MAT_ID(ffn_moe_down-32): 201 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 360 us MUL_MAT_ID(ffn_moe_down-33): 559 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 406 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 385 us MUL_MAT_ID(ffn_moe_down-35): 204 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 390 us MUL_MAT_ID(ffn_moe_down-36): 215 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 351 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 398 us MUL_MAT_ID(ffn_moe_down-38): 533 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 392 us MUL_MAT_ID(ffn_moe_down-39): 283 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 357 us MUL_MAT_ID(ffn_moe_down-40): 241 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 359 us MUL_MAT_ID(ffn_moe_down-41): 225 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 382 us MUL_MAT_ID(ffn_moe_down-42): 184 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 373 us MUL_MAT_ID(ffn_moe_down-43): 224 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 398 us MUL_MAT_ID(ffn_moe_down-44): 205 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 357 us MUL_MAT_ID(ffn_moe_down-45): 199 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 383 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 352 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 395 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 365 us MUL_MAT_ID(ffn_moe_down-49): 534 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 362 us MUL_MAT_ID(ffn_moe_down-50): 222 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 368 us MUL_MAT_ID(ffn_moe_down-51): 191 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 385 us MUL_MAT_ID(ffn_moe_down-52): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 386 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 349 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 400 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 347 us MUL_MAT_ID(ffn_moe_down-56): 244 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 351 us MUL_MAT_ID(ffn_moe_down-57): 208 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 379 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 370 us MUL_MAT_ID(ffn_moe_down-59): 204 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 400 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 385 us MUL_MAT_ID(ffn_moe_down-61): 179 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 353 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 399 us MUL_MAT_ID(ffn_moe_down-63): 207 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 390 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 354 us MUL_MAT_ID(ffn_moe_down-65): 211 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 384 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 357 us MUL_MAT_ID(ffn_moe_down-67): 209 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 408 us MUL_MAT_ID(ffn_moe_down-68): 312 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 437 us MUL_MAT_ID(ffn_moe_down-69): 305 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 366 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 385 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 403 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 385 us MUL_MAT_ID(ffn_moe_down-73): 214 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 355 us MUL_MAT_ID(ffn_moe_down-74): 528 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 380 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 367 us MUL_MAT_ID(ffn_moe_down-76): 216 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 336 us MUL_MAT_ID(ffn_moe_down-77): 552 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 352 us MUL_MAT_ID(ffn_moe_down-78): 210 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 348 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 372 us MUL_MAT_ID(ffn_moe_down-80): 202 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 402 us MUL_MAT_ID(ffn_moe_down-81): 208 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 384 us MUL_MAT_ID(ffn_moe_down-82): 199 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 398 us MUL_MAT_ID(ffn_moe_down-83): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 485 us MUL_MAT_ID(ffn_moe_down-84): 197 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 390 us MUL_MAT_ID(ffn_moe_down-85): 243 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 433 us MUL_MAT_ID(ffn_moe_down-86): 206 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 370 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 414 us MUL_MAT_ID(ffn_moe_down-88): 399 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 360 us MUL_MAT_ID(ffn_moe_down-89): 191 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 363 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 394 us MUL_MAT_ID(ffn_moe_down-91): 262 us ggml_barrier(...): 39 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 390 us MUL_MAT_ID(ffn_moe_down-25): 186 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 401 us MUL_MAT_ID(ffn_moe_down-26): 183 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 412 us MUL_MAT_ID(ffn_moe_down-27): 200 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 364 us MUL_MAT_ID(ffn_moe_down-28): 220 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 362 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 383 us MUL_MAT_ID(ffn_moe_down-30): 179 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 381 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 398 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 434 us MUL_MAT_ID(ffn_moe_down-33): 206 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 379 us MUL_MAT_ID(ffn_moe_down-34): 205 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 382 us MUL_MAT_ID(ffn_moe_down-35): 342 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 536 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 363 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 404 us MUL_MAT_ID(ffn_moe_down-38): 264 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 363 us MUL_MAT_ID(ffn_moe_down-39): 229 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 380 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 410 us MUL_MAT_ID(ffn_moe_down-41): 221 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 359 us MUL_MAT_ID(ffn_moe_down-42): 195 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 375 us MUL_MAT_ID(ffn_moe_down-43): 233 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 382 us MUL_MAT_ID(ffn_moe_down-44): 187 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 392 us MUL_MAT_ID(ffn_moe_down-45): 197 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 385 us MUL_MAT_ID(ffn_moe_down-46): 202 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 386 us MUL_MAT_ID(ffn_moe_down-47): 211 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 360 us MUL_MAT_ID(ffn_moe_down-48): 226 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 417 us MUL_MAT_ID(ffn_moe_down-49): 229 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 393 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 479 us MUL_MAT_ID(ffn_moe_down-51): 186 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 396 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 382 us MUL_MAT_ID(ffn_moe_down-53): 293 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 421 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 409 us MUL_MAT_ID(ffn_moe_down-55): 355 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 423 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 384 us MUL_MAT_ID(ffn_moe_down-57): 220 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 416 us MUL_MAT_ID(ffn_moe_down-58): 185 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 395 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 495 us MUL_MAT_ID(ffn_moe_down-60): 203 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 379 us MUL_MAT_ID(ffn_moe_down-61): 209 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 369 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 386 us MUL_MAT_ID(ffn_moe_down-63): 347 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 383 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 389 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 355 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 403 us MUL_MAT_ID(ffn_moe_down-67): 305 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 366 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 379 us MUL_MAT_ID(ffn_moe_down-69): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 369 us MUL_MAT_ID(ffn_moe_down-70): 395 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 354 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 370 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 362 us MUL_MAT_ID(ffn_moe_down-73): 370 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 358 us MUL_MAT_ID(ffn_moe_down-74): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 366 us MUL_MAT_ID(ffn_moe_down-75): 220 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 377 us MUL_MAT_ID(ffn_moe_down-76): 200 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 403 us MUL_MAT_ID(ffn_moe_down-77): 216 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 383 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 392 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 358 us MUL_MAT_ID(ffn_moe_down-80): 325 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 362 us MUL_MAT_ID(ffn_moe_down-81): 207 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 402 us MUL_MAT_ID(ffn_moe_down-82): 214 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 384 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 413 us MUL_MAT_ID(ffn_moe_down-84): 223 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 415 us MUL_MAT_ID(ffn_moe_down-85): 574 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 385 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 387 us MUL_MAT_ID(ffn_moe_down-87): 197 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 400 us MUL_MAT_ID(ffn_moe_down-88): 344 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 374 us MUL_MAT_ID(ffn_moe_down-89): 186 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 408 us MUL_MAT_ID(ffn_moe_down-90): 443 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 531 us MUL_MAT_ID(ffn_moe_down-91): 269 us ggml_barrier(...): 96 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 405 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 386 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 366 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 367 us MUL_MAT_ID(ffn_moe_down-27): 232 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 398 us MUL_MAT_ID(ffn_moe_down-28): 211 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 412 us MUL_MAT_ID(ffn_moe_down-29): 203 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 384 us MUL_MAT_ID(ffn_moe_down-30): 204 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 423 us MUL_MAT_ID(ffn_moe_down-31): 276 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 480 us MUL_MAT_ID(ffn_moe_down-32): 183 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 471 us MUL_MAT_ID(ffn_moe_down-33): 212 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 211 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 379 us MUL_MAT_ID(ffn_moe_down-35): 237 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 519 us MUL_MAT_ID(ffn_moe_down-36): 218 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 386 us MUL_MAT_ID(ffn_moe_down-37): 213 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 381 us MUL_MAT_ID(ffn_moe_down-38): 220 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 372 us MUL_MAT_ID(ffn_moe_down-39): 226 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 363 us MUL_MAT_ID(ffn_moe_down-40): 210 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 361 us MUL_MAT_ID(ffn_moe_down-41): 282 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 422 us MUL_MAT_ID(ffn_moe_down-42): 247 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 414 us MUL_MAT_ID(ffn_moe_down-43): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 428 us MUL_MAT_ID(ffn_moe_down-44): 237 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 349 us MUL_MAT_ID(ffn_moe_down-45): 314 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 367 us MUL_MAT_ID(ffn_moe_down-46): 552 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 362 us MUL_MAT_ID(ffn_moe_down-47): 200 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 510 us MUL_MAT_ID(ffn_moe_down-48): 194 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 349 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 396 us MUL_MAT_ID(ffn_moe_down-50): 228 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 389 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 358 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 449 us MUL_MAT_ID(ffn_moe_down-53): 182 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 380 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 409 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 360 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 365 us MUL_MAT_ID(ffn_moe_down-57): 193 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 431 us MUL_MAT_ID(ffn_moe_down-58): 354 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 453 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 373 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 480 us MUL_MAT_ID(ffn_moe_down-61): 170 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 386 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 390 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 361 us MUL_MAT_ID(ffn_moe_down-64): 274 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 403 us MUL_MAT_ID(ffn_moe_down-65): 268 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 402 us MUL_MAT_ID(ffn_moe_down-66): 202 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 370 us MUL_MAT_ID(ffn_moe_down-67): 371 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 359 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 370 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 498 us MUL_MAT_ID(ffn_moe_down-70): 332 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 417 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 389 us MUL_MAT_ID(ffn_moe_down-72): 203 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 370 us MUL_MAT_ID(ffn_moe_down-73): 538 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 373 us MUL_MAT_ID(ffn_moe_down-74): 375 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 432 us MUL_MAT_ID(ffn_moe_down-75): 243 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 436 us MUL_MAT_ID(ffn_moe_down-76): 261 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 389 us MUL_MAT_ID(ffn_moe_down-77): 268 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 412 us MUL_MAT_ID(ffn_moe_down-78): 474 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 413 us MUL_MAT_ID(ffn_moe_down-79): 197 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 378 us MUL_MAT_ID(ffn_moe_down-80): 292 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 465 us MUL_MAT_ID(ffn_moe_down-81): 251 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 390 us MUL_MAT_ID(ffn_moe_down-82): 181 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 373 us MUL_MAT_ID(ffn_moe_down-83): 224 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 379 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 378 us MUL_MAT_ID(ffn_moe_down-85): 246 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 389 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 373 us MUL_MAT_ID(ffn_moe_down-87): 184 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 382 us MUL_MAT_ID(ffn_moe_down-88): 431 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 365 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 377 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 385 us MUL_MAT_ID(ffn_moe_down-91): 302 us ggml_barrier(...): 246 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 397 us MUL_MAT_ID(ffn_moe_down-25): 205 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 369 us MUL_MAT_ID(ffn_moe_down-26): 181 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 422 us MUL_MAT_ID(ffn_moe_down-27): 251 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 355 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 359 us MUL_MAT_ID(ffn_moe_down-29): 191 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 373 us MUL_MAT_ID(ffn_moe_down-30): 191 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 387 us MUL_MAT_ID(ffn_moe_down-31): 187 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 379 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 400 us MUL_MAT_ID(ffn_moe_down-33): 220 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 503 us MUL_MAT_ID(ffn_moe_down-34): 213 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 384 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 207 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 409 us MUL_MAT_ID(ffn_moe_down-37): 225 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 386 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 374 us MUL_MAT_ID(ffn_moe_down-39): 251 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 380 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 397 us MUL_MAT_ID(ffn_moe_down-41): 207 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 376 us MUL_MAT_ID(ffn_moe_down-42): 214 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 419 us MUL_MAT_ID(ffn_moe_down-43): 277 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 391 us MUL_MAT_ID(ffn_moe_down-44): 267 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 392 us MUL_MAT_ID(ffn_moe_down-45): 210 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 394 us MUL_MAT_ID(ffn_moe_down-46): 203 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 362 us MUL_MAT_ID(ffn_moe_down-47): 313 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 419 us MUL_MAT_ID(ffn_moe_down-48): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 445 us MUL_MAT_ID(ffn_moe_down-49): 185 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 452 us MUL_MAT_ID(ffn_moe_down-50): 250 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 489 us MUL_MAT_ID(ffn_moe_down-51): 356 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 426 us MUL_MAT_ID(ffn_moe_down-52): 436 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 382 us MUL_MAT_ID(ffn_moe_down-53): 191 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 383 us MUL_MAT_ID(ffn_moe_down-54): 192 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 403 us MUL_MAT_ID(ffn_moe_down-55): 181 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 394 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 377 us MUL_MAT_ID(ffn_moe_down-57): 229 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 411 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 377 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 382 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 374 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 455 us MUL_MAT_ID(ffn_moe_down-62): 166 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 366 us MUL_MAT_ID(ffn_moe_down-63): 228 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 411 us MUL_MAT_ID(ffn_moe_down-64): 189 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 376 us MUL_MAT_ID(ffn_moe_down-65): 351 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 423 us MUL_MAT_ID(ffn_moe_down-66): 204 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 359 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 353 us MUL_MAT_ID(ffn_moe_down-68): 207 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 368 us MUL_MAT_ID(ffn_moe_down-69): 232 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 385 us MUL_MAT_ID(ffn_moe_down-70): 242 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 377 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 356 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 365 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 349 us MUL_MAT_ID(ffn_moe_down-74): 449 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 363 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 370 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 358 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 370 us MUL_MAT_ID(ffn_moe_down-78): 198 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 398 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 393 us MUL_MAT_ID(ffn_moe_down-80): 207 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 364 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 391 us MUL_MAT_ID(ffn_moe_down-82): 204 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 365 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 367 us MUL_MAT_ID(ffn_moe_down-84): 225 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 337 us MUL_MAT_ID(ffn_moe_down-85): 256 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 345 us MUL_MAT_ID(ffn_moe_down-86): 227 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 410 us MUL_MAT_ID(ffn_moe_down-87): 221 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 369 us MUL_MAT_ID(ffn_moe_down-88): 214 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 397 us MUL_MAT_ID(ffn_moe_down-89): 225 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 365 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 378 us MUL_MAT_ID(ffn_moe_down-91): 244 us ggml_barrier(...): 28 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 418 us MUL_MAT_ID(ffn_moe_down-25): 356 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 435 us MUL_MAT_ID(ffn_moe_down-26): 319 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 382 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 422 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 200 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 457 us MUL_MAT_ID(ffn_moe_down-30): 302 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 399 us MUL_MAT_ID(ffn_moe_down-31): 199 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 385 us MUL_MAT_ID(ffn_moe_down-32): 183 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 402 us MUL_MAT_ID(ffn_moe_down-33): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 384 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 361 us MUL_MAT_ID(ffn_moe_down-35): 218 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 411 us MUL_MAT_ID(ffn_moe_down-36): 201 us ggml_barrier(...): 348 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 395 us MUL_MAT_ID(ffn_moe_down-37): 238 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 380 us MUL_MAT_ID(ffn_moe_down-38): 347 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 411 us MUL_MAT_ID(ffn_moe_down-39): 394 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 361 us MUL_MAT_ID(ffn_moe_down-40): 246 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 424 us MUL_MAT_ID(ffn_moe_down-41): 209 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 371 us MUL_MAT_ID(ffn_moe_down-42): 367 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 437 us MUL_MAT_ID(ffn_moe_down-43): 230 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 374 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 501 us MUL_MAT_ID(ffn_moe_down-45): 187 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 405 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 368 us MUL_MAT_ID(ffn_moe_down-47): 284 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 502 us MUL_MAT_ID(ffn_moe_down-48): 216 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 395 us MUL_MAT_ID(ffn_moe_down-49): 204 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 475 us MUL_MAT_ID(ffn_moe_down-50): 262 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 381 us MUL_MAT_ID(ffn_moe_down-51): 189 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 386 us MUL_MAT_ID(ffn_moe_down-52): 372 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 350 us MUL_MAT_ID(ffn_moe_down-53): 183 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 394 us MUL_MAT_ID(ffn_moe_down-54): 397 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 359 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 398 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 457 us MUL_MAT_ID(ffn_moe_down-57): 295 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 401 us MUL_MAT_ID(ffn_moe_down-58): 204 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 397 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 497 us MUL_MAT_ID(ffn_moe_down-60): 195 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 386 us MUL_MAT_ID(ffn_moe_down-61): 338 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 356 us MUL_MAT_ID(ffn_moe_down-62): 243 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 390 us MUL_MAT_ID(ffn_moe_down-63): 213 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 392 us MUL_MAT_ID(ffn_moe_down-64): 199 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 351 us MUL_MAT_ID(ffn_moe_down-65): 240 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 348 us MUL_MAT_ID(ffn_moe_down-66): 178 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 375 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 390 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 429 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 393 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 362 us MUL_MAT_ID(ffn_moe_down-71): 277 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 394 us MUL_MAT_ID(ffn_moe_down-72): 391 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 392 us MUL_MAT_ID(ffn_moe_down-73): 331 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 442 us MUL_MAT_ID(ffn_moe_down-74): 194 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 374 us MUL_MAT_ID(ffn_moe_down-75): 200 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 379 us MUL_MAT_ID(ffn_moe_down-76): 410 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 406 us MUL_MAT_ID(ffn_moe_down-77): 242 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 402 us MUL_MAT_ID(ffn_moe_down-78): 376 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 412 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 354 us MUL_MAT_ID(ffn_moe_down-80): 580 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 379 us MUL_MAT_ID(ffn_moe_down-81): 205 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 383 us MUL_MAT_ID(ffn_moe_down-82): 398 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 439 us MUL_MAT_ID(ffn_moe_down-83): 257 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 415 us MUL_MAT_ID(ffn_moe_down-84): 348 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 355 us MUL_MAT_ID(ffn_moe_down-85): 237 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 390 us MUL_MAT_ID(ffn_moe_down-86): 272 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 437 us MUL_MAT_ID(ffn_moe_down-87): 280 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 391 us MUL_MAT_ID(ffn_moe_down-88): 211 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 378 us MUL_MAT_ID(ffn_moe_down-89): 220 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 463 us MUL_MAT_ID(ffn_moe_down-90): 210 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 402 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 17 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 385 us MUL_MAT_ID(ffn_moe_down-25): 193 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 408 us MUL_MAT_ID(ffn_moe_down-26): 184 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 392 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 397 us MUL_MAT_ID(ffn_moe_down-28): 180 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 391 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 372 us MUL_MAT_ID(ffn_moe_down-30): 212 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 499 us MUL_MAT_ID(ffn_moe_down-31): 197 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 366 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 394 us MUL_MAT_ID(ffn_moe_down-33): 239 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 389 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 358 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 383 us MUL_MAT_ID(ffn_moe_down-36): 210 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 445 us MUL_MAT_ID(ffn_moe_down-37): 279 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 387 us MUL_MAT_ID(ffn_moe_down-38): 186 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 373 us MUL_MAT_ID(ffn_moe_down-39): 242 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 409 us MUL_MAT_ID(ffn_moe_down-40): 204 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 392 us MUL_MAT_ID(ffn_moe_down-41): 191 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 369 us MUL_MAT_ID(ffn_moe_down-42): 219 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 388 us MUL_MAT_ID(ffn_moe_down-43): 260 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 362 us MUL_MAT_ID(ffn_moe_down-44): 224 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 379 us MUL_MAT_ID(ffn_moe_down-45): 213 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 356 us MUL_MAT_ID(ffn_moe_down-46): 234 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 368 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 357 us MUL_MAT_ID(ffn_moe_down-48): 283 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 388 us MUL_MAT_ID(ffn_moe_down-49): 215 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 391 us MUL_MAT_ID(ffn_moe_down-50): 271 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 397 us MUL_MAT_ID(ffn_moe_down-51): 208 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 358 us MUL_MAT_ID(ffn_moe_down-52): 228 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 399 us MUL_MAT_ID(ffn_moe_down-53): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 391 us MUL_MAT_ID(ffn_moe_down-54): 212 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 390 us MUL_MAT_ID(ffn_moe_down-55): 180 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 412 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 379 us MUL_MAT_ID(ffn_moe_down-57): 195 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 384 us MUL_MAT_ID(ffn_moe_down-58): 275 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 395 us MUL_MAT_ID(ffn_moe_down-59): 258 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 376 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 397 us MUL_MAT_ID(ffn_moe_down-61): 184 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 389 us MUL_MAT_ID(ffn_moe_down-62): 186 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 372 us MUL_MAT_ID(ffn_moe_down-63): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 362 us MUL_MAT_ID(ffn_moe_down-64): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 353 us MUL_MAT_ID(ffn_moe_down-65): 198 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 395 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 366 us MUL_MAT_ID(ffn_moe_down-67): 192 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 371 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 352 us MUL_MAT_ID(ffn_moe_down-69): 205 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 407 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 393 us MUL_MAT_ID(ffn_moe_down-71): 187 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 397 us MUL_MAT_ID(ffn_moe_down-72): 430 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 367 us MUL_MAT_ID(ffn_moe_down-73): 219 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 383 us MUL_MAT_ID(ffn_moe_down-74): 253 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 389 us MUL_MAT_ID(ffn_moe_down-75): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 368 us MUL_MAT_ID(ffn_moe_down-76): 181 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 359 us MUL_MAT_ID(ffn_moe_down-77): 214 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 414 us MUL_MAT_ID(ffn_moe_down-78): 390 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 382 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 408 us MUL_MAT_ID(ffn_moe_down-80): 218 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 348 us MUL_MAT_ID(ffn_moe_down-81): 188 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 379 us MUL_MAT_ID(ffn_moe_down-82): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 391 us MUL_MAT_ID(ffn_moe_down-83): 187 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 377 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 354 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 376 us MUL_MAT_ID(ffn_moe_down-86): 196 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 335 us MUL_MAT_ID(ffn_moe_down-87): 206 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 363 us MUL_MAT_ID(ffn_moe_down-88): 179 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 341 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 354 us MUL_MAT_ID(ffn_moe_down-90): 220 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 370 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 78 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 469 us MUL_MAT_ID(ffn_moe_down-25): 238 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 452 us MUL_MAT_ID(ffn_moe_down-26): 168 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 420 us MUL_MAT_ID(ffn_moe_down-27): 188 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 433 us MUL_MAT_ID(ffn_moe_down-28): 155 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 373 us MUL_MAT_ID(ffn_moe_down-29): 227 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 178 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 387 us MUL_MAT_ID(ffn_moe_down-31): 171 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 422 us MUL_MAT_ID(ffn_moe_down-32): 236 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 396 us MUL_MAT_ID(ffn_moe_down-33): 193 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 368 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 398 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 396 us MUL_MAT_ID(ffn_moe_down-36): 273 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 387 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 374 us MUL_MAT_ID(ffn_moe_down-38): 197 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 368 us MUL_MAT_ID(ffn_moe_down-39): 265 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 406 us MUL_MAT_ID(ffn_moe_down-40): 258 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 388 us MUL_MAT_ID(ffn_moe_down-41): 301 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 455 us MUL_MAT_ID(ffn_moe_down-42): 278 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 348 us MUL_MAT_ID(ffn_moe_down-43): 187 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 500 us MUL_MAT_ID(ffn_moe_down-44): 299 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 476 us MUL_MAT_ID(ffn_moe_down-45): 182 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 431 us MUL_MAT_ID(ffn_moe_down-46): 290 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 395 us MUL_MAT_ID(ffn_moe_down-47): 223 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 373 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 376 us MUL_MAT_ID(ffn_moe_down-49): 203 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 407 us MUL_MAT_ID(ffn_moe_down-50): 356 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 375 us MUL_MAT_ID(ffn_moe_down-51): 358 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 457 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 388 us MUL_MAT_ID(ffn_moe_down-53): 500 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 356 us MUL_MAT_ID(ffn_moe_down-54): 382 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 401 us MUL_MAT_ID(ffn_moe_down-55): 205 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 357 us MUL_MAT_ID(ffn_moe_down-56): 207 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 441 us MUL_MAT_ID(ffn_moe_down-57): 434 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 353 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 361 us MUL_MAT_ID(ffn_moe_down-59): 219 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 386 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 428 us MUL_MAT_ID(ffn_moe_down-61): 232 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 391 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 504 us MUL_MAT_ID(ffn_moe_down-63): 202 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 373 us MUL_MAT_ID(ffn_moe_down-64): 264 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 418 us MUL_MAT_ID(ffn_moe_down-65): 351 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 362 us MUL_MAT_ID(ffn_moe_down-66): 180 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 403 us MUL_MAT_ID(ffn_moe_down-67): 189 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 372 us MUL_MAT_ID(ffn_moe_down-68): 226 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 386 us MUL_MAT_ID(ffn_moe_down-69): 560 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 430 us MUL_MAT_ID(ffn_moe_down-70): 284 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 371 us MUL_MAT_ID(ffn_moe_down-71): 508 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 400 us MUL_MAT_ID(ffn_moe_down-72): 206 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 377 us MUL_MAT_ID(ffn_moe_down-73): 219 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 388 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 424 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 356 us MUL_MAT_ID(ffn_moe_down-76): 200 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 341 us MUL_MAT_ID(ffn_moe_down-77): 230 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 414 us MUL_MAT_ID(ffn_moe_down-78): 337 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 397 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 446 us MUL_MAT_ID(ffn_moe_down-80): 256 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 403 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 393 us MUL_MAT_ID(ffn_moe_down-82): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 377 us MUL_MAT_ID(ffn_moe_down-83): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 357 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 383 us MUL_MAT_ID(ffn_moe_down-85): 237 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 391 us MUL_MAT_ID(ffn_moe_down-86): 206 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 393 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 386 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 364 us MUL_MAT_ID(ffn_moe_down-89): 309 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 388 us MUL_MAT_ID(ffn_moe_down-90): 230 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 402 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 17 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 408 us MUL_MAT_ID(ffn_moe_down-25): 177 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 507 us MUL_MAT_ID(ffn_moe_down-26): 175 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 406 us MUL_MAT_ID(ffn_moe_down-27): 240 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 385 us MUL_MAT_ID(ffn_moe_down-28): 188 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 374 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 432 us MUL_MAT_ID(ffn_moe_down-30): 164 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 365 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 377 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 446 us MUL_MAT_ID(ffn_moe_down-33): 232 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 381 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 404 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 359 us MUL_MAT_ID(ffn_moe_down-36): 366 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 381 us MUL_MAT_ID(ffn_moe_down-37): 258 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 358 us MUL_MAT_ID(ffn_moe_down-38): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 364 us MUL_MAT_ID(ffn_moe_down-39): 638 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 394 us MUL_MAT_ID(ffn_moe_down-40): 237 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 393 us MUL_MAT_ID(ffn_moe_down-41): 218 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 423 us MUL_MAT_ID(ffn_moe_down-42): 192 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 344 us MUL_MAT_ID(ffn_moe_down-43): 245 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 354 us MUL_MAT_ID(ffn_moe_down-44): 299 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 368 us MUL_MAT_ID(ffn_moe_down-45): 212 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 380 us MUL_MAT_ID(ffn_moe_down-46): 213 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 385 us MUL_MAT_ID(ffn_moe_down-47): 230 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 386 us MUL_MAT_ID(ffn_moe_down-48): 221 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 387 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 420 us MUL_MAT_ID(ffn_moe_down-50): 234 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 358 us MUL_MAT_ID(ffn_moe_down-51): 180 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 385 us MUL_MAT_ID(ffn_moe_down-52): 186 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 373 us MUL_MAT_ID(ffn_moe_down-53): 187 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 359 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 382 us MUL_MAT_ID(ffn_moe_down-55): 203 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 392 us MUL_MAT_ID(ffn_moe_down-57): 176 us ggml_barrier(...): 379 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 374 us MUL_MAT_ID(ffn_moe_down-58): 306 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 372 us MUL_MAT_ID(ffn_moe_down-59): 213 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 365 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 359 us MUL_MAT_ID(ffn_moe_down-61): 179 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 380 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 376 us MUL_MAT_ID(ffn_moe_down-63): 236 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 485 us MUL_MAT_ID(ffn_moe_down-64): 319 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 367 us MUL_MAT_ID(ffn_moe_down-65): 230 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 417 us MUL_MAT_ID(ffn_moe_down-66): 211 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 363 us MUL_MAT_ID(ffn_moe_down-67): 217 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 451 us MUL_MAT_ID(ffn_moe_down-68): 196 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 396 us MUL_MAT_ID(ffn_moe_down-69): 219 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 402 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 387 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 381 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 346 us MUL_MAT_ID(ffn_moe_down-73): 545 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 399 us MUL_MAT_ID(ffn_moe_down-74): 191 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 372 us MUL_MAT_ID(ffn_moe_down-75): 198 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 446 us MUL_MAT_ID(ffn_moe_down-76): 286 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 370 us MUL_MAT_ID(ffn_moe_down-77): 191 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 370 us MUL_MAT_ID(ffn_moe_down-78): 178 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 398 us MUL_MAT_ID(ffn_moe_down-79): 189 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 375 us MUL_MAT_ID(ffn_moe_down-80): 410 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 360 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 375 us MUL_MAT_ID(ffn_moe_down-82): 200 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 387 us MUL_MAT_ID(ffn_moe_down-83): 214 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 391 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 389 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 372 us MUL_MAT_ID(ffn_moe_down-86): 307 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 388 us MUL_MAT_ID(ffn_moe_down-87): 183 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 554 us MUL_MAT_ID(ffn_moe_down-88): 539 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 380 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 365 us MUL_MAT_ID(ffn_moe_down-90): 244 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 399 us MUL_MAT_ID(ffn_moe_down-91): 245 us ggml_barrier(...): 136 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 391 us MUL_MAT_ID(ffn_moe_down-25): 191 us ggml_barrier(...): 370 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 386 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 392 us MUL_MAT_ID(ffn_moe_down-27): 207 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 363 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 404 us MUL_MAT_ID(ffn_moe_down-29): 227 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 401 us MUL_MAT_ID(ffn_moe_down-30): 210 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 400 us MUL_MAT_ID(ffn_moe_down-31): 213 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 437 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 414 us MUL_MAT_ID(ffn_moe_down-33): 205 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 409 us MUL_MAT_ID(ffn_moe_down-34): 190 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 413 us MUL_MAT_ID(ffn_moe_down-35): 371 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 390 us MUL_MAT_ID(ffn_moe_down-36): 205 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 417 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 428 us MUL_MAT_ID(ffn_moe_down-38): 182 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 389 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 404 us MUL_MAT_ID(ffn_moe_down-40): 247 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 443 us MUL_MAT_ID(ffn_moe_down-41): 316 us ggml_barrier(...): 367 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 404 us MUL_MAT_ID(ffn_moe_down-42): 205 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 400 us MUL_MAT_ID(ffn_moe_down-43): 389 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 413 us MUL_MAT_ID(ffn_moe_down-44): 184 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 384 us MUL_MAT_ID(ffn_moe_down-45): 198 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 382 us MUL_MAT_ID(ffn_moe_down-46): 186 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 357 us MUL_MAT_ID(ffn_moe_down-47): 192 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 401 us MUL_MAT_ID(ffn_moe_down-48): 231 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 427 us MUL_MAT_ID(ffn_moe_down-49): 213 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 416 us MUL_MAT_ID(ffn_moe_down-50): 192 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 403 us MUL_MAT_ID(ffn_moe_down-51): 219 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 404 us MUL_MAT_ID(ffn_moe_down-52): 197 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 387 us MUL_MAT_ID(ffn_moe_down-53): 196 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 400 us MUL_MAT_ID(ffn_moe_down-54): 218 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 377 us MUL_MAT_ID(ffn_moe_down-55): 347 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 388 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 378 us MUL_MAT_ID(ffn_moe_down-57): 209 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 370 us MUL_MAT_ID(ffn_moe_down-58): 187 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 394 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 357 us MUL_MAT_ID(ffn_moe_down-60): 538 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 532 us MUL_MAT_ID(ffn_moe_down-61): 498 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 401 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 388 us MUL_MAT_ID(ffn_moe_down-63): 234 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 375 us MUL_MAT_ID(ffn_moe_down-64): 340 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 384 us MUL_MAT_ID(ffn_moe_down-65): 186 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 381 us MUL_MAT_ID(ffn_moe_down-66): 282 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 394 us MUL_MAT_ID(ffn_moe_down-67): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 271 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 419 us MUL_MAT_ID(ffn_moe_down-69): 209 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 418 us MUL_MAT_ID(ffn_moe_down-70): 210 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 405 us MUL_MAT_ID(ffn_moe_down-71): 211 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 360 us MUL_MAT_ID(ffn_moe_down-72): 358 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 369 us MUL_MAT_ID(ffn_moe_down-73): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 426 us MUL_MAT_ID(ffn_moe_down-74): 212 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 497 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 361 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 389 us MUL_MAT_ID(ffn_moe_down-77): 206 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 383 us MUL_MAT_ID(ffn_moe_down-78): 354 us ggml_barrier(...): 353 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 407 us MUL_MAT_ID(ffn_moe_down-79): 215 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 340 us MUL_MAT_ID(ffn_moe_down-80): 219 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 355 us MUL_MAT_ID(ffn_moe_down-81): 249 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 359 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 404 us MUL_MAT_ID(ffn_moe_down-83): 208 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 368 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 396 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 405 us MUL_MAT_ID(ffn_moe_down-86): 197 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 374 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 434 us MUL_MAT_ID(ffn_moe_down-88): 239 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 383 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 359 us MUL_MAT_ID(ffn_moe_down-90): 225 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 368 us MUL_MAT_ID(ffn_moe_down-91): 370 us ggml_barrier(...): 399 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 439 us MUL_MAT_ID(ffn_moe_down-25): 428 us ggml_barrier(...): 423 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 412 us MUL_MAT_ID(ffn_moe_down-26): 191 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 386 us MUL_MAT_ID(ffn_moe_down-27): 202 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 357 us MUL_MAT_ID(ffn_moe_down-28): 200 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 415 us MUL_MAT_ID(ffn_moe_down-29): 273 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 372 us MUL_MAT_ID(ffn_moe_down-30): 228 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 364 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 380 us MUL_MAT_ID(ffn_moe_down-32): 203 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 435 us MUL_MAT_ID(ffn_moe_down-33): 187 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 394 us MUL_MAT_ID(ffn_moe_down-34): 213 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 428 us MUL_MAT_ID(ffn_moe_down-35): 337 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 369 us MUL_MAT_ID(ffn_moe_down-36): 189 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 359 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 376 us MUL_MAT_ID(ffn_moe_down-38): 188 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 370 us MUL_MAT_ID(ffn_moe_down-39): 670 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 440 us MUL_MAT_ID(ffn_moe_down-40): 242 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 384 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 409 us MUL_MAT_ID(ffn_moe_down-42): 203 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 400 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 405 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 365 us MUL_MAT_ID(ffn_moe_down-45): 224 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 406 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 383 us MUL_MAT_ID(ffn_moe_down-47): 221 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 367 us MUL_MAT_ID(ffn_moe_down-48): 170 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 404 us MUL_MAT_ID(ffn_moe_down-49): 196 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 371 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 367 us MUL_MAT_ID(ffn_moe_down-51): 225 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 369 us MUL_MAT_ID(ffn_moe_down-52): 184 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 398 us MUL_MAT_ID(ffn_moe_down-53): 230 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 388 us MUL_MAT_ID(ffn_moe_down-54): 322 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 392 us MUL_MAT_ID(ffn_moe_down-55): 362 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 400 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 340 us MUL_MAT_ID(ffn_moe_down-57): 201 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 391 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 389 us MUL_MAT_ID(ffn_moe_down-59): 186 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 390 us MUL_MAT_ID(ffn_moe_down-60): 190 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 386 us MUL_MAT_ID(ffn_moe_down-61): 395 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 364 us MUL_MAT_ID(ffn_moe_down-62): 231 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 491 us MUL_MAT_ID(ffn_moe_down-63): 283 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 387 us MUL_MAT_ID(ffn_moe_down-64): 435 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 447 us MUL_MAT_ID(ffn_moe_down-65): 281 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 408 us MUL_MAT_ID(ffn_moe_down-66): 166 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 389 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 364 us MUL_MAT_ID(ffn_moe_down-68): 222 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 381 us MUL_MAT_ID(ffn_moe_down-69): 195 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 368 us MUL_MAT_ID(ffn_moe_down-70): 460 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 444 us MUL_MAT_ID(ffn_moe_down-71): 206 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 389 us MUL_MAT_ID(ffn_moe_down-72): 355 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 388 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 487 us MUL_MAT_ID(ffn_moe_down-74): 181 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 437 us MUL_MAT_ID(ffn_moe_down-75): 382 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 397 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 379 us MUL_MAT_ID(ffn_moe_down-77): 229 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 423 us MUL_MAT_ID(ffn_moe_down-78): 207 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 370 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 407 us MUL_MAT_ID(ffn_moe_down-80): 181 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 360 us MUL_MAT_ID(ffn_moe_down-81): 237 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 481 us MUL_MAT_ID(ffn_moe_down-82): 162 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 527 us MUL_MAT_ID(ffn_moe_down-83): 235 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 376 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 376 us MUL_MAT_ID(ffn_moe_down-85): 668 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 393 us MUL_MAT_ID(ffn_moe_down-86): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 352 us MUL_MAT_ID(ffn_moe_down-87): 408 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 380 us MUL_MAT_ID(ffn_moe_down-88): 207 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 409 us MUL_MAT_ID(ffn_moe_down-89): 213 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 343 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 371 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 63 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 367 us MUL_MAT_ID(ffn_moe_down-25): 204 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 423 us MUL_MAT_ID(ffn_moe_down-26): 178 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 372 us MUL_MAT_ID(ffn_moe_down-27): 210 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 412 us MUL_MAT_ID(ffn_moe_down-28): 337 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 370 us MUL_MAT_ID(ffn_moe_down-29): 397 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 427 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 384 us MUL_MAT_ID(ffn_moe_down-31): 344 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 389 us MUL_MAT_ID(ffn_moe_down-32): 183 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 411 us MUL_MAT_ID(ffn_moe_down-33): 216 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 391 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 405 us MUL_MAT_ID(ffn_moe_down-35): 197 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 409 us MUL_MAT_ID(ffn_moe_down-36): 199 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 377 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 339 us MUL_MAT_ID(ffn_moe_down-38): 181 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 367 us MUL_MAT_ID(ffn_moe_down-39): 239 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 354 us MUL_MAT_ID(ffn_moe_down-40): 282 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 402 us MUL_MAT_ID(ffn_moe_down-41): 202 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 367 us MUL_MAT_ID(ffn_moe_down-42): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 497 us MUL_MAT_ID(ffn_moe_down-43): 171 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 364 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 386 us MUL_MAT_ID(ffn_moe_down-45): 394 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 414 us MUL_MAT_ID(ffn_moe_down-46): 213 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 372 us MUL_MAT_ID(ffn_moe_down-47): 208 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 396 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 398 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 432 us MUL_MAT_ID(ffn_moe_down-50): 213 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 361 us MUL_MAT_ID(ffn_moe_down-51): 211 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 376 us MUL_MAT_ID(ffn_moe_down-52): 206 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 377 us MUL_MAT_ID(ffn_moe_down-53): 219 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 371 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 378 us MUL_MAT_ID(ffn_moe_down-55): 524 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 378 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 406 us MUL_MAT_ID(ffn_moe_down-57): 294 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 415 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 376 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 389 us MUL_MAT_ID(ffn_moe_down-60): 270 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 417 us MUL_MAT_ID(ffn_moe_down-61): 382 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 371 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 384 us MUL_MAT_ID(ffn_moe_down-63): 516 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 377 us MUL_MAT_ID(ffn_moe_down-64): 235 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 408 us MUL_MAT_ID(ffn_moe_down-65): 393 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 358 us MUL_MAT_ID(ffn_moe_down-66): 503 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 384 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 406 us MUL_MAT_ID(ffn_moe_down-68): 475 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 393 us MUL_MAT_ID(ffn_moe_down-69): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 404 us MUL_MAT_ID(ffn_moe_down-70): 218 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 383 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 467 us MUL_MAT_ID(ffn_moe_down-72): 163 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 360 us MUL_MAT_ID(ffn_moe_down-73): 210 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 391 us MUL_MAT_ID(ffn_moe_down-74): 207 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 376 us MUL_MAT_ID(ffn_moe_down-75): 187 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 371 us MUL_MAT_ID(ffn_moe_down-76): 182 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 397 us MUL_MAT_ID(ffn_moe_down-77): 197 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 392 us MUL_MAT_ID(ffn_moe_down-78): 258 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 408 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 489 us MUL_MAT_ID(ffn_moe_down-80): 202 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 368 us MUL_MAT_ID(ffn_moe_down-81): 524 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 384 us MUL_MAT_ID(ffn_moe_down-82): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 379 us MUL_MAT_ID(ffn_moe_down-83): 188 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 362 us MUL_MAT_ID(ffn_moe_down-84): 215 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 500 us MUL_MAT_ID(ffn_moe_down-85): 266 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 366 us MUL_MAT_ID(ffn_moe_down-86): 388 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 419 us MUL_MAT_ID(ffn_moe_down-87): 185 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 380 us MUL_MAT_ID(ffn_moe_down-88): 208 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 396 us MUL_MAT_ID(ffn_moe_down-89): 186 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 388 us MUL_MAT_ID(ffn_moe_down-90): 224 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 357 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 52 us | glm4moe 355B.A32B Q3_K - Medium | 155.60 GiB | 369.08 B | CUDA | 94 | 8192 | 8192 | q8_0 | q8_0 | graph | 0.00 | tg128 | 7.13 ± 0.00 | Device 0: 11.8203 MiB Device 1: 12.2188 MiB Device 2: 12.6172 MiB Device 3: 12.2188 MiB GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 396 us MUL_MAT_ID(ffn_moe_down-25): 214 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 415 us MUL_MAT_ID(ffn_moe_down-26): 544 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 404 us MUL_MAT_ID(ffn_moe_down-27): 532 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 460 us MUL_MAT_ID(ffn_moe_down-28): 183 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 370 us MUL_MAT_ID(ffn_moe_down-29): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 422 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 367 us MUL_MAT_ID(ffn_moe_down-31): 200 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 369 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 400 us MUL_MAT_ID(ffn_moe_down-33): 366 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 385 us MUL_MAT_ID(ffn_moe_down-34): 168 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 415 us MUL_MAT_ID(ffn_moe_down-35): 225 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 366 us MUL_MAT_ID(ffn_moe_down-36): 209 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 348 us MUL_MAT_ID(ffn_moe_down-37): 249 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 367 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 363 us MUL_MAT_ID(ffn_moe_down-40): 196 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 396 us MUL_MAT_ID(ffn_moe_down-41): 222 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 400 us MUL_MAT_ID(ffn_moe_down-42): 527 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 369 us MUL_MAT_ID(ffn_moe_down-43): 208 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 400 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 378 us MUL_MAT_ID(ffn_moe_down-45): 213 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 365 us MUL_MAT_ID(ffn_moe_down-46): 195 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 387 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 416 us MUL_MAT_ID(ffn_moe_down-48): 224 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 417 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 390 us MUL_MAT_ID(ffn_moe_down-50): 358 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 385 us MUL_MAT_ID(ffn_moe_down-51): 265 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 376 us MUL_MAT_ID(ffn_moe_down-52): 203 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 358 us MUL_MAT_ID(ffn_moe_down-53): 216 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 383 us MUL_MAT_ID(ffn_moe_down-54): 190 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 388 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 400 us MUL_MAT_ID(ffn_moe_down-56): 208 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 391 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 339 us MUL_MAT_ID(ffn_moe_down-58): 202 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 386 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 351 us MUL_MAT_ID(ffn_moe_down-60): 551 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 408 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 361 us MUL_MAT_ID(ffn_moe_down-62): 204 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 391 us MUL_MAT_ID(ffn_moe_down-63): 501 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 432 us MUL_MAT_ID(ffn_moe_down-64): 201 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 382 us MUL_MAT_ID(ffn_moe_down-65): 387 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 419 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 401 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 392 us MUL_MAT_ID(ffn_moe_down-68): 211 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 404 us MUL_MAT_ID(ffn_moe_down-69): 195 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 381 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 346 us MUL_MAT_ID(ffn_moe_down-71): 215 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 362 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 408 us MUL_MAT_ID(ffn_moe_down-73): 186 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 400 us MUL_MAT_ID(ffn_moe_down-74): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 378 us MUL_MAT_ID(ffn_moe_down-75): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 380 us MUL_MAT_ID(ffn_moe_down-76): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 402 us MUL_MAT_ID(ffn_moe_down-77): 315 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 358 us MUL_MAT_ID(ffn_moe_down-78): 190 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 378 us MUL_MAT_ID(ffn_moe_down-79): 188 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 393 us MUL_MAT_ID(ffn_moe_down-80): 213 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 389 us MUL_MAT_ID(ffn_moe_down-81): 219 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 360 us MUL_MAT_ID(ffn_moe_down-82): 182 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 366 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 355 us MUL_MAT_ID(ffn_moe_down-84): 252 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 404 us MUL_MAT_ID(ffn_moe_down-85): 245 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 404 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 370 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 393 us MUL_MAT_ID(ffn_moe_down-88): 205 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 382 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 396 us MUL_MAT_ID(ffn_moe_down-90): 209 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 373 us MUL_MAT_ID(ffn_moe_down-91): 415 us ggml_barrier(...): 47 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 408 us MUL_MAT_ID(ffn_moe_down-25): 193 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 409 us MUL_MAT_ID(ffn_moe_down-26): 291 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 410 us MUL_MAT_ID(ffn_moe_down-27): 190 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 362 us MUL_MAT_ID(ffn_moe_down-28): 242 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 366 us MUL_MAT_ID(ffn_moe_down-29): 175 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 417 us MUL_MAT_ID(ffn_moe_down-30): 294 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 376 us MUL_MAT_ID(ffn_moe_down-31): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 379 us MUL_MAT_ID(ffn_moe_down-32): 186 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 359 us MUL_MAT_ID(ffn_moe_down-33): 406 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 391 us MUL_MAT_ID(ffn_moe_down-34): 212 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 488 us MUL_MAT_ID(ffn_moe_down-35): 207 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 379 us MUL_MAT_ID(ffn_moe_down-36): 331 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 377 us MUL_MAT_ID(ffn_moe_down-37): 243 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 362 us MUL_MAT_ID(ffn_moe_down-38): 232 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 400 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 367 us MUL_MAT_ID(ffn_moe_down-40): 245 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 349 us MUL_MAT_ID(ffn_moe_down-41): 182 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 367 us MUL_MAT_ID(ffn_moe_down-42): 183 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 355 us MUL_MAT_ID(ffn_moe_down-43): 296 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 458 us MUL_MAT_ID(ffn_moe_down-44): 194 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 390 us MUL_MAT_ID(ffn_moe_down-45): 203 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 361 us MUL_MAT_ID(ffn_moe_down-46): 259 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 511 us MUL_MAT_ID(ffn_moe_down-47): 228 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 502 us MUL_MAT_ID(ffn_moe_down-48): 163 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 435 us MUL_MAT_ID(ffn_moe_down-49): 244 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 394 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 357 us MUL_MAT_ID(ffn_moe_down-51): 259 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 372 us MUL_MAT_ID(ffn_moe_down-52): 268 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 390 us MUL_MAT_ID(ffn_moe_down-53): 243 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 406 us MUL_MAT_ID(ffn_moe_down-54): 295 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 417 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 379 us MUL_MAT_ID(ffn_moe_down-56): 274 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 477 us MUL_MAT_ID(ffn_moe_down-57): 315 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 395 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 430 us MUL_MAT_ID(ffn_moe_down-59): 300 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 410 us MUL_MAT_ID(ffn_moe_down-60): 284 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 382 us MUL_MAT_ID(ffn_moe_down-61): 581 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 460 us MUL_MAT_ID(ffn_moe_down-62): 201 us ggml_barrier(...): 327 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 438 us MUL_MAT_ID(ffn_moe_down-63): 245 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 396 us MUL_MAT_ID(ffn_moe_down-64): 233 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 432 us MUL_MAT_ID(ffn_moe_down-65): 182 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 421 us MUL_MAT_ID(ffn_moe_down-66): 435 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 353 us MUL_MAT_ID(ffn_moe_down-67): 182 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 377 us MUL_MAT_ID(ffn_moe_down-68): 197 us ggml_barrier(...): 361 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 401 us MUL_MAT_ID(ffn_moe_down-69): 187 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 374 us MUL_MAT_ID(ffn_moe_down-70): 172 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 390 us MUL_MAT_ID(ffn_moe_down-71): 197 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 398 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 376 us MUL_MAT_ID(ffn_moe_down-73): 262 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 389 us MUL_MAT_ID(ffn_moe_down-74): 207 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 349 us MUL_MAT_ID(ffn_moe_down-75): 220 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 375 us MUL_MAT_ID(ffn_moe_down-76): 261 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 389 us MUL_MAT_ID(ffn_moe_down-77): 184 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 433 us MUL_MAT_ID(ffn_moe_down-78): 206 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 397 us MUL_MAT_ID(ffn_moe_down-79): 265 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 376 us MUL_MAT_ID(ffn_moe_down-80): 269 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 354 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 470 us MUL_MAT_ID(ffn_moe_down-82): 205 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 383 us MUL_MAT_ID(ffn_moe_down-83): 236 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 484 us MUL_MAT_ID(ffn_moe_down-84): 214 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 387 us MUL_MAT_ID(ffn_moe_down-85): 234 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 370 us MUL_MAT_ID(ffn_moe_down-86): 205 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 394 us MUL_MAT_ID(ffn_moe_down-87): 212 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 380 us MUL_MAT_ID(ffn_moe_down-88): 224 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 356 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 404 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 444 us MUL_MAT_ID(ffn_moe_down-91): 249 us ggml_barrier(...): 210 us GET_ROWS(inp_embd): 11 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 371 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 374 us MUL_MAT_ID(ffn_moe_down-26): 516 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 402 us MUL_MAT_ID(ffn_moe_down-27): 209 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 402 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 425 us MUL_MAT_ID(ffn_moe_down-29): 286 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 367 us MUL_MAT_ID(ffn_moe_down-30): 219 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 379 us MUL_MAT_ID(ffn_moe_down-31): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 422 us MUL_MAT_ID(ffn_moe_down-32): 188 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 363 us MUL_MAT_ID(ffn_moe_down-33): 217 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 388 us MUL_MAT_ID(ffn_moe_down-34): 184 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 381 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 369 us MUL_MAT_ID(ffn_moe_down-36): 180 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 374 us MUL_MAT_ID(ffn_moe_down-37): 366 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 369 us MUL_MAT_ID(ffn_moe_down-38): 196 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 357 us MUL_MAT_ID(ffn_moe_down-39): 549 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 443 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 386 us MUL_MAT_ID(ffn_moe_down-41): 380 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 394 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 381 us MUL_MAT_ID(ffn_moe_down-43): 261 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 443 us MUL_MAT_ID(ffn_moe_down-44): 242 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 360 us MUL_MAT_ID(ffn_moe_down-45): 409 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 400 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 403 us MUL_MAT_ID(ffn_moe_down-47): 178 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 391 us MUL_MAT_ID(ffn_moe_down-48): 182 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 237 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 397 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 442 us MUL_MAT_ID(ffn_moe_down-51): 234 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 379 us MUL_MAT_ID(ffn_moe_down-52): 262 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 381 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 371 us MUL_MAT_ID(ffn_moe_down-54): 205 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 381 us MUL_MAT_ID(ffn_moe_down-55): 555 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 395 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 357 us MUL_MAT_ID(ffn_moe_down-57): 189 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 434 us MUL_MAT_ID(ffn_moe_down-58): 206 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 487 us MUL_MAT_ID(ffn_moe_down-59): 237 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 365 us MUL_MAT_ID(ffn_moe_down-60): 293 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 398 us MUL_MAT_ID(ffn_moe_down-61): 204 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 402 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 367 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 490 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 366 us MUL_MAT_ID(ffn_moe_down-65): 205 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 371 us MUL_MAT_ID(ffn_moe_down-66): 215 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 404 us MUL_MAT_ID(ffn_moe_down-67): 220 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 380 us MUL_MAT_ID(ffn_moe_down-68): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 449 us MUL_MAT_ID(ffn_moe_down-69): 208 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 384 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 370 us MUL_MAT_ID(ffn_moe_down-71): 385 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 427 us MUL_MAT_ID(ffn_moe_down-72): 168 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 182 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 389 us MUL_MAT_ID(ffn_moe_down-74): 209 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 467 us MUL_MAT_ID(ffn_moe_down-75): 220 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 379 us MUL_MAT_ID(ffn_moe_down-76): 298 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 408 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 415 us MUL_MAT_ID(ffn_moe_down-78): 312 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 391 us MUL_MAT_ID(ffn_moe_down-79): 249 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 367 us MUL_MAT_ID(ffn_moe_down-80): 344 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 461 us MUL_MAT_ID(ffn_moe_down-81): 181 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 379 us MUL_MAT_ID(ffn_moe_down-82): 207 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 390 us MUL_MAT_ID(ffn_moe_down-83): 269 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 446 us MUL_MAT_ID(ffn_moe_down-84): 307 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 449 us MUL_MAT_ID(ffn_moe_down-85): 278 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 433 us MUL_MAT_ID(ffn_moe_down-86): 218 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 419 us MUL_MAT_ID(ffn_moe_down-87): 624 us ggml_barrier(...): 327 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 385 us MUL_MAT_ID(ffn_moe_down-88): 279 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 385 us MUL_MAT_ID(ffn_moe_down-89): 298 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 401 us MUL_MAT_ID(ffn_moe_down-90): 227 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 400 us MUL_MAT_ID(ffn_moe_down-91): 422 us ggml_barrier(...): 192 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 410 us MUL_MAT_ID(ffn_moe_down-25): 197 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 390 us MUL_MAT_ID(ffn_moe_down-26): 210 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 416 us MUL_MAT_ID(ffn_moe_down-27): 200 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 364 us MUL_MAT_ID(ffn_moe_down-28): 220 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 376 us MUL_MAT_ID(ffn_moe_down-29): 242 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 377 us MUL_MAT_ID(ffn_moe_down-30): 184 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 387 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 361 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 382 us MUL_MAT_ID(ffn_moe_down-33): 223 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 361 us MUL_MAT_ID(ffn_moe_down-34): 234 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 408 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 394 us MUL_MAT_ID(ffn_moe_down-36): 207 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 404 us MUL_MAT_ID(ffn_moe_down-37): 356 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 372 us MUL_MAT_ID(ffn_moe_down-38): 200 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 363 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 379 us MUL_MAT_ID(ffn_moe_down-40): 602 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 399 us MUL_MAT_ID(ffn_moe_down-41): 209 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 389 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 387 us MUL_MAT_ID(ffn_moe_down-43): 212 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 392 us MUL_MAT_ID(ffn_moe_down-44): 204 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 355 us MUL_MAT_ID(ffn_moe_down-45): 530 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 358 us MUL_MAT_ID(ffn_moe_down-46): 186 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 399 us MUL_MAT_ID(ffn_moe_down-47): 208 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 438 us MUL_MAT_ID(ffn_moe_down-48): 261 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 381 us MUL_MAT_ID(ffn_moe_down-49): 390 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 401 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 345 us MUL_MAT_ID(ffn_moe_down-51): 199 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 374 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 369 us MUL_MAT_ID(ffn_moe_down-53): 529 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 387 us MUL_MAT_ID(ffn_moe_down-54): 190 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 416 us MUL_MAT_ID(ffn_moe_down-55): 185 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 376 us MUL_MAT_ID(ffn_moe_down-56): 171 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 388 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 388 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 400 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 381 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 381 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 378 us MUL_MAT_ID(ffn_moe_down-62): 241 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 383 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 362 us MUL_MAT_ID(ffn_moe_down-64): 265 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 396 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 380 us MUL_MAT_ID(ffn_moe_down-66): 205 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 369 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 406 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 388 us MUL_MAT_ID(ffn_moe_down-69): 390 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 391 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 398 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 449 us MUL_MAT_ID(ffn_moe_down-72): 211 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 397 us MUL_MAT_ID(ffn_moe_down-73): 219 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 361 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 405 us MUL_MAT_ID(ffn_moe_down-75): 239 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 391 us MUL_MAT_ID(ffn_moe_down-76): 197 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 397 us MUL_MAT_ID(ffn_moe_down-77): 379 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 397 us MUL_MAT_ID(ffn_moe_down-78): 204 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 396 us MUL_MAT_ID(ffn_moe_down-79): 214 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 379 us MUL_MAT_ID(ffn_moe_down-80): 211 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 508 us MUL_MAT_ID(ffn_moe_down-81): 186 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 396 us MUL_MAT_ID(ffn_moe_down-82): 209 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 379 us MUL_MAT_ID(ffn_moe_down-83): 223 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 376 us MUL_MAT_ID(ffn_moe_down-84): 246 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 409 us MUL_MAT_ID(ffn_moe_down-85): 215 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 395 us MUL_MAT_ID(ffn_moe_down-86): 204 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 369 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 373 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 396 us MUL_MAT_ID(ffn_moe_down-89): 427 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 405 us MUL_MAT_ID(ffn_moe_down-90): 197 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 350 us MUL_MAT_ID(ffn_moe_down-91): 345 us ggml_barrier(...): 247 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 373 us MUL_MAT_ID(ffn_moe_down-25): 197 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 377 us MUL_MAT_ID(ffn_moe_down-26): 181 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 384 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 385 us MUL_MAT_ID(ffn_moe_down-28): 223 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 388 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 512 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 401 us MUL_MAT_ID(ffn_moe_down-31): 516 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 386 us MUL_MAT_ID(ffn_moe_down-32): 390 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 375 us MUL_MAT_ID(ffn_moe_down-33): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 385 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 398 us MUL_MAT_ID(ffn_moe_down-35): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 423 us MUL_MAT_ID(ffn_moe_down-36): 270 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 369 us MUL_MAT_ID(ffn_moe_down-37): 649 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 373 us MUL_MAT_ID(ffn_moe_down-38): 353 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 407 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 397 us MUL_MAT_ID(ffn_moe_down-40): 240 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 406 us MUL_MAT_ID(ffn_moe_down-41): 200 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 370 us MUL_MAT_ID(ffn_moe_down-42): 181 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 361 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 346 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 391 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 364 us MUL_MAT_ID(ffn_moe_down-46): 228 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 377 us MUL_MAT_ID(ffn_moe_down-47): 319 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 372 us MUL_MAT_ID(ffn_moe_down-48): 187 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 338 us MUL_MAT_ID(ffn_moe_down-49): 208 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 413 us MUL_MAT_ID(ffn_moe_down-50): 184 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 361 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 383 us MUL_MAT_ID(ffn_moe_down-52): 227 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 370 us MUL_MAT_ID(ffn_moe_down-53): 218 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 346 us MUL_MAT_ID(ffn_moe_down-54): 223 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 385 us MUL_MAT_ID(ffn_moe_down-55): 380 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 387 us MUL_MAT_ID(ffn_moe_down-56): 249 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 378 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 381 us MUL_MAT_ID(ffn_moe_down-58): 353 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 397 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 408 us MUL_MAT_ID(ffn_moe_down-60): 202 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 378 us MUL_MAT_ID(ffn_moe_down-61): 537 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 388 us MUL_MAT_ID(ffn_moe_down-62): 224 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 247 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 394 us MUL_MAT_ID(ffn_moe_down-64): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 369 us MUL_MAT_ID(ffn_moe_down-65): 198 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 403 us MUL_MAT_ID(ffn_moe_down-66): 223 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 390 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 389 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 338 us MUL_MAT_ID(ffn_moe_down-69): 219 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 394 us MUL_MAT_ID(ffn_moe_down-70): 186 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 392 us MUL_MAT_ID(ffn_moe_down-71): 219 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 353 us MUL_MAT_ID(ffn_moe_down-72): 257 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 366 us MUL_MAT_ID(ffn_moe_down-73): 196 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 357 us MUL_MAT_ID(ffn_moe_down-74): 208 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 365 us MUL_MAT_ID(ffn_moe_down-75): 395 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 364 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 391 us MUL_MAT_ID(ffn_moe_down-77): 216 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 388 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 394 us MUL_MAT_ID(ffn_moe_down-79): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 371 us MUL_MAT_ID(ffn_moe_down-80): 210 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 341 us MUL_MAT_ID(ffn_moe_down-81): 285 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 360 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 407 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 371 us MUL_MAT_ID(ffn_moe_down-84): 244 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 373 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 428 us MUL_MAT_ID(ffn_moe_down-86): 210 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 370 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 366 us MUL_MAT_ID(ffn_moe_down-88): 543 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 375 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 397 us MUL_MAT_ID(ffn_moe_down-90): 609 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 394 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 18 us GET_ROWS(inp_embd): 5 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 417 us MUL_MAT_ID(ffn_moe_down-25): 368 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 544 us MUL_MAT_ID(ffn_moe_down-26): 572 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 348 us MUL_MAT_ID(ffn_moe_down-27): 232 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 385 us MUL_MAT_ID(ffn_moe_down-29): 205 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 404 us MUL_MAT_ID(ffn_moe_down-30): 250 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 396 us MUL_MAT_ID(ffn_moe_down-31): 200 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 404 us MUL_MAT_ID(ffn_moe_down-32): 280 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 419 us MUL_MAT_ID(ffn_moe_down-33): 280 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 506 us MUL_MAT_ID(ffn_moe_down-34): 262 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 418 us MUL_MAT_ID(ffn_moe_down-35): 243 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 401 us MUL_MAT_ID(ffn_moe_down-36): 446 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 453 us MUL_MAT_ID(ffn_moe_down-37): 253 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 369 us MUL_MAT_ID(ffn_moe_down-38): 302 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 416 us MUL_MAT_ID(ffn_moe_down-39): 368 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 434 us MUL_MAT_ID(ffn_moe_down-40): 292 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 384 us MUL_MAT_ID(ffn_moe_down-41): 437 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 383 us MUL_MAT_ID(ffn_moe_down-42): 216 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 358 us MUL_MAT_ID(ffn_moe_down-43): 240 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 377 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 476 us MUL_MAT_ID(ffn_moe_down-45): 331 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 491 us MUL_MAT_ID(ffn_moe_down-46): 208 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 393 us MUL_MAT_ID(ffn_moe_down-47): 189 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 403 us MUL_MAT_ID(ffn_moe_down-48): 208 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 407 us MUL_MAT_ID(ffn_moe_down-49): 206 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 373 us MUL_MAT_ID(ffn_moe_down-50): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 359 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 399 us MUL_MAT_ID(ffn_moe_down-52): 421 us ggml_barrier(...): 348 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 401 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 374 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 403 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 516 us MUL_MAT_ID(ffn_moe_down-56): 206 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 375 us MUL_MAT_ID(ffn_moe_down-57): 211 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 352 us MUL_MAT_ID(ffn_moe_down-58): 181 us ggml_barrier(...): 343 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 449 us MUL_MAT_ID(ffn_moe_down-59): 431 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 438 us MUL_MAT_ID(ffn_moe_down-60): 187 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 392 us MUL_MAT_ID(ffn_moe_down-61): 271 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 401 us MUL_MAT_ID(ffn_moe_down-62): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 383 us MUL_MAT_ID(ffn_moe_down-63): 345 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 462 us MUL_MAT_ID(ffn_moe_down-64): 187 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 408 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 421 us MUL_MAT_ID(ffn_moe_down-66): 405 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 368 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 478 us MUL_MAT_ID(ffn_moe_down-68): 339 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 412 us MUL_MAT_ID(ffn_moe_down-69): 283 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 403 us MUL_MAT_ID(ffn_moe_down-70): 220 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 348 us MUL_MAT_ID(ffn_moe_down-71): 282 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 392 us MUL_MAT_ID(ffn_moe_down-72): 299 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 386 us MUL_MAT_ID(ffn_moe_down-73): 221 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 363 us MUL_MAT_ID(ffn_moe_down-74): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 448 us MUL_MAT_ID(ffn_moe_down-75): 400 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 394 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 374 us MUL_MAT_ID(ffn_moe_down-77): 261 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 380 us MUL_MAT_ID(ffn_moe_down-78): 188 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 353 us MUL_MAT_ID(ffn_moe_down-79): 173 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 358 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 332 us MUL_MAT_ID(ffn_moe_down-81): 164 us ggml_barrier(...): 434 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 366 us MUL_MAT_ID(ffn_moe_down-82): 187 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 479 us MUL_MAT_ID(ffn_moe_down-83): 174 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 358 us MUL_MAT_ID(ffn_moe_down-84): 677 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 368 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 368 us MUL_MAT_ID(ffn_moe_down-86): 177 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 388 us MUL_MAT_ID(ffn_moe_down-87): 290 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 380 us MUL_MAT_ID(ffn_moe_down-88): 321 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 357 us MUL_MAT_ID(ffn_moe_down-89): 282 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 385 us MUL_MAT_ID(ffn_moe_down-90): 241 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 411 us MUL_MAT_ID(ffn_moe_down-91): 250 us ggml_barrier(...): 281 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 413 us MUL_MAT_ID(ffn_moe_down-25): 191 us ggml_barrier(...): 468 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 378 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 399 us MUL_MAT_ID(ffn_moe_down-27): 187 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 381 us MUL_MAT_ID(ffn_moe_down-28): 377 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 360 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 418 us MUL_MAT_ID(ffn_moe_down-30): 191 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 404 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 377 us MUL_MAT_ID(ffn_moe_down-32): 403 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 384 us MUL_MAT_ID(ffn_moe_down-33): 207 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 379 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 394 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 370 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 362 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 404 us MUL_MAT_ID(ffn_moe_down-38): 203 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 361 us MUL_MAT_ID(ffn_moe_down-39): 229 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 396 us MUL_MAT_ID(ffn_moe_down-40): 227 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 393 us MUL_MAT_ID(ffn_moe_down-41): 203 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 398 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 367 us MUL_MAT_ID(ffn_moe_down-43): 219 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 472 us MUL_MAT_ID(ffn_moe_down-44): 177 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 383 us MUL_MAT_ID(ffn_moe_down-45): 192 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 359 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 361 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 349 us MUL_MAT_ID(ffn_moe_down-48): 358 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 381 us MUL_MAT_ID(ffn_moe_down-49): 199 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 384 us MUL_MAT_ID(ffn_moe_down-50): 201 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 376 us MUL_MAT_ID(ffn_moe_down-51): 396 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 383 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 377 us MUL_MAT_ID(ffn_moe_down-53): 190 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 366 us MUL_MAT_ID(ffn_moe_down-54): 220 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 357 us MUL_MAT_ID(ffn_moe_down-55): 194 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 380 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 391 us MUL_MAT_ID(ffn_moe_down-57): 201 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 377 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 397 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 377 us MUL_MAT_ID(ffn_moe_down-60): 218 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 395 us MUL_MAT_ID(ffn_moe_down-61): 199 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 378 us MUL_MAT_ID(ffn_moe_down-62): 206 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 400 us MUL_MAT_ID(ffn_moe_down-63): 357 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 359 us MUL_MAT_ID(ffn_moe_down-64): 235 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 346 us MUL_MAT_ID(ffn_moe_down-65): 203 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 358 us MUL_MAT_ID(ffn_moe_down-66): 210 us ggml_barrier(...): 436 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 369 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 411 us MUL_MAT_ID(ffn_moe_down-68): 239 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 368 us MUL_MAT_ID(ffn_moe_down-69): 208 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 370 us MUL_MAT_ID(ffn_moe_down-70): 332 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 400 us MUL_MAT_ID(ffn_moe_down-71): 402 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 386 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 353 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 375 us MUL_MAT_ID(ffn_moe_down-74): 204 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 393 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 394 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 289 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 394 us MUL_MAT_ID(ffn_moe_down-78): 202 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 384 us MUL_MAT_ID(ffn_moe_down-79): 205 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 358 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 418 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 350 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 422 us MUL_MAT_ID(ffn_moe_down-83): 204 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 366 us MUL_MAT_ID(ffn_moe_down-84): 201 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 421 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 395 us MUL_MAT_ID(ffn_moe_down-86): 387 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 517 us MUL_MAT_ID(ffn_moe_down-87): 206 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 369 us MUL_MAT_ID(ffn_moe_down-88): 186 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 371 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 463 us MUL_MAT_ID(ffn_moe_down-90): 258 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 403 us MUL_MAT_ID(ffn_moe_down-91): 702 us ggml_barrier(...): 17 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 395 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 350 us MUL_MAT_ID(ffn_moe_down-26): 218 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 378 us MUL_MAT_ID(ffn_moe_down-27): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 366 us MUL_MAT_ID(ffn_moe_down-28): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 390 us MUL_MAT_ID(ffn_moe_down-29): 208 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 355 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 395 us MUL_MAT_ID(ffn_moe_down-31): 235 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 370 us MUL_MAT_ID(ffn_moe_down-32): 201 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 394 us MUL_MAT_ID(ffn_moe_down-33): 534 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 375 us MUL_MAT_ID(ffn_moe_down-34): 203 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 379 us MUL_MAT_ID(ffn_moe_down-35): 185 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 384 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 416 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 360 us MUL_MAT_ID(ffn_moe_down-38): 185 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 382 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 378 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 366 us MUL_MAT_ID(ffn_moe_down-41): 199 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 376 us MUL_MAT_ID(ffn_moe_down-42): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 381 us MUL_MAT_ID(ffn_moe_down-43): 190 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 363 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 369 us MUL_MAT_ID(ffn_moe_down-45): 200 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 361 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 353 us MUL_MAT_ID(ffn_moe_down-47): 237 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 369 us MUL_MAT_ID(ffn_moe_down-48): 337 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 404 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 379 us MUL_MAT_ID(ffn_moe_down-50): 201 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 370 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 364 us MUL_MAT_ID(ffn_moe_down-52): 274 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 397 us MUL_MAT_ID(ffn_moe_down-53): 200 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 393 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 371 us MUL_MAT_ID(ffn_moe_down-55): 210 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 391 us MUL_MAT_ID(ffn_moe_down-56): 205 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 369 us MUL_MAT_ID(ffn_moe_down-57): 203 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 373 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 397 us MUL_MAT_ID(ffn_moe_down-59): 322 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 377 us MUL_MAT_ID(ffn_moe_down-60): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 351 us MUL_MAT_ID(ffn_moe_down-61): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 391 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 375 us MUL_MAT_ID(ffn_moe_down-63): 198 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 369 us MUL_MAT_ID(ffn_moe_down-64): 189 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 372 us MUL_MAT_ID(ffn_moe_down-65): 235 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 567 us MUL_MAT_ID(ffn_moe_down-66): 202 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 403 us MUL_MAT_ID(ffn_moe_down-67): 383 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 369 us MUL_MAT_ID(ffn_moe_down-68): 472 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 407 us MUL_MAT_ID(ffn_moe_down-69): 206 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 383 us MUL_MAT_ID(ffn_moe_down-70): 205 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 418 us MUL_MAT_ID(ffn_moe_down-71): 219 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 368 us MUL_MAT_ID(ffn_moe_down-72): 189 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 369 us MUL_MAT_ID(ffn_moe_down-73): 210 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 393 us MUL_MAT_ID(ffn_moe_down-74): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 373 us MUL_MAT_ID(ffn_moe_down-75): 211 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 401 us MUL_MAT_ID(ffn_moe_down-76): 201 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 426 us MUL_MAT_ID(ffn_moe_down-77): 259 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 396 us MUL_MAT_ID(ffn_moe_down-78): 190 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 417 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 531 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 371 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 365 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 362 us MUL_MAT_ID(ffn_moe_down-83): 204 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 367 us MUL_MAT_ID(ffn_moe_down-84): 244 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 383 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 405 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 381 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 391 us MUL_MAT_ID(ffn_moe_down-88): 186 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 409 us MUL_MAT_ID(ffn_moe_down-89): 196 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 381 us MUL_MAT_ID(ffn_moe_down-90): 263 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 406 us MUL_MAT_ID(ffn_moe_down-91): 242 us ggml_barrier(...): 12 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 367 us MUL_MAT_ID(ffn_moe_down-25): 176 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 455 us MUL_MAT_ID(ffn_moe_down-26): 296 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 378 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 382 us MUL_MAT_ID(ffn_moe_down-28): 210 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 367 us MUL_MAT_ID(ffn_moe_down-29): 217 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 378 us MUL_MAT_ID(ffn_moe_down-30): 344 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 486 us MUL_MAT_ID(ffn_moe_down-31): 174 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 419 us MUL_MAT_ID(ffn_moe_down-32): 256 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 460 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 371 us MUL_MAT_ID(ffn_moe_down-34): 209 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 409 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 402 us MUL_MAT_ID(ffn_moe_down-36): 199 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 404 us MUL_MAT_ID(ffn_moe_down-37): 258 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 411 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 367 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 395 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 387 us MUL_MAT_ID(ffn_moe_down-41): 205 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 426 us MUL_MAT_ID(ffn_moe_down-42): 195 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 360 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 403 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 377 us MUL_MAT_ID(ffn_moe_down-45): 190 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 353 us MUL_MAT_ID(ffn_moe_down-46): 538 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 378 us MUL_MAT_ID(ffn_moe_down-47): 360 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 374 us MUL_MAT_ID(ffn_moe_down-48): 197 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 422 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 387 us MUL_MAT_ID(ffn_moe_down-50): 393 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 351 us MUL_MAT_ID(ffn_moe_down-51): 187 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 377 us MUL_MAT_ID(ffn_moe_down-52): 187 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 380 us MUL_MAT_ID(ffn_moe_down-53): 185 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 355 us MUL_MAT_ID(ffn_moe_down-54): 183 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 413 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 413 us MUL_MAT_ID(ffn_moe_down-57): 331 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 420 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 382 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 348 us MUL_MAT_ID(ffn_moe_down-60): 193 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 406 us MUL_MAT_ID(ffn_moe_down-61): 200 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 343 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 354 us MUL_MAT_ID(ffn_moe_down-63): 180 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 416 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 380 us MUL_MAT_ID(ffn_moe_down-65): 203 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 400 us MUL_MAT_ID(ffn_moe_down-66): 186 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 415 us MUL_MAT_ID(ffn_moe_down-67): 207 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 357 us MUL_MAT_ID(ffn_moe_down-68): 328 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 378 us MUL_MAT_ID(ffn_moe_down-69): 215 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 377 us MUL_MAT_ID(ffn_moe_down-70): 188 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 424 us MUL_MAT_ID(ffn_moe_down-71): 275 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 392 us MUL_MAT_ID(ffn_moe_down-72): 515 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 396 us MUL_MAT_ID(ffn_moe_down-73): 232 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 360 us MUL_MAT_ID(ffn_moe_down-74): 235 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 382 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 373 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 390 us MUL_MAT_ID(ffn_moe_down-77): 220 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 371 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 334 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 366 us MUL_MAT_ID(ffn_moe_down-80): 187 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 413 us MUL_MAT_ID(ffn_moe_down-81): 208 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 398 us MUL_MAT_ID(ffn_moe_down-82): 199 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 389 us MUL_MAT_ID(ffn_moe_down-83): 332 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 395 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 404 us MUL_MAT_ID(ffn_moe_down-85): 247 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 429 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 374 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 410 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 377 us MUL_MAT_ID(ffn_moe_down-89): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 385 us MUL_MAT_ID(ffn_moe_down-90): 276 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 372 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 75 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 421 us MUL_MAT_ID(ffn_moe_down-25): 183 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 390 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 394 us MUL_MAT_ID(ffn_moe_down-27): 202 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 407 us MUL_MAT_ID(ffn_moe_down-28): 317 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 410 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 362 us MUL_MAT_ID(ffn_moe_down-30): 187 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 349 us MUL_MAT_ID(ffn_moe_down-31): 183 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 373 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 403 us MUL_MAT_ID(ffn_moe_down-33): 412 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 390 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 368 us MUL_MAT_ID(ffn_moe_down-35): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 358 us MUL_MAT_ID(ffn_moe_down-36): 189 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 350 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 378 us MUL_MAT_ID(ffn_moe_down-38): 216 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 350 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 370 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 383 us MUL_MAT_ID(ffn_moe_down-41): 207 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 376 us MUL_MAT_ID(ffn_moe_down-42): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 393 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 353 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 379 us MUL_MAT_ID(ffn_moe_down-45): 217 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 400 us MUL_MAT_ID(ffn_moe_down-46): 209 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 364 us MUL_MAT_ID(ffn_moe_down-47): 219 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 433 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 355 us MUL_MAT_ID(ffn_moe_down-49): 450 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 418 us MUL_MAT_ID(ffn_moe_down-50): 519 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 385 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 350 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 382 us MUL_MAT_ID(ffn_moe_down-53): 435 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 364 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 398 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 369 us MUL_MAT_ID(ffn_moe_down-56): 202 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 386 us MUL_MAT_ID(ffn_moe_down-57): 198 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 368 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 369 us MUL_MAT_ID(ffn_moe_down-59): 460 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 382 us MUL_MAT_ID(ffn_moe_down-60): 180 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 414 us MUL_MAT_ID(ffn_moe_down-61): 276 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 394 us MUL_MAT_ID(ffn_moe_down-62): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 357 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 375 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 362 us MUL_MAT_ID(ffn_moe_down-65): 189 us ggml_barrier(...): 318 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 420 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 383 us MUL_MAT_ID(ffn_moe_down-67): 184 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 348 us MUL_MAT_ID(ffn_moe_down-68): 196 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 357 us MUL_MAT_ID(ffn_moe_down-69): 400 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 396 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 417 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 387 us MUL_MAT_ID(ffn_moe_down-72): 187 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 404 us MUL_MAT_ID(ffn_moe_down-73): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 367 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 365 us MUL_MAT_ID(ffn_moe_down-75): 188 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 392 us MUL_MAT_ID(ffn_moe_down-76): 360 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 356 us MUL_MAT_ID(ffn_moe_down-77): 203 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 399 us MUL_MAT_ID(ffn_moe_down-78): 284 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 357 us MUL_MAT_ID(ffn_moe_down-79): 209 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 369 us MUL_MAT_ID(ffn_moe_down-80): 291 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 405 us MUL_MAT_ID(ffn_moe_down-81): 214 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 392 us MUL_MAT_ID(ffn_moe_down-82): 203 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 397 us MUL_MAT_ID(ffn_moe_down-83): 210 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 391 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 372 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 408 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 507 us MUL_MAT_ID(ffn_moe_down-87): 181 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 234 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 411 us MUL_MAT_ID(ffn_moe_down-89): 187 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 418 us MUL_MAT_ID(ffn_moe_down-90): 686 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 400 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 27 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 404 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 398 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 406 us MUL_MAT_ID(ffn_moe_down-27): 200 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 393 us MUL_MAT_ID(ffn_moe_down-28): 185 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 407 us MUL_MAT_ID(ffn_moe_down-29): 197 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 371 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 399 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 391 us MUL_MAT_ID(ffn_moe_down-32): 208 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 392 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 377 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 390 us MUL_MAT_ID(ffn_moe_down-35): 184 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 358 us MUL_MAT_ID(ffn_moe_down-36): 202 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 380 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 397 us MUL_MAT_ID(ffn_moe_down-38): 387 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 398 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 361 us MUL_MAT_ID(ffn_moe_down-40): 227 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 429 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 385 us MUL_MAT_ID(ffn_moe_down-42): 210 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 391 us MUL_MAT_ID(ffn_moe_down-43): 209 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 373 us MUL_MAT_ID(ffn_moe_down-44): 255 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 463 us MUL_MAT_ID(ffn_moe_down-45): 231 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 346 us MUL_MAT_ID(ffn_moe_down-46): 327 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 374 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 377 us MUL_MAT_ID(ffn_moe_down-48): 218 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 352 us MUL_MAT_ID(ffn_moe_down-49): 197 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 385 us MUL_MAT_ID(ffn_moe_down-50): 180 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 365 us MUL_MAT_ID(ffn_moe_down-51): 204 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 362 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 386 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 389 us MUL_MAT_ID(ffn_moe_down-53): 371 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 412 us MUL_MAT_ID(ffn_moe_down-54): 311 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 382 us MUL_MAT_ID(ffn_moe_down-55): 204 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 394 us MUL_MAT_ID(ffn_moe_down-56): 347 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 402 us MUL_MAT_ID(ffn_moe_down-57): 188 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 402 us MUL_MAT_ID(ffn_moe_down-58): 210 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 371 us MUL_MAT_ID(ffn_moe_down-59): 217 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 386 us MUL_MAT_ID(ffn_moe_down-60): 228 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 359 us MUL_MAT_ID(ffn_moe_down-61): 206 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 416 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 461 us MUL_MAT_ID(ffn_moe_down-63): 174 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 390 us MUL_MAT_ID(ffn_moe_down-64): 228 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 382 us MUL_MAT_ID(ffn_moe_down-65): 388 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 425 us MUL_MAT_ID(ffn_moe_down-66): 400 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 393 us MUL_MAT_ID(ffn_moe_down-67): 204 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 404 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 375 us MUL_MAT_ID(ffn_moe_down-69): 190 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 385 us MUL_MAT_ID(ffn_moe_down-70): 374 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 368 us MUL_MAT_ID(ffn_moe_down-71): 392 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 375 us MUL_MAT_ID(ffn_moe_down-72): 252 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 414 us MUL_MAT_ID(ffn_moe_down-73): 299 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 515 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 419 us MUL_MAT_ID(ffn_moe_down-75): 240 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 381 us MUL_MAT_ID(ffn_moe_down-76): 294 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 407 us MUL_MAT_ID(ffn_moe_down-77): 246 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 421 us MUL_MAT_ID(ffn_moe_down-78): 253 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 413 us MUL_MAT_ID(ffn_moe_down-79): 202 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 408 us MUL_MAT_ID(ffn_moe_down-80): 213 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 411 us MUL_MAT_ID(ffn_moe_down-81): 215 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 369 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 396 us MUL_MAT_ID(ffn_moe_down-83): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 361 us MUL_MAT_ID(ffn_moe_down-84): 248 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 380 us MUL_MAT_ID(ffn_moe_down-85): 230 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 377 us MUL_MAT_ID(ffn_moe_down-86): 179 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 418 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 498 us MUL_MAT_ID(ffn_moe_down-88): 206 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 372 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 474 us MUL_MAT_ID(ffn_moe_down-90): 283 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 385 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 41 us GET_ROWS(inp_embd): 5 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 379 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 389 us MUL_MAT_ID(ffn_moe_down-26): 181 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 395 us MUL_MAT_ID(ffn_moe_down-27): 202 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 406 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 405 us MUL_MAT_ID(ffn_moe_down-29): 188 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 353 us MUL_MAT_ID(ffn_moe_down-30): 200 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 415 us MUL_MAT_ID(ffn_moe_down-31): 188 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 400 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 410 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 404 us MUL_MAT_ID(ffn_moe_down-34): 275 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 386 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 384 us MUL_MAT_ID(ffn_moe_down-36): 199 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 377 us MUL_MAT_ID(ffn_moe_down-37): 250 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 380 us MUL_MAT_ID(ffn_moe_down-38): 337 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 428 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 512 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 346 us MUL_MAT_ID(ffn_moe_down-41): 189 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 349 us MUL_MAT_ID(ffn_moe_down-42): 241 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 381 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 399 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 361 us MUL_MAT_ID(ffn_moe_down-45): 203 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 375 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 331 us MUL_MAT_ID(ffn_moe_down-47): 182 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 365 us MUL_MAT_ID(ffn_moe_down-48): 185 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 201 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 394 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 366 us MUL_MAT_ID(ffn_moe_down-51): 188 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 377 us MUL_MAT_ID(ffn_moe_down-52): 202 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 370 us MUL_MAT_ID(ffn_moe_down-53): 169 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 382 us MUL_MAT_ID(ffn_moe_down-54): 533 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 374 us MUL_MAT_ID(ffn_moe_down-55): 214 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 403 us MUL_MAT_ID(ffn_moe_down-56): 245 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 413 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 408 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 387 us MUL_MAT_ID(ffn_moe_down-59): 187 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 388 us MUL_MAT_ID(ffn_moe_down-60): 304 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 405 us MUL_MAT_ID(ffn_moe_down-61): 225 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 377 us MUL_MAT_ID(ffn_moe_down-62): 507 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 386 us MUL_MAT_ID(ffn_moe_down-63): 215 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 408 us MUL_MAT_ID(ffn_moe_down-64): 221 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 347 us MUL_MAT_ID(ffn_moe_down-65): 293 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 187 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 368 us MUL_MAT_ID(ffn_moe_down-67): 216 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 360 us MUL_MAT_ID(ffn_moe_down-68): 352 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 420 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 368 us MUL_MAT_ID(ffn_moe_down-70): 182 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 363 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 402 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 390 us MUL_MAT_ID(ffn_moe_down-73): 205 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 461 us MUL_MAT_ID(ffn_moe_down-74): 431 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 374 us MUL_MAT_ID(ffn_moe_down-75): 432 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 399 us MUL_MAT_ID(ffn_moe_down-76): 382 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 381 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 405 us MUL_MAT_ID(ffn_moe_down-78): 362 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 398 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 407 us MUL_MAT_ID(ffn_moe_down-80): 191 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 406 us MUL_MAT_ID(ffn_moe_down-81): 221 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 370 us MUL_MAT_ID(ffn_moe_down-82): 209 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 380 us MUL_MAT_ID(ffn_moe_down-83): 389 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 404 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 378 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 396 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 375 us MUL_MAT_ID(ffn_moe_down-87): 200 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 395 us MUL_MAT_ID(ffn_moe_down-88): 455 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 387 us MUL_MAT_ID(ffn_moe_down-89): 196 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 436 us MUL_MAT_ID(ffn_moe_down-90): 753 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 364 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 52 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 409 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 402 us MUL_MAT_ID(ffn_moe_down-26): 188 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 402 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 401 us MUL_MAT_ID(ffn_moe_down-28): 239 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 372 us MUL_MAT_ID(ffn_moe_down-29): 200 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 399 us MUL_MAT_ID(ffn_moe_down-30): 217 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 409 us MUL_MAT_ID(ffn_moe_down-31): 218 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 394 us MUL_MAT_ID(ffn_moe_down-32): 248 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 406 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 421 us MUL_MAT_ID(ffn_moe_down-34): 187 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 382 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 407 us MUL_MAT_ID(ffn_moe_down-36): 195 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 368 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 396 us MUL_MAT_ID(ffn_moe_down-38): 197 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 359 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 371 us MUL_MAT_ID(ffn_moe_down-40): 237 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 408 us MUL_MAT_ID(ffn_moe_down-41): 203 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 358 us MUL_MAT_ID(ffn_moe_down-42): 540 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 393 us MUL_MAT_ID(ffn_moe_down-43): 415 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 384 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 396 us MUL_MAT_ID(ffn_moe_down-45): 234 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 383 us MUL_MAT_ID(ffn_moe_down-46): 186 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 389 us MUL_MAT_ID(ffn_moe_down-47): 503 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 411 us MUL_MAT_ID(ffn_moe_down-48): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 389 us MUL_MAT_ID(ffn_moe_down-49): 214 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 408 us MUL_MAT_ID(ffn_moe_down-50): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 385 us MUL_MAT_ID(ffn_moe_down-51): 202 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 409 us MUL_MAT_ID(ffn_moe_down-52): 186 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 411 us MUL_MAT_ID(ffn_moe_down-53): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 404 us MUL_MAT_ID(ffn_moe_down-54): 232 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 426 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 398 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 390 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 404 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 369 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 389 us MUL_MAT_ID(ffn_moe_down-60): 195 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 417 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 389 us MUL_MAT_ID(ffn_moe_down-62): 242 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 361 us MUL_MAT_ID(ffn_moe_down-63): 182 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 383 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 368 us MUL_MAT_ID(ffn_moe_down-65): 187 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 394 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 409 us MUL_MAT_ID(ffn_moe_down-67): 185 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 393 us MUL_MAT_ID(ffn_moe_down-68): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 395 us MUL_MAT_ID(ffn_moe_down-69): 255 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 394 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 388 us MUL_MAT_ID(ffn_moe_down-71): 197 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 410 us MUL_MAT_ID(ffn_moe_down-72): 186 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 373 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 385 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 343 us MUL_MAT_ID(ffn_moe_down-75): 206 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 374 us MUL_MAT_ID(ffn_moe_down-76): 503 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 359 us MUL_MAT_ID(ffn_moe_down-77): 217 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 400 us MUL_MAT_ID(ffn_moe_down-78): 182 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 371 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 377 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 382 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 364 us MUL_MAT_ID(ffn_moe_down-82): 493 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 358 us MUL_MAT_ID(ffn_moe_down-83): 212 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 381 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 384 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 421 us MUL_MAT_ID(ffn_moe_down-86): 185 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 400 us MUL_MAT_ID(ffn_moe_down-87): 214 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 383 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 371 us MUL_MAT_ID(ffn_moe_down-89): 185 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 448 us MUL_MAT_ID(ffn_moe_down-90): 625 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 380 us MUL_MAT_ID(ffn_moe_down-91): 327 us ggml_barrier(...): 35 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 412 us MUL_MAT_ID(ffn_moe_down-25): 182 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 402 us MUL_MAT_ID(ffn_moe_down-26): 186 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 393 us MUL_MAT_ID(ffn_moe_down-27): 191 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 407 us MUL_MAT_ID(ffn_moe_down-28): 184 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 415 us MUL_MAT_ID(ffn_moe_down-29): 194 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 415 us MUL_MAT_ID(ffn_moe_down-30): 204 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 434 us MUL_MAT_ID(ffn_moe_down-31): 183 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 389 us MUL_MAT_ID(ffn_moe_down-32): 259 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 396 us MUL_MAT_ID(ffn_moe_down-33): 265 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 382 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 391 us MUL_MAT_ID(ffn_moe_down-35): 257 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 378 us MUL_MAT_ID(ffn_moe_down-36): 214 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 378 us MUL_MAT_ID(ffn_moe_down-37): 220 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 452 us MUL_MAT_ID(ffn_moe_down-38): 321 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 454 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 384 us MUL_MAT_ID(ffn_moe_down-40): 239 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 381 us MUL_MAT_ID(ffn_moe_down-41): 212 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 368 us MUL_MAT_ID(ffn_moe_down-42): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 432 us MUL_MAT_ID(ffn_moe_down-43): 280 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 385 us MUL_MAT_ID(ffn_moe_down-44): 163 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 396 us MUL_MAT_ID(ffn_moe_down-45): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 380 us MUL_MAT_ID(ffn_moe_down-46): 207 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 409 us MUL_MAT_ID(ffn_moe_down-47): 201 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 348 us MUL_MAT_ID(ffn_moe_down-48): 223 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 409 us MUL_MAT_ID(ffn_moe_down-49): 182 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 470 us MUL_MAT_ID(ffn_moe_down-50): 204 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 352 us MUL_MAT_ID(ffn_moe_down-51): 189 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 395 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 442 us MUL_MAT_ID(ffn_moe_down-53): 176 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 384 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 372 us MUL_MAT_ID(ffn_moe_down-55): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 384 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 387 us MUL_MAT_ID(ffn_moe_down-57): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 461 us MUL_MAT_ID(ffn_moe_down-58): 273 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 428 us MUL_MAT_ID(ffn_moe_down-59): 213 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 382 us MUL_MAT_ID(ffn_moe_down-60): 208 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 401 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 389 us MUL_MAT_ID(ffn_moe_down-62): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 382 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 386 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 441 us MUL_MAT_ID(ffn_moe_down-65): 468 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 392 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 386 us MUL_MAT_ID(ffn_moe_down-67): 211 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 373 us MUL_MAT_ID(ffn_moe_down-68): 183 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 382 us MUL_MAT_ID(ffn_moe_down-69): 194 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 405 us MUL_MAT_ID(ffn_moe_down-70): 188 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 397 us MUL_MAT_ID(ffn_moe_down-71): 209 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 339 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 513 us MUL_MAT_ID(ffn_moe_down-73): 159 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 401 us MUL_MAT_ID(ffn_moe_down-74): 189 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 382 us MUL_MAT_ID(ffn_moe_down-75): 191 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 413 us MUL_MAT_ID(ffn_moe_down-76): 224 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 364 us MUL_MAT_ID(ffn_moe_down-77): 398 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 377 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 354 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 367 us MUL_MAT_ID(ffn_moe_down-80): 247 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 359 us MUL_MAT_ID(ffn_moe_down-81): 391 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 375 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 439 us MUL_MAT_ID(ffn_moe_down-83): 449 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 459 us MUL_MAT_ID(ffn_moe_down-84): 299 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 394 us MUL_MAT_ID(ffn_moe_down-85): 691 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 376 us MUL_MAT_ID(ffn_moe_down-86): 209 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 341 us MUL_MAT_ID(ffn_moe_down-87): 224 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 354 us MUL_MAT_ID(ffn_moe_down-88): 212 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 380 us MUL_MAT_ID(ffn_moe_down-89): 202 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 419 us MUL_MAT_ID(ffn_moe_down-90): 230 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 402 us MUL_MAT_ID(ffn_moe_down-91): 241 us ggml_barrier(...): 15 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 410 us MUL_MAT_ID(ffn_moe_down-25): 373 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 385 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 380 us MUL_MAT_ID(ffn_moe_down-27): 222 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 395 us MUL_MAT_ID(ffn_moe_down-28): 260 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 362 us MUL_MAT_ID(ffn_moe_down-29): 351 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 405 us MUL_MAT_ID(ffn_moe_down-30): 217 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 375 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 372 us MUL_MAT_ID(ffn_moe_down-32): 383 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 398 us MUL_MAT_ID(ffn_moe_down-33): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 479 us MUL_MAT_ID(ffn_moe_down-34): 368 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 386 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 393 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 378 us MUL_MAT_ID(ffn_moe_down-37): 267 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 411 us MUL_MAT_ID(ffn_moe_down-38): 418 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 406 us MUL_MAT_ID(ffn_moe_down-39): 223 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 379 us MUL_MAT_ID(ffn_moe_down-40): 241 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 386 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 490 us MUL_MAT_ID(ffn_moe_down-42): 221 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 381 us MUL_MAT_ID(ffn_moe_down-43): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 522 us MUL_MAT_ID(ffn_moe_down-44): 232 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 443 us MUL_MAT_ID(ffn_moe_down-45): 357 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 355 us MUL_MAT_ID(ffn_moe_down-46): 369 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 378 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 444 us MUL_MAT_ID(ffn_moe_down-48): 218 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 357 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 406 us MUL_MAT_ID(ffn_moe_down-50): 236 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 405 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 470 us MUL_MAT_ID(ffn_moe_down-52): 180 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 393 us MUL_MAT_ID(ffn_moe_down-53): 206 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 354 us MUL_MAT_ID(ffn_moe_down-54): 505 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 399 us MUL_MAT_ID(ffn_moe_down-55): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 422 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 359 us MUL_MAT_ID(ffn_moe_down-57): 346 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 392 us MUL_MAT_ID(ffn_moe_down-58): 215 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 370 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 363 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 378 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 382 us MUL_MAT_ID(ffn_moe_down-62): 271 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 422 us MUL_MAT_ID(ffn_moe_down-63): 168 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 406 us MUL_MAT_ID(ffn_moe_down-64): 205 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 392 us MUL_MAT_ID(ffn_moe_down-65): 216 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 352 us MUL_MAT_ID(ffn_moe_down-66): 345 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 377 us MUL_MAT_ID(ffn_moe_down-67): 517 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 379 us MUL_MAT_ID(ffn_moe_down-68): 190 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 349 us MUL_MAT_ID(ffn_moe_down-69): 182 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 365 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 478 us MUL_MAT_ID(ffn_moe_down-71): 471 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 212 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 435 us MUL_MAT_ID(ffn_moe_down-73): 284 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 382 us MUL_MAT_ID(ffn_moe_down-74): 349 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 398 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 389 us MUL_MAT_ID(ffn_moe_down-76): 247 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 495 us MUL_MAT_ID(ffn_moe_down-77): 420 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 375 us MUL_MAT_ID(ffn_moe_down-78): 286 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 398 us MUL_MAT_ID(ffn_moe_down-79): 304 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 458 us MUL_MAT_ID(ffn_moe_down-80): 294 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 389 us MUL_MAT_ID(ffn_moe_down-81): 185 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 358 us MUL_MAT_ID(ffn_moe_down-82): 206 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 401 us MUL_MAT_ID(ffn_moe_down-83): 382 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 349 us MUL_MAT_ID(ffn_moe_down-84): 663 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 400 us MUL_MAT_ID(ffn_moe_down-85): 234 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 367 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 434 us MUL_MAT_ID(ffn_moe_down-87): 320 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 387 us MUL_MAT_ID(ffn_moe_down-88): 189 us ggml_barrier(...): 477 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 332 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 406 us MUL_MAT_ID(ffn_moe_down-90): 238 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 358 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 58 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 391 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 386 us MUL_MAT_ID(ffn_moe_down-26): 210 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 397 us MUL_MAT_ID(ffn_moe_down-27): 244 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 375 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 373 us MUL_MAT_ID(ffn_moe_down-29): 370 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 417 us MUL_MAT_ID(ffn_moe_down-30): 197 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 359 us MUL_MAT_ID(ffn_moe_down-31): 363 us ggml_barrier(...): 344 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 369 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 379 us MUL_MAT_ID(ffn_moe_down-33): 160 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 418 us MUL_MAT_ID(ffn_moe_down-34): 227 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 374 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 379 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 396 us MUL_MAT_ID(ffn_moe_down-37): 502 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 361 us MUL_MAT_ID(ffn_moe_down-38): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 376 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 409 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 369 us MUL_MAT_ID(ffn_moe_down-41): 263 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 376 us MUL_MAT_ID(ffn_moe_down-42): 246 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 355 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 383 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 378 us MUL_MAT_ID(ffn_moe_down-45): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 465 us MUL_MAT_ID(ffn_moe_down-46): 282 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 360 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 413 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 369 us MUL_MAT_ID(ffn_moe_down-49): 349 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 492 us MUL_MAT_ID(ffn_moe_down-50): 214 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 400 us MUL_MAT_ID(ffn_moe_down-51): 180 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 386 us MUL_MAT_ID(ffn_moe_down-52): 176 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 400 us MUL_MAT_ID(ffn_moe_down-53): 183 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 352 us MUL_MAT_ID(ffn_moe_down-54): 189 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 364 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 370 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 386 us MUL_MAT_ID(ffn_moe_down-57): 181 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 526 us MUL_MAT_ID(ffn_moe_down-58): 417 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 364 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 376 us MUL_MAT_ID(ffn_moe_down-60): 181 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 354 us MUL_MAT_ID(ffn_moe_down-61): 215 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 344 us MUL_MAT_ID(ffn_moe_down-62): 182 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 353 us MUL_MAT_ID(ffn_moe_down-63): 208 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 395 us MUL_MAT_ID(ffn_moe_down-64): 373 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 411 us MUL_MAT_ID(ffn_moe_down-65): 205 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 400 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 408 us MUL_MAT_ID(ffn_moe_down-67): 379 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 389 us MUL_MAT_ID(ffn_moe_down-68): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 423 us MUL_MAT_ID(ffn_moe_down-69): 232 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 392 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 401 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 471 us MUL_MAT_ID(ffn_moe_down-72): 172 us ggml_barrier(...): 384 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 390 us MUL_MAT_ID(ffn_moe_down-73): 222 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 400 us MUL_MAT_ID(ffn_moe_down-74): 190 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 340 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 400 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 387 us MUL_MAT_ID(ffn_moe_down-77): 197 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 406 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 369 us MUL_MAT_ID(ffn_moe_down-79): 207 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 406 us MUL_MAT_ID(ffn_moe_down-80): 208 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 384 us MUL_MAT_ID(ffn_moe_down-81): 213 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 369 us MUL_MAT_ID(ffn_moe_down-82): 200 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 355 us MUL_MAT_ID(ffn_moe_down-83): 213 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 359 us MUL_MAT_ID(ffn_moe_down-84): 607 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 515 us MUL_MAT_ID(ffn_moe_down-85): 192 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 404 us MUL_MAT_ID(ffn_moe_down-86): 553 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 454 us MUL_MAT_ID(ffn_moe_down-88): 290 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 368 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 392 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 353 us MUL_MAT_ID(ffn_moe_down-91): 233 us ggml_barrier(...): 57 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 408 us MUL_MAT_ID(ffn_moe_down-25): 378 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 395 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 377 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 436 us MUL_MAT_ID(ffn_moe_down-28): 202 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 362 us MUL_MAT_ID(ffn_moe_down-30): 190 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 411 us MUL_MAT_ID(ffn_moe_down-31): 222 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 398 us MUL_MAT_ID(ffn_moe_down-32): 307 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 350 us MUL_MAT_ID(ffn_moe_down-33): 188 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 391 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 366 us MUL_MAT_ID(ffn_moe_down-35): 184 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 397 us MUL_MAT_ID(ffn_moe_down-36): 212 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 388 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 384 us MUL_MAT_ID(ffn_moe_down-38): 373 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 394 us MUL_MAT_ID(ffn_moe_down-39): 345 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 395 us MUL_MAT_ID(ffn_moe_down-40): 643 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 368 us MUL_MAT_ID(ffn_moe_down-41): 194 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 376 us MUL_MAT_ID(ffn_moe_down-42): 213 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 358 us MUL_MAT_ID(ffn_moe_down-43): 208 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 340 us MUL_MAT_ID(ffn_moe_down-44): 200 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 389 us MUL_MAT_ID(ffn_moe_down-45): 212 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 347 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 351 us MUL_MAT_ID(ffn_moe_down-47): 185 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 378 us MUL_MAT_ID(ffn_moe_down-48): 204 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 351 us MUL_MAT_ID(ffn_moe_down-49): 224 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 386 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 347 us MUL_MAT_ID(ffn_moe_down-51): 215 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 356 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 373 us MUL_MAT_ID(ffn_moe_down-53): 192 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 394 us MUL_MAT_ID(ffn_moe_down-54): 338 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 374 us MUL_MAT_ID(ffn_moe_down-55): 189 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 386 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 359 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 402 us MUL_MAT_ID(ffn_moe_down-58): 186 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 408 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 361 us MUL_MAT_ID(ffn_moe_down-60): 193 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 390 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 357 us MUL_MAT_ID(ffn_moe_down-62): 216 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 364 us MUL_MAT_ID(ffn_moe_down-63): 182 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 387 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 368 us MUL_MAT_ID(ffn_moe_down-65): 181 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 360 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 400 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 394 us MUL_MAT_ID(ffn_moe_down-68): 341 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 395 us MUL_MAT_ID(ffn_moe_down-69): 416 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 397 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 399 us MUL_MAT_ID(ffn_moe_down-71): 208 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 218 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 360 us MUL_MAT_ID(ffn_moe_down-73): 200 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 401 us MUL_MAT_ID(ffn_moe_down-74): 209 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 352 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 395 us MUL_MAT_ID(ffn_moe_down-76): 204 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 399 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 362 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 384 us MUL_MAT_ID(ffn_moe_down-79): 210 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 362 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 404 us MUL_MAT_ID(ffn_moe_down-81): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 424 us MUL_MAT_ID(ffn_moe_down-82): 192 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 390 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 367 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 373 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 439 us MUL_MAT_ID(ffn_moe_down-86): 169 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 410 us MUL_MAT_ID(ffn_moe_down-87): 183 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 399 us MUL_MAT_ID(ffn_moe_down-88): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 367 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 359 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 393 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 48 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 400 us MUL_MAT_ID(ffn_moe_down-25): 197 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 399 us MUL_MAT_ID(ffn_moe_down-26): 199 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 394 us MUL_MAT_ID(ffn_moe_down-27): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 382 us MUL_MAT_ID(ffn_moe_down-28): 187 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 205 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 376 us MUL_MAT_ID(ffn_moe_down-30): 246 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 400 us MUL_MAT_ID(ffn_moe_down-31): 383 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 368 us MUL_MAT_ID(ffn_moe_down-32): 302 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 395 us MUL_MAT_ID(ffn_moe_down-33): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 357 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 392 us MUL_MAT_ID(ffn_moe_down-35): 204 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 400 us MUL_MAT_ID(ffn_moe_down-36): 205 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 388 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 388 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 373 us MUL_MAT_ID(ffn_moe_down-39): 221 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 404 us MUL_MAT_ID(ffn_moe_down-40): 225 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 350 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 360 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 386 us MUL_MAT_ID(ffn_moe_down-43): 187 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 345 us MUL_MAT_ID(ffn_moe_down-44): 157 us ggml_barrier(...): 342 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 352 us MUL_MAT_ID(ffn_moe_down-45): 218 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 354 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 375 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 378 us MUL_MAT_ID(ffn_moe_down-48): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 402 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 452 us MUL_MAT_ID(ffn_moe_down-50): 290 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 355 us MUL_MAT_ID(ffn_moe_down-51): 521 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 404 us MUL_MAT_ID(ffn_moe_down-52): 184 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 374 us MUL_MAT_ID(ffn_moe_down-53): 215 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 382 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 370 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 376 us MUL_MAT_ID(ffn_moe_down-56): 214 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 359 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 341 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 355 us MUL_MAT_ID(ffn_moe_down-59): 292 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 382 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 400 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 361 us MUL_MAT_ID(ffn_moe_down-62): 209 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 391 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 356 us MUL_MAT_ID(ffn_moe_down-64): 200 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 351 us MUL_MAT_ID(ffn_moe_down-65): 239 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 384 us MUL_MAT_ID(ffn_moe_down-66): 374 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 390 us MUL_MAT_ID(ffn_moe_down-67): 342 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 517 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 364 us MUL_MAT_ID(ffn_moe_down-69): 217 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 362 us MUL_MAT_ID(ffn_moe_down-70): 304 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 490 us MUL_MAT_ID(ffn_moe_down-71): 176 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 359 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 396 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 348 us MUL_MAT_ID(ffn_moe_down-74): 419 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 398 us MUL_MAT_ID(ffn_moe_down-75): 213 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 395 us MUL_MAT_ID(ffn_moe_down-76): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 373 us MUL_MAT_ID(ffn_moe_down-77): 225 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 354 us MUL_MAT_ID(ffn_moe_down-78): 188 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 403 us MUL_MAT_ID(ffn_moe_down-79): 207 us ggml_barrier(...): 401 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 359 us MUL_MAT_ID(ffn_moe_down-80): 177 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 385 us MUL_MAT_ID(ffn_moe_down-81): 187 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 433 us MUL_MAT_ID(ffn_moe_down-82): 461 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 405 us MUL_MAT_ID(ffn_moe_down-83): 201 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 356 us MUL_MAT_ID(ffn_moe_down-84): 248 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 393 us MUL_MAT_ID(ffn_moe_down-85): 240 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 374 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 389 us MUL_MAT_ID(ffn_moe_down-87): 324 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 418 us MUL_MAT_ID(ffn_moe_down-88): 182 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 388 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 398 us MUL_MAT_ID(ffn_moe_down-90): 269 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 380 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 40 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 397 us MUL_MAT_ID(ffn_moe_down-25): 335 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 377 us MUL_MAT_ID(ffn_moe_down-26): 277 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 397 us MUL_MAT_ID(ffn_moe_down-27): 202 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 376 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 413 us MUL_MAT_ID(ffn_moe_down-29): 186 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 386 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 379 us MUL_MAT_ID(ffn_moe_down-31): 372 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 367 us MUL_MAT_ID(ffn_moe_down-32): 187 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 408 us MUL_MAT_ID(ffn_moe_down-33): 185 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 405 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 377 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 409 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 398 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 434 us MUL_MAT_ID(ffn_moe_down-38): 210 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 364 us MUL_MAT_ID(ffn_moe_down-39): 228 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 391 us MUL_MAT_ID(ffn_moe_down-40): 238 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 396 us MUL_MAT_ID(ffn_moe_down-41): 189 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 368 us MUL_MAT_ID(ffn_moe_down-42): 209 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 407 us MUL_MAT_ID(ffn_moe_down-43): 520 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 372 us MUL_MAT_ID(ffn_moe_down-44): 182 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 406 us MUL_MAT_ID(ffn_moe_down-45): 208 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 372 us MUL_MAT_ID(ffn_moe_down-46): 446 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 378 us MUL_MAT_ID(ffn_moe_down-47): 208 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 365 us MUL_MAT_ID(ffn_moe_down-48): 313 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 387 us MUL_MAT_ID(ffn_moe_down-49): 184 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 370 us MUL_MAT_ID(ffn_moe_down-50): 209 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 387 us MUL_MAT_ID(ffn_moe_down-51): 223 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 375 us MUL_MAT_ID(ffn_moe_down-52): 214 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 399 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 352 us MUL_MAT_ID(ffn_moe_down-54): 187 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 398 us MUL_MAT_ID(ffn_moe_down-55): 180 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 390 us MUL_MAT_ID(ffn_moe_down-56): 189 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 400 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 392 us MUL_MAT_ID(ffn_moe_down-58): 219 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 383 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 346 us MUL_MAT_ID(ffn_moe_down-60): 223 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 364 us MUL_MAT_ID(ffn_moe_down-61): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 391 us MUL_MAT_ID(ffn_moe_down-62): 211 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 391 us MUL_MAT_ID(ffn_moe_down-63): 406 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 356 us MUL_MAT_ID(ffn_moe_down-64): 386 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 330 us MUL_MAT_ID(ffn_moe_down-65): 206 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 202 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 353 us MUL_MAT_ID(ffn_moe_down-67): 210 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 354 us MUL_MAT_ID(ffn_moe_down-68): 178 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 397 us MUL_MAT_ID(ffn_moe_down-69): 208 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 378 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 515 us MUL_MAT_ID(ffn_moe_down-71): 209 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 385 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 395 us MUL_MAT_ID(ffn_moe_down-73): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 526 us MUL_MAT_ID(ffn_moe_down-74): 276 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 351 us MUL_MAT_ID(ffn_moe_down-75): 196 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 395 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 437 us MUL_MAT_ID(ffn_moe_down-77): 172 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 381 us MUL_MAT_ID(ffn_moe_down-78): 246 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 381 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 373 us MUL_MAT_ID(ffn_moe_down-80): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 411 us MUL_MAT_ID(ffn_moe_down-81): 205 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 370 us MUL_MAT_ID(ffn_moe_down-82): 190 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 404 us MUL_MAT_ID(ffn_moe_down-83): 205 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 540 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 350 us MUL_MAT_ID(ffn_moe_down-85): 417 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 453 us MUL_MAT_ID(ffn_moe_down-86): 212 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 377 us MUL_MAT_ID(ffn_moe_down-87): 229 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 407 us MUL_MAT_ID(ffn_moe_down-88): 201 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 403 us MUL_MAT_ID(ffn_moe_down-89): 280 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 389 us MUL_MAT_ID(ffn_moe_down-90): 300 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 390 us MUL_MAT_ID(ffn_moe_down-91): 242 us ggml_barrier(...): 81 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 390 us MUL_MAT_ID(ffn_moe_down-25): 245 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 370 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 381 us MUL_MAT_ID(ffn_moe_down-27): 225 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 379 us MUL_MAT_ID(ffn_moe_down-28): 184 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 362 us MUL_MAT_ID(ffn_moe_down-29): 204 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 375 us MUL_MAT_ID(ffn_moe_down-30): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 391 us MUL_MAT_ID(ffn_moe_down-31): 209 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 394 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 344 us MUL_MAT_ID(ffn_moe_down-33): 186 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 399 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 398 us MUL_MAT_ID(ffn_moe_down-35): 229 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 388 us MUL_MAT_ID(ffn_moe_down-36): 191 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 378 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 353 us MUL_MAT_ID(ffn_moe_down-38): 174 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 374 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 362 us MUL_MAT_ID(ffn_moe_down-40): 708 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 402 us MUL_MAT_ID(ffn_moe_down-41): 208 us ggml_barrier(...): 507 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 350 us MUL_MAT_ID(ffn_moe_down-42): 236 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 376 us MUL_MAT_ID(ffn_moe_down-43): 211 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 338 us MUL_MAT_ID(ffn_moe_down-44): 233 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 388 us MUL_MAT_ID(ffn_moe_down-45): 227 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 393 us MUL_MAT_ID(ffn_moe_down-46): 210 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 348 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 381 us MUL_MAT_ID(ffn_moe_down-48): 191 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 390 us MUL_MAT_ID(ffn_moe_down-49): 207 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 400 us MUL_MAT_ID(ffn_moe_down-50): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 382 us MUL_MAT_ID(ffn_moe_down-51): 202 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 375 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 338 us MUL_MAT_ID(ffn_moe_down-53): 379 us ggml_barrier(...): 330 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 351 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 368 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 424 us MUL_MAT_ID(ffn_moe_down-56): 187 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 387 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 383 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 365 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 364 us MUL_MAT_ID(ffn_moe_down-60): 512 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 379 us MUL_MAT_ID(ffn_moe_down-61): 250 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 406 us MUL_MAT_ID(ffn_moe_down-62): 234 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 370 us MUL_MAT_ID(ffn_moe_down-63): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 364 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 378 us MUL_MAT_ID(ffn_moe_down-65): 204 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 489 us MUL_MAT_ID(ffn_moe_down-66): 179 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 392 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 392 us MUL_MAT_ID(ffn_moe_down-68): 256 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 479 us MUL_MAT_ID(ffn_moe_down-69): 375 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 386 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 376 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 380 us MUL_MAT_ID(ffn_moe_down-72): 215 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 373 us MUL_MAT_ID(ffn_moe_down-73): 224 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 372 us MUL_MAT_ID(ffn_moe_down-74): 190 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 382 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 367 us MUL_MAT_ID(ffn_moe_down-76): 252 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 363 us MUL_MAT_ID(ffn_moe_down-77): 346 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 349 us MUL_MAT_ID(ffn_moe_down-78): 195 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 375 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 405 us MUL_MAT_ID(ffn_moe_down-80): 369 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 386 us MUL_MAT_ID(ffn_moe_down-81): 211 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 376 us MUL_MAT_ID(ffn_moe_down-82): 197 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 403 us MUL_MAT_ID(ffn_moe_down-83): 210 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 369 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 404 us MUL_MAT_ID(ffn_moe_down-85): 230 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 375 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 387 us MUL_MAT_ID(ffn_moe_down-87): 210 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 409 us MUL_MAT_ID(ffn_moe_down-88): 169 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 401 us MUL_MAT_ID(ffn_moe_down-89): 202 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 407 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 400 us MUL_MAT_ID(ffn_moe_down-91): 241 us ggml_barrier(...): 44 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 182 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 395 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 353 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 377 us MUL_MAT_ID(ffn_moe_down-28): 202 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 395 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 399 us MUL_MAT_ID(ffn_moe_down-30): 208 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 364 us MUL_MAT_ID(ffn_moe_down-31): 221 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 379 us MUL_MAT_ID(ffn_moe_down-32): 186 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 369 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 405 us MUL_MAT_ID(ffn_moe_down-34): 388 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 369 us MUL_MAT_ID(ffn_moe_down-35): 371 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 382 us MUL_MAT_ID(ffn_moe_down-36): 208 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 360 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 362 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 380 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 391 us MUL_MAT_ID(ffn_moe_down-40): 225 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 349 us MUL_MAT_ID(ffn_moe_down-41): 239 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 407 us MUL_MAT_ID(ffn_moe_down-42): 283 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 367 us MUL_MAT_ID(ffn_moe_down-43): 513 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 391 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 367 us MUL_MAT_ID(ffn_moe_down-45): 370 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 388 us MUL_MAT_ID(ffn_moe_down-46): 185 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 393 us MUL_MAT_ID(ffn_moe_down-47): 225 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 399 us MUL_MAT_ID(ffn_moe_down-48): 190 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 395 us MUL_MAT_ID(ffn_moe_down-49): 318 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 374 us MUL_MAT_ID(ffn_moe_down-50): 211 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 379 us MUL_MAT_ID(ffn_moe_down-51): 189 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 396 us MUL_MAT_ID(ffn_moe_down-52): 206 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 395 us MUL_MAT_ID(ffn_moe_down-53): 501 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 380 us MUL_MAT_ID(ffn_moe_down-54): 215 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 361 us MUL_MAT_ID(ffn_moe_down-55): 183 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 399 us MUL_MAT_ID(ffn_moe_down-56): 238 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 425 us MUL_MAT_ID(ffn_moe_down-57): 222 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 398 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 408 us MUL_MAT_ID(ffn_moe_down-59): 215 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 376 us MUL_MAT_ID(ffn_moe_down-60): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 401 us MUL_MAT_ID(ffn_moe_down-61): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 561 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 388 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 343 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 396 us MUL_MAT_ID(ffn_moe_down-65): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 397 us MUL_MAT_ID(ffn_moe_down-66): 318 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 376 us MUL_MAT_ID(ffn_moe_down-67): 200 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 402 us MUL_MAT_ID(ffn_moe_down-68): 358 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 365 us MUL_MAT_ID(ffn_moe_down-69): 189 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 422 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 454 us MUL_MAT_ID(ffn_moe_down-71): 203 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 400 us MUL_MAT_ID(ffn_moe_down-72): 207 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 406 us MUL_MAT_ID(ffn_moe_down-73): 193 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 501 us MUL_MAT_ID(ffn_moe_down-74): 181 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 369 us MUL_MAT_ID(ffn_moe_down-75): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 349 us MUL_MAT_ID(ffn_moe_down-76): 260 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 405 us MUL_MAT_ID(ffn_moe_down-77): 325 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 396 us MUL_MAT_ID(ffn_moe_down-78): 206 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 402 us MUL_MAT_ID(ffn_moe_down-79): 187 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 405 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 383 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 438 us MUL_MAT_ID(ffn_moe_down-82): 179 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 394 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 343 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 351 us MUL_MAT_ID(ffn_moe_down-85): 476 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 362 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 400 us MUL_MAT_ID(ffn_moe_down-87): 210 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 390 us MUL_MAT_ID(ffn_moe_down-88): 351 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 429 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 368 us MUL_MAT_ID(ffn_moe_down-90): 253 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 361 us MUL_MAT_ID(ffn_moe_down-91): 280 us ggml_barrier(...): 187 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 262 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 506 us MUL_MAT_ID(ffn_moe_down-26): 178 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 406 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 364 us MUL_MAT_ID(ffn_moe_down-28): 199 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 404 us MUL_MAT_ID(ffn_moe_down-29): 409 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 403 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 401 us MUL_MAT_ID(ffn_moe_down-31): 220 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 387 us MUL_MAT_ID(ffn_moe_down-32): 402 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 357 us MUL_MAT_ID(ffn_moe_down-33): 209 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 366 us MUL_MAT_ID(ffn_moe_down-34): 329 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 362 us MUL_MAT_ID(ffn_moe_down-35): 546 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 359 us MUL_MAT_ID(ffn_moe_down-36): 270 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 390 us MUL_MAT_ID(ffn_moe_down-37): 222 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 376 us MUL_MAT_ID(ffn_moe_down-39): 258 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 415 us MUL_MAT_ID(ffn_moe_down-40): 301 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 365 us MUL_MAT_ID(ffn_moe_down-41): 183 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 402 us MUL_MAT_ID(ffn_moe_down-42): 291 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 404 us MUL_MAT_ID(ffn_moe_down-43): 202 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 382 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 388 us MUL_MAT_ID(ffn_moe_down-45): 211 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 369 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 381 us MUL_MAT_ID(ffn_moe_down-47): 186 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 396 us MUL_MAT_ID(ffn_moe_down-48): 215 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 368 us MUL_MAT_ID(ffn_moe_down-49): 224 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 397 us MUL_MAT_ID(ffn_moe_down-50): 515 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 374 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 370 us MUL_MAT_ID(ffn_moe_down-52): 203 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 354 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 369 us MUL_MAT_ID(ffn_moe_down-54): 206 us ggml_barrier(...): 373 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 399 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 477 us MUL_MAT_ID(ffn_moe_down-56): 186 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 421 us MUL_MAT_ID(ffn_moe_down-57): 185 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 396 us MUL_MAT_ID(ffn_moe_down-58): 210 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 361 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 417 us MUL_MAT_ID(ffn_moe_down-60): 544 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 423 us MUL_MAT_ID(ffn_moe_down-61): 412 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 376 us MUL_MAT_ID(ffn_moe_down-62): 187 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 407 us MUL_MAT_ID(ffn_moe_down-63): 230 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 374 us MUL_MAT_ID(ffn_moe_down-64): 204 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 350 us MUL_MAT_ID(ffn_moe_down-65): 198 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 365 us MUL_MAT_ID(ffn_moe_down-66): 186 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 384 us MUL_MAT_ID(ffn_moe_down-67): 209 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 366 us MUL_MAT_ID(ffn_moe_down-68): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 406 us MUL_MAT_ID(ffn_moe_down-69): 215 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 383 us MUL_MAT_ID(ffn_moe_down-70): 344 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 384 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 355 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 386 us MUL_MAT_ID(ffn_moe_down-73): 236 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 421 us MUL_MAT_ID(ffn_moe_down-74): 189 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 468 us MUL_MAT_ID(ffn_moe_down-75): 368 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 365 us MUL_MAT_ID(ffn_moe_down-76): 412 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 383 us MUL_MAT_ID(ffn_moe_down-77): 375 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 392 us MUL_MAT_ID(ffn_moe_down-78): 189 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 390 us MUL_MAT_ID(ffn_moe_down-79): 166 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 363 us MUL_MAT_ID(ffn_moe_down-80): 208 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 522 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 386 us MUL_MAT_ID(ffn_moe_down-82): 192 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 396 us MUL_MAT_ID(ffn_moe_down-83): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 365 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 373 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 390 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 392 us MUL_MAT_ID(ffn_moe_down-87): 197 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 364 us MUL_MAT_ID(ffn_moe_down-88): 337 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 356 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 403 us MUL_MAT_ID(ffn_moe_down-90): 236 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 530 us MUL_MAT_ID(ffn_moe_down-91): 276 us ggml_barrier(...): 58 us GET_ROWS(inp_embd): 4 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 404 us MUL_MAT_ID(ffn_moe_down-25): 281 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 367 us MUL_MAT_ID(ffn_moe_down-26): 178 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 376 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 403 us MUL_MAT_ID(ffn_moe_down-28): 229 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 365 us MUL_MAT_ID(ffn_moe_down-29): 179 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 377 us MUL_MAT_ID(ffn_moe_down-30): 223 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 414 us MUL_MAT_ID(ffn_moe_down-31): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 374 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 349 us MUL_MAT_ID(ffn_moe_down-33): 154 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 409 us MUL_MAT_ID(ffn_moe_down-34): 202 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 389 us MUL_MAT_ID(ffn_moe_down-35): 349 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 342 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 368 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 389 us MUL_MAT_ID(ffn_moe_down-38): 218 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 435 us MUL_MAT_ID(ffn_moe_down-39): 407 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 351 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 370 us MUL_MAT_ID(ffn_moe_down-41): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 382 us MUL_MAT_ID(ffn_moe_down-42): 187 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 383 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 467 us MUL_MAT_ID(ffn_moe_down-44): 298 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 376 us MUL_MAT_ID(ffn_moe_down-45): 224 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 402 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 417 us MUL_MAT_ID(ffn_moe_down-47): 204 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 386 us MUL_MAT_ID(ffn_moe_down-48): 209 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 364 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 400 us MUL_MAT_ID(ffn_moe_down-50): 210 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 408 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 361 us MUL_MAT_ID(ffn_moe_down-52): 203 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 460 us MUL_MAT_ID(ffn_moe_down-53): 296 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 373 us MUL_MAT_ID(ffn_moe_down-54): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 382 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 379 us MUL_MAT_ID(ffn_moe_down-57): 203 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 375 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 383 us MUL_MAT_ID(ffn_moe_down-59): 363 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 423 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 380 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 394 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 382 us MUL_MAT_ID(ffn_moe_down-63): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 404 us MUL_MAT_ID(ffn_moe_down-64): 207 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 362 us MUL_MAT_ID(ffn_moe_down-65): 188 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 396 us MUL_MAT_ID(ffn_moe_down-66): 229 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 384 us MUL_MAT_ID(ffn_moe_down-67): 206 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 393 us MUL_MAT_ID(ffn_moe_down-68): 507 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 382 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 384 us MUL_MAT_ID(ffn_moe_down-70): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 391 us MUL_MAT_ID(ffn_moe_down-71): 191 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 382 us MUL_MAT_ID(ffn_moe_down-72): 238 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 406 us MUL_MAT_ID(ffn_moe_down-73): 490 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 386 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 375 us MUL_MAT_ID(ffn_moe_down-75): 226 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 369 us MUL_MAT_ID(ffn_moe_down-76): 181 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 377 us MUL_MAT_ID(ffn_moe_down-77): 205 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 365 us MUL_MAT_ID(ffn_moe_down-78): 556 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 397 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 365 us MUL_MAT_ID(ffn_moe_down-80): 388 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 382 us MUL_MAT_ID(ffn_moe_down-81): 206 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 348 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 414 us MUL_MAT_ID(ffn_moe_down-83): 202 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 385 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 382 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 372 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 378 us MUL_MAT_ID(ffn_moe_down-87): 210 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 414 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 401 us MUL_MAT_ID(ffn_moe_down-89): 215 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 363 us MUL_MAT_ID(ffn_moe_down-90): 662 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 473 us MUL_MAT_ID(ffn_moe_down-91): 192 us ggml_barrier(...): 151 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 378 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 407 us MUL_MAT_ID(ffn_moe_down-27): 191 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 367 us MUL_MAT_ID(ffn_moe_down-28): 186 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 348 us MUL_MAT_ID(ffn_moe_down-29): 291 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 425 us MUL_MAT_ID(ffn_moe_down-30): 203 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 393 us MUL_MAT_ID(ffn_moe_down-31): 188 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 426 us MUL_MAT_ID(ffn_moe_down-32): 205 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 416 us MUL_MAT_ID(ffn_moe_down-33): 373 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 483 us MUL_MAT_ID(ffn_moe_down-34): 285 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 351 us MUL_MAT_ID(ffn_moe_down-35): 211 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 395 us MUL_MAT_ID(ffn_moe_down-36): 211 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 366 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 424 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 399 us MUL_MAT_ID(ffn_moe_down-39): 372 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 393 us MUL_MAT_ID(ffn_moe_down-40): 228 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 381 us MUL_MAT_ID(ffn_moe_down-41): 204 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 412 us MUL_MAT_ID(ffn_moe_down-42): 183 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 406 us MUL_MAT_ID(ffn_moe_down-43): 220 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 390 us MUL_MAT_ID(ffn_moe_down-44): 289 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 355 us MUL_MAT_ID(ffn_moe_down-45): 209 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 425 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 418 us MUL_MAT_ID(ffn_moe_down-47): 186 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 371 us MUL_MAT_ID(ffn_moe_down-48): 219 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 439 us MUL_MAT_ID(ffn_moe_down-49): 173 us ggml_barrier(...): 523 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 396 us MUL_MAT_ID(ffn_moe_down-50): 216 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 388 us MUL_MAT_ID(ffn_moe_down-51): 378 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 380 us MUL_MAT_ID(ffn_moe_down-52): 180 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 387 us MUL_MAT_ID(ffn_moe_down-53): 268 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 389 us MUL_MAT_ID(ffn_moe_down-54): 212 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 357 us MUL_MAT_ID(ffn_moe_down-55): 187 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 371 us MUL_MAT_ID(ffn_moe_down-56): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 385 us MUL_MAT_ID(ffn_moe_down-57): 235 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 418 us MUL_MAT_ID(ffn_moe_down-58): 204 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 402 us MUL_MAT_ID(ffn_moe_down-59): 217 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 402 us MUL_MAT_ID(ffn_moe_down-60): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 361 us MUL_MAT_ID(ffn_moe_down-61): 200 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 369 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 402 us MUL_MAT_ID(ffn_moe_down-63): 199 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 527 us MUL_MAT_ID(ffn_moe_down-64): 208 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 388 us MUL_MAT_ID(ffn_moe_down-65): 206 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 402 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 350 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 388 us MUL_MAT_ID(ffn_moe_down-68): 222 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 369 us MUL_MAT_ID(ffn_moe_down-69): 192 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 374 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 387 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 391 us MUL_MAT_ID(ffn_moe_down-72): 514 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 399 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 360 us MUL_MAT_ID(ffn_moe_down-74): 221 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 385 us MUL_MAT_ID(ffn_moe_down-75): 199 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 370 us MUL_MAT_ID(ffn_moe_down-76): 187 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 454 us MUL_MAT_ID(ffn_moe_down-77): 374 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 389 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 415 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 366 us MUL_MAT_ID(ffn_moe_down-80): 200 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 392 us MUL_MAT_ID(ffn_moe_down-81): 408 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 382 us MUL_MAT_ID(ffn_moe_down-82): 375 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 377 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 366 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 370 us MUL_MAT_ID(ffn_moe_down-85): 734 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 412 us MUL_MAT_ID(ffn_moe_down-86): 395 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 385 us MUL_MAT_ID(ffn_moe_down-87): 209 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 374 us MUL_MAT_ID(ffn_moe_down-88): 355 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 392 us MUL_MAT_ID(ffn_moe_down-89): 183 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 402 us MUL_MAT_ID(ffn_moe_down-90): 248 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 398 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 20 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 403 us MUL_MAT_ID(ffn_moe_down-25): 171 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 418 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 367 us MUL_MAT_ID(ffn_moe_down-27): 220 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 392 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 422 us MUL_MAT_ID(ffn_moe_down-29): 182 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 393 us MUL_MAT_ID(ffn_moe_down-30): 285 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 446 us MUL_MAT_ID(ffn_moe_down-31): 539 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 412 us MUL_MAT_ID(ffn_moe_down-32): 262 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 411 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 392 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 393 us MUL_MAT_ID(ffn_moe_down-35): 223 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 395 us MUL_MAT_ID(ffn_moe_down-36): 539 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 436 us MUL_MAT_ID(ffn_moe_down-37): 254 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 481 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 388 us MUL_MAT_ID(ffn_moe_down-39): 219 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 411 us MUL_MAT_ID(ffn_moe_down-40): 217 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 363 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 392 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 405 us MUL_MAT_ID(ffn_moe_down-43): 188 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 359 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 463 us MUL_MAT_ID(ffn_moe_down-45): 178 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 349 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 420 us MUL_MAT_ID(ffn_moe_down-47): 299 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 401 us MUL_MAT_ID(ffn_moe_down-48): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 421 us MUL_MAT_ID(ffn_moe_down-49): 357 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 416 us MUL_MAT_ID(ffn_moe_down-50): 184 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 338 us MUL_MAT_ID(ffn_moe_down-51): 177 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 371 us MUL_MAT_ID(ffn_moe_down-52): 181 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 362 us MUL_MAT_ID(ffn_moe_down-53): 515 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 388 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 512 us MUL_MAT_ID(ffn_moe_down-55): 412 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 386 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 364 us MUL_MAT_ID(ffn_moe_down-57): 201 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 397 us MUL_MAT_ID(ffn_moe_down-58): 170 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 387 us MUL_MAT_ID(ffn_moe_down-59): 415 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 346 us MUL_MAT_ID(ffn_moe_down-60): 376 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 384 us MUL_MAT_ID(ffn_moe_down-61): 231 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 364 us MUL_MAT_ID(ffn_moe_down-62): 380 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 377 us MUL_MAT_ID(ffn_moe_down-63): 208 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 470 us MUL_MAT_ID(ffn_moe_down-64): 158 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 376 us MUL_MAT_ID(ffn_moe_down-65): 205 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 378 us MUL_MAT_ID(ffn_moe_down-66): 185 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 399 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 365 us MUL_MAT_ID(ffn_moe_down-68): 282 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 395 us MUL_MAT_ID(ffn_moe_down-69): 227 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 385 us MUL_MAT_ID(ffn_moe_down-70): 192 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 408 us MUL_MAT_ID(ffn_moe_down-71): 191 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 428 us MUL_MAT_ID(ffn_moe_down-72): 188 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 394 us MUL_MAT_ID(ffn_moe_down-73): 186 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 404 us MUL_MAT_ID(ffn_moe_down-74): 210 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 448 us MUL_MAT_ID(ffn_moe_down-75): 360 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 417 us MUL_MAT_ID(ffn_moe_down-76): 178 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 346 us MUL_MAT_ID(ffn_moe_down-77): 531 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 353 us MUL_MAT_ID(ffn_moe_down-78): 262 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 537 us MUL_MAT_ID(ffn_moe_down-79): 206 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 354 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 468 us MUL_MAT_ID(ffn_moe_down-81): 297 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 387 us MUL_MAT_ID(ffn_moe_down-82): 306 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 398 us MUL_MAT_ID(ffn_moe_down-83): 324 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 387 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 399 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 365 us MUL_MAT_ID(ffn_moe_down-86): 313 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 471 us MUL_MAT_ID(ffn_moe_down-87): 159 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 197 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 395 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 357 us MUL_MAT_ID(ffn_moe_down-90): 238 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 373 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 26 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 445 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 404 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 357 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 363 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 393 us MUL_MAT_ID(ffn_moe_down-28): 271 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 375 us MUL_MAT_ID(ffn_moe_down-30): 208 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 461 us MUL_MAT_ID(ffn_moe_down-31): 188 us ggml_barrier(...): 270 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 387 us MUL_MAT_ID(ffn_moe_down-32): 203 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 374 us MUL_MAT_ID(ffn_moe_down-33): 204 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 410 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 375 us MUL_MAT_ID(ffn_moe_down-35): 172 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 358 us MUL_MAT_ID(ffn_moe_down-36): 270 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 375 us MUL_MAT_ID(ffn_moe_down-37): 224 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 267 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 415 us MUL_MAT_ID(ffn_moe_down-39): 249 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 406 us MUL_MAT_ID(ffn_moe_down-40): 228 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 373 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 365 us MUL_MAT_ID(ffn_moe_down-42): 201 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 390 us MUL_MAT_ID(ffn_moe_down-43): 186 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 357 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 373 us MUL_MAT_ID(ffn_moe_down-45): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 375 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 349 us MUL_MAT_ID(ffn_moe_down-47): 228 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 388 us MUL_MAT_ID(ffn_moe_down-48): 379 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 391 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 511 us MUL_MAT_ID(ffn_moe_down-50): 217 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 396 us MUL_MAT_ID(ffn_moe_down-51): 229 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 384 us MUL_MAT_ID(ffn_moe_down-52): 188 us ggml_barrier(...): 365 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 402 us MUL_MAT_ID(ffn_moe_down-53): 178 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 392 us MUL_MAT_ID(ffn_moe_down-54): 221 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 453 us MUL_MAT_ID(ffn_moe_down-55): 402 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 186 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 399 us MUL_MAT_ID(ffn_moe_down-57): 182 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 424 us MUL_MAT_ID(ffn_moe_down-58): 188 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 387 us MUL_MAT_ID(ffn_moe_down-59): 205 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 389 us MUL_MAT_ID(ffn_moe_down-60): 259 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 360 us MUL_MAT_ID(ffn_moe_down-61): 204 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 396 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 377 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 390 us MUL_MAT_ID(ffn_moe_down-64): 211 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 383 us MUL_MAT_ID(ffn_moe_down-65): 202 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 346 us MUL_MAT_ID(ffn_moe_down-66): 183 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 392 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 384 us MUL_MAT_ID(ffn_moe_down-68): 260 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 375 us MUL_MAT_ID(ffn_moe_down-69): 255 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 366 us MUL_MAT_ID(ffn_moe_down-70): 205 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 386 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 410 us MUL_MAT_ID(ffn_moe_down-72): 214 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 383 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 382 us MUL_MAT_ID(ffn_moe_down-74): 213 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 402 us MUL_MAT_ID(ffn_moe_down-75): 200 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 397 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 363 us MUL_MAT_ID(ffn_moe_down-77): 210 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 495 us MUL_MAT_ID(ffn_moe_down-78): 171 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 373 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 365 us MUL_MAT_ID(ffn_moe_down-80): 186 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 421 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 399 us MUL_MAT_ID(ffn_moe_down-82): 190 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 479 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 401 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 394 us MUL_MAT_ID(ffn_moe_down-85): 624 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 408 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 346 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 363 us MUL_MAT_ID(ffn_moe_down-88): 229 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 350 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 408 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 363 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 48 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 397 us MUL_MAT_ID(ffn_moe_down-25): 417 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 371 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 375 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 375 us MUL_MAT_ID(ffn_moe_down-28): 199 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 368 us MUL_MAT_ID(ffn_moe_down-29): 219 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 371 us MUL_MAT_ID(ffn_moe_down-30): 184 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 394 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 399 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 374 us MUL_MAT_ID(ffn_moe_down-33): 188 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 411 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 369 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 362 us MUL_MAT_ID(ffn_moe_down-36): 214 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 379 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 357 us MUL_MAT_ID(ffn_moe_down-38): 236 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 362 us MUL_MAT_ID(ffn_moe_down-39): 686 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 364 us MUL_MAT_ID(ffn_moe_down-40): 222 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 394 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 418 us MUL_MAT_ID(ffn_moe_down-42): 216 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 394 us MUL_MAT_ID(ffn_moe_down-43): 212 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 372 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 371 us MUL_MAT_ID(ffn_moe_down-45): 210 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 367 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 522 us MUL_MAT_ID(ffn_moe_down-47): 223 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 355 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 390 us MUL_MAT_ID(ffn_moe_down-49): 187 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 407 us MUL_MAT_ID(ffn_moe_down-50): 197 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 468 us MUL_MAT_ID(ffn_moe_down-51): 300 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 395 us MUL_MAT_ID(ffn_moe_down-52): 379 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 369 us MUL_MAT_ID(ffn_moe_down-53): 167 us ggml_barrier(...): 338 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 431 us MUL_MAT_ID(ffn_moe_down-54): 271 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 369 us MUL_MAT_ID(ffn_moe_down-55): 193 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 403 us MUL_MAT_ID(ffn_moe_down-56): 370 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 404 us MUL_MAT_ID(ffn_moe_down-57): 180 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 407 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 378 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 408 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 389 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 343 us MUL_MAT_ID(ffn_moe_down-62): 187 us ggml_barrier(...): 365 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 396 us MUL_MAT_ID(ffn_moe_down-63): 168 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 376 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 353 us MUL_MAT_ID(ffn_moe_down-65): 215 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 415 us MUL_MAT_ID(ffn_moe_down-66): 183 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 383 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 381 us MUL_MAT_ID(ffn_moe_down-68): 266 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 378 us MUL_MAT_ID(ffn_moe_down-69): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 423 us MUL_MAT_ID(ffn_moe_down-70): 270 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 444 us MUL_MAT_ID(ffn_moe_down-71): 246 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 397 us MUL_MAT_ID(ffn_moe_down-72): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 409 us MUL_MAT_ID(ffn_moe_down-73): 179 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 413 us MUL_MAT_ID(ffn_moe_down-74): 200 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 377 us MUL_MAT_ID(ffn_moe_down-75): 183 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 394 us MUL_MAT_ID(ffn_moe_down-76): 382 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 371 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 392 us MUL_MAT_ID(ffn_moe_down-78): 205 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 423 us MUL_MAT_ID(ffn_moe_down-79): 227 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 365 us MUL_MAT_ID(ffn_moe_down-80): 318 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 414 us MUL_MAT_ID(ffn_moe_down-81): 203 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 395 us MUL_MAT_ID(ffn_moe_down-82): 205 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 422 us MUL_MAT_ID(ffn_moe_down-83): 210 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 397 us MUL_MAT_ID(ffn_moe_down-84): 262 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 429 us MUL_MAT_ID(ffn_moe_down-85): 227 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 381 us MUL_MAT_ID(ffn_moe_down-86): 280 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 511 us MUL_MAT_ID(ffn_moe_down-87): 215 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 352 us MUL_MAT_ID(ffn_moe_down-88): 223 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 406 us MUL_MAT_ID(ffn_moe_down-89): 205 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 408 us MUL_MAT_ID(ffn_moe_down-90): 237 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 396 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 22 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 559 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 386 us MUL_MAT_ID(ffn_moe_down-26): 191 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 382 us MUL_MAT_ID(ffn_moe_down-27): 221 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 355 us MUL_MAT_ID(ffn_moe_down-28): 190 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 396 us MUL_MAT_ID(ffn_moe_down-29): 181 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 378 us MUL_MAT_ID(ffn_moe_down-30): 223 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 396 us MUL_MAT_ID(ffn_moe_down-31): 203 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 427 us MUL_MAT_ID(ffn_moe_down-32): 181 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 361 us MUL_MAT_ID(ffn_moe_down-33): 209 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 378 us MUL_MAT_ID(ffn_moe_down-34): 207 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 381 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 384 us MUL_MAT_ID(ffn_moe_down-36): 384 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 405 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 393 us MUL_MAT_ID(ffn_moe_down-38): 282 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 360 us MUL_MAT_ID(ffn_moe_down-39): 256 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 395 us MUL_MAT_ID(ffn_moe_down-40): 227 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 384 us MUL_MAT_ID(ffn_moe_down-41): 226 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 407 us MUL_MAT_ID(ffn_moe_down-42): 204 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 372 us MUL_MAT_ID(ffn_moe_down-43): 216 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 340 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 536 us MUL_MAT_ID(ffn_moe_down-45): 166 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 404 us MUL_MAT_ID(ffn_moe_down-46): 302 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 388 us MUL_MAT_ID(ffn_moe_down-47): 401 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 407 us MUL_MAT_ID(ffn_moe_down-48): 185 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 392 us MUL_MAT_ID(ffn_moe_down-49): 209 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 418 us MUL_MAT_ID(ffn_moe_down-50): 301 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 424 us MUL_MAT_ID(ffn_moe_down-51): 202 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 401 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 428 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 384 us MUL_MAT_ID(ffn_moe_down-54): 217 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 368 us MUL_MAT_ID(ffn_moe_down-56): 197 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 371 us MUL_MAT_ID(ffn_moe_down-57): 198 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 377 us MUL_MAT_ID(ffn_moe_down-58): 275 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 361 us MUL_MAT_ID(ffn_moe_down-59): 278 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 376 us MUL_MAT_ID(ffn_moe_down-60): 248 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 442 us MUL_MAT_ID(ffn_moe_down-61): 186 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 383 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 382 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 396 us MUL_MAT_ID(ffn_moe_down-64): 514 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 399 us MUL_MAT_ID(ffn_moe_down-65): 183 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 223 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 387 us MUL_MAT_ID(ffn_moe_down-67): 230 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 391 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 444 us MUL_MAT_ID(ffn_moe_down-69): 171 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 350 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 432 us MUL_MAT_ID(ffn_moe_down-71): 236 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 371 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 345 us MUL_MAT_ID(ffn_moe_down-73): 213 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 408 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 402 us MUL_MAT_ID(ffn_moe_down-75): 191 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 396 us MUL_MAT_ID(ffn_moe_down-76): 229 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 380 us MUL_MAT_ID(ffn_moe_down-77): 197 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 399 us MUL_MAT_ID(ffn_moe_down-78): 219 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 393 us MUL_MAT_ID(ffn_moe_down-79): 202 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 480 us MUL_MAT_ID(ffn_moe_down-80): 201 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 354 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 353 us MUL_MAT_ID(ffn_moe_down-82): 189 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 384 us MUL_MAT_ID(ffn_moe_down-83): 206 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 355 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 407 us MUL_MAT_ID(ffn_moe_down-85): 237 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 379 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 427 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 422 us MUL_MAT_ID(ffn_moe_down-88): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 385 us MUL_MAT_ID(ffn_moe_down-89): 383 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 404 us MUL_MAT_ID(ffn_moe_down-90): 236 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 363 us MUL_MAT_ID(ffn_moe_down-91): 252 us ggml_barrier(...): 65 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 415 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 364 us MUL_MAT_ID(ffn_moe_down-26): 330 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 402 us MUL_MAT_ID(ffn_moe_down-27): 201 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 384 us MUL_MAT_ID(ffn_moe_down-28): 244 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 377 us MUL_MAT_ID(ffn_moe_down-29): 217 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 398 us MUL_MAT_ID(ffn_moe_down-30): 202 us ggml_barrier(...): 419 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 353 us MUL_MAT_ID(ffn_moe_down-31): 202 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 404 us MUL_MAT_ID(ffn_moe_down-32): 186 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 391 us MUL_MAT_ID(ffn_moe_down-33): 202 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 506 us MUL_MAT_ID(ffn_moe_down-34): 240 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 376 us MUL_MAT_ID(ffn_moe_down-35): 204 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 388 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 366 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 388 us MUL_MAT_ID(ffn_moe_down-38): 196 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 341 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 408 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 383 us MUL_MAT_ID(ffn_moe_down-41): 214 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 356 us MUL_MAT_ID(ffn_moe_down-42): 193 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 389 us MUL_MAT_ID(ffn_moe_down-43): 256 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 361 us MUL_MAT_ID(ffn_moe_down-44): 393 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 353 us MUL_MAT_ID(ffn_moe_down-45): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 376 us MUL_MAT_ID(ffn_moe_down-46): 363 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 391 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 417 us MUL_MAT_ID(ffn_moe_down-48): 365 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 389 us MUL_MAT_ID(ffn_moe_down-49): 203 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 358 us MUL_MAT_ID(ffn_moe_down-50): 193 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 386 us MUL_MAT_ID(ffn_moe_down-51): 292 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 384 us MUL_MAT_ID(ffn_moe_down-52): 470 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 403 us MUL_MAT_ID(ffn_moe_down-53): 181 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 381 us MUL_MAT_ID(ffn_moe_down-54): 232 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 388 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 396 us MUL_MAT_ID(ffn_moe_down-56): 210 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 361 us MUL_MAT_ID(ffn_moe_down-57): 201 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 367 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 367 us MUL_MAT_ID(ffn_moe_down-59): 213 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 381 us MUL_MAT_ID(ffn_moe_down-60): 191 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 384 us MUL_MAT_ID(ffn_moe_down-61): 235 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 422 us MUL_MAT_ID(ffn_moe_down-62): 225 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 435 us MUL_MAT_ID(ffn_moe_down-63): 220 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 400 us MUL_MAT_ID(ffn_moe_down-64): 234 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 384 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 403 us MUL_MAT_ID(ffn_moe_down-67): 219 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 417 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 369 us MUL_MAT_ID(ffn_moe_down-69): 209 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 420 us MUL_MAT_ID(ffn_moe_down-70): 208 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 356 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 434 us MUL_MAT_ID(ffn_moe_down-72): 174 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 369 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 344 us MUL_MAT_ID(ffn_moe_down-74): 182 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 375 us MUL_MAT_ID(ffn_moe_down-75): 195 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 386 us MUL_MAT_ID(ffn_moe_down-76): 185 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 360 us MUL_MAT_ID(ffn_moe_down-77): 205 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 356 us MUL_MAT_ID(ffn_moe_down-78): 509 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 445 us MUL_MAT_ID(ffn_moe_down-79): 330 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 510 us MUL_MAT_ID(ffn_moe_down-80): 163 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 401 us MUL_MAT_ID(ffn_moe_down-81): 188 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 385 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 392 us MUL_MAT_ID(ffn_moe_down-83): 203 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 392 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 396 us MUL_MAT_ID(ffn_moe_down-85): 271 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 400 us MUL_MAT_ID(ffn_moe_down-86): 220 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 470 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 349 us MUL_MAT_ID(ffn_moe_down-88): 183 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 344 us MUL_MAT_ID(ffn_moe_down-89): 207 us ggml_barrier(...): 346 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 406 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 529 us MUL_MAT_ID(ffn_moe_down-91): 274 us ggml_barrier(...): 59 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 395 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 387 us MUL_MAT_ID(ffn_moe_down-26): 185 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 414 us MUL_MAT_ID(ffn_moe_down-27): 405 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 404 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 396 us MUL_MAT_ID(ffn_moe_down-29): 204 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 411 us MUL_MAT_ID(ffn_moe_down-30): 200 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 420 us MUL_MAT_ID(ffn_moe_down-31): 362 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 405 us MUL_MAT_ID(ffn_moe_down-32): 355 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 398 us MUL_MAT_ID(ffn_moe_down-33): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 399 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 375 us MUL_MAT_ID(ffn_moe_down-35): 202 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 390 us MUL_MAT_ID(ffn_moe_down-36): 457 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 395 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 395 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 410 us MUL_MAT_ID(ffn_moe_down-39): 241 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 389 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 379 us MUL_MAT_ID(ffn_moe_down-41): 221 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 364 us MUL_MAT_ID(ffn_moe_down-42): 194 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 369 us MUL_MAT_ID(ffn_moe_down-43): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 353 us MUL_MAT_ID(ffn_moe_down-44): 193 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 401 us MUL_MAT_ID(ffn_moe_down-45): 557 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 346 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 413 us MUL_MAT_ID(ffn_moe_down-47): 211 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 395 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 355 us MUL_MAT_ID(ffn_moe_down-49): 184 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 390 us MUL_MAT_ID(ffn_moe_down-50): 206 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 396 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 377 us MUL_MAT_ID(ffn_moe_down-52): 459 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 377 us MUL_MAT_ID(ffn_moe_down-53): 231 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 399 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 374 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 380 us MUL_MAT_ID(ffn_moe_down-56): 211 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 380 us MUL_MAT_ID(ffn_moe_down-57): 198 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 395 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 353 us MUL_MAT_ID(ffn_moe_down-59): 188 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 393 us MUL_MAT_ID(ffn_moe_down-60): 189 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 384 us MUL_MAT_ID(ffn_moe_down-61): 436 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 386 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 394 us MUL_MAT_ID(ffn_moe_down-63): 533 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 407 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 378 us MUL_MAT_ID(ffn_moe_down-65): 212 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 387 us MUL_MAT_ID(ffn_moe_down-66): 184 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 368 us MUL_MAT_ID(ffn_moe_down-67): 202 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 401 us MUL_MAT_ID(ffn_moe_down-68): 183 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 359 us MUL_MAT_ID(ffn_moe_down-69): 318 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 379 us MUL_MAT_ID(ffn_moe_down-70): 434 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 362 us MUL_MAT_ID(ffn_moe_down-71): 411 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 448 us MUL_MAT_ID(ffn_moe_down-72): 219 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 422 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 400 us MUL_MAT_ID(ffn_moe_down-74): 190 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 347 us MUL_MAT_ID(ffn_moe_down-75): 336 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 409 us MUL_MAT_ID(ffn_moe_down-76): 203 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 415 us MUL_MAT_ID(ffn_moe_down-77): 191 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 408 us MUL_MAT_ID(ffn_moe_down-78): 211 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 368 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 368 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 408 us MUL_MAT_ID(ffn_moe_down-81): 205 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 351 us MUL_MAT_ID(ffn_moe_down-82): 194 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 424 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 372 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 359 us MUL_MAT_ID(ffn_moe_down-85): 227 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 400 us MUL_MAT_ID(ffn_moe_down-86): 215 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 391 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 377 us MUL_MAT_ID(ffn_moe_down-88): 317 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 375 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 368 us MUL_MAT_ID(ffn_moe_down-90): 248 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 416 us MUL_MAT_ID(ffn_moe_down-91): 247 us ggml_barrier(...): 25 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 377 us MUL_MAT_ID(ffn_moe_down-25): 176 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 384 us MUL_MAT_ID(ffn_moe_down-26): 183 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 365 us MUL_MAT_ID(ffn_moe_down-27): 187 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 399 us MUL_MAT_ID(ffn_moe_down-28): 188 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 395 us MUL_MAT_ID(ffn_moe_down-29): 251 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 395 us MUL_MAT_ID(ffn_moe_down-30): 209 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 372 us MUL_MAT_ID(ffn_moe_down-31): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 370 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 404 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 421 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 362 us MUL_MAT_ID(ffn_moe_down-35): 204 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 401 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 377 us MUL_MAT_ID(ffn_moe_down-37): 242 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 232 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 380 us MUL_MAT_ID(ffn_moe_down-39): 667 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 380 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 378 us MUL_MAT_ID(ffn_moe_down-41): 210 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 386 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 405 us MUL_MAT_ID(ffn_moe_down-43): 196 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 375 us MUL_MAT_ID(ffn_moe_down-44): 251 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 382 us MUL_MAT_ID(ffn_moe_down-45): 203 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 408 us MUL_MAT_ID(ffn_moe_down-46): 210 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 384 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 421 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 372 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 367 us MUL_MAT_ID(ffn_moe_down-50): 198 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 389 us MUL_MAT_ID(ffn_moe_down-51): 211 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 387 us MUL_MAT_ID(ffn_moe_down-52): 206 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 415 us MUL_MAT_ID(ffn_moe_down-53): 201 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 420 us MUL_MAT_ID(ffn_moe_down-54): 523 us ggml_barrier(...): 279 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 403 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 391 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 396 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 404 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 376 us MUL_MAT_ID(ffn_moe_down-59): 224 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 395 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 405 us MUL_MAT_ID(ffn_moe_down-61): 184 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 359 us MUL_MAT_ID(ffn_moe_down-62): 206 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 360 us MUL_MAT_ID(ffn_moe_down-63): 217 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 358 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 384 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 355 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 385 us MUL_MAT_ID(ffn_moe_down-67): 234 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 367 us MUL_MAT_ID(ffn_moe_down-68): 226 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 406 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 370 us MUL_MAT_ID(ffn_moe_down-70): 206 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 355 us MUL_MAT_ID(ffn_moe_down-71): 207 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 229 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 399 us MUL_MAT_ID(ffn_moe_down-73): 279 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 492 us MUL_MAT_ID(ffn_moe_down-74): 182 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 376 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 379 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 387 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 370 us MUL_MAT_ID(ffn_moe_down-78): 204 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 369 us MUL_MAT_ID(ffn_moe_down-79): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 372 us MUL_MAT_ID(ffn_moe_down-80): 206 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 365 us MUL_MAT_ID(ffn_moe_down-81): 185 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 377 us MUL_MAT_ID(ffn_moe_down-82): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 402 us MUL_MAT_ID(ffn_moe_down-83): 192 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 381 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 366 us MUL_MAT_ID(ffn_moe_down-85): 249 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 344 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 361 us MUL_MAT_ID(ffn_moe_down-87): 207 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 363 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 368 us MUL_MAT_ID(ffn_moe_down-89): 306 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 380 us MUL_MAT_ID(ffn_moe_down-90): 221 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 396 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 22 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 390 us MUL_MAT_ID(ffn_moe_down-26): 382 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 397 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 365 us MUL_MAT_ID(ffn_moe_down-28): 191 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 384 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 370 us MUL_MAT_ID(ffn_moe_down-31): 179 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 407 us MUL_MAT_ID(ffn_moe_down-32): 205 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 392 us MUL_MAT_ID(ffn_moe_down-33): 219 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 412 us MUL_MAT_ID(ffn_moe_down-34): 213 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 390 us MUL_MAT_ID(ffn_moe_down-35): 229 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 403 us MUL_MAT_ID(ffn_moe_down-36): 213 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 389 us MUL_MAT_ID(ffn_moe_down-37): 238 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 388 us MUL_MAT_ID(ffn_moe_down-38): 205 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 377 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 372 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 381 us MUL_MAT_ID(ffn_moe_down-41): 208 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 370 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 351 us MUL_MAT_ID(ffn_moe_down-43): 201 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 382 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 403 us MUL_MAT_ID(ffn_moe_down-45): 281 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 383 us MUL_MAT_ID(ffn_moe_down-46): 355 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 359 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 401 us MUL_MAT_ID(ffn_moe_down-48): 283 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 369 us MUL_MAT_ID(ffn_moe_down-49): 206 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 388 us MUL_MAT_ID(ffn_moe_down-50): 435 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 412 us MUL_MAT_ID(ffn_moe_down-51): 203 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 351 us MUL_MAT_ID(ffn_moe_down-52): 201 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 410 us MUL_MAT_ID(ffn_moe_down-53): 349 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 530 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 384 us MUL_MAT_ID(ffn_moe_down-55): 269 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 376 us MUL_MAT_ID(ffn_moe_down-56): 203 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 399 us MUL_MAT_ID(ffn_moe_down-57): 401 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 408 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 360 us MUL_MAT_ID(ffn_moe_down-59): 208 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 406 us MUL_MAT_ID(ffn_moe_down-60): 403 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 381 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 405 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 369 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 369 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 384 us MUL_MAT_ID(ffn_moe_down-65): 268 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 359 us MUL_MAT_ID(ffn_moe_down-66): 343 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 406 us MUL_MAT_ID(ffn_moe_down-67): 351 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 463 us MUL_MAT_ID(ffn_moe_down-68): 331 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 403 us MUL_MAT_ID(ffn_moe_down-69): 221 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 373 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 375 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 540 us MUL_MAT_ID(ffn_moe_down-72): 177 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 390 us MUL_MAT_ID(ffn_moe_down-73): 528 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 371 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 369 us MUL_MAT_ID(ffn_moe_down-75): 371 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 351 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 384 us MUL_MAT_ID(ffn_moe_down-77): 271 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 373 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 434 us MUL_MAT_ID(ffn_moe_down-79): 179 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 346 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 367 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 389 us MUL_MAT_ID(ffn_moe_down-82): 205 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 357 us MUL_MAT_ID(ffn_moe_down-83): 224 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 370 us MUL_MAT_ID(ffn_moe_down-84): 315 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 389 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 389 us MUL_MAT_ID(ffn_moe_down-86): 411 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 391 us MUL_MAT_ID(ffn_moe_down-87): 434 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 390 us MUL_MAT_ID(ffn_moe_down-88): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 372 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 403 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 359 us MUL_MAT_ID(ffn_moe_down-91): 229 us ggml_barrier(...): 69 us GET_ROWS(inp_embd): 22 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 385 us MUL_MAT_ID(ffn_moe_down-25): 186 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 387 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 378 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 372 us MUL_MAT_ID(ffn_moe_down-28): 193 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 368 us MUL_MAT_ID(ffn_moe_down-29): 194 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 363 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 395 us MUL_MAT_ID(ffn_moe_down-31): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 372 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 405 us MUL_MAT_ID(ffn_moe_down-33): 213 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 401 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 371 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 364 us MUL_MAT_ID(ffn_moe_down-36): 404 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 392 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 361 us MUL_MAT_ID(ffn_moe_down-39): 243 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 351 us MUL_MAT_ID(ffn_moe_down-40): 424 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 371 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 385 us MUL_MAT_ID(ffn_moe_down-42): 327 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 381 us MUL_MAT_ID(ffn_moe_down-43): 212 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 418 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 370 us MUL_MAT_ID(ffn_moe_down-45): 222 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 370 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 394 us MUL_MAT_ID(ffn_moe_down-47): 201 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 352 us MUL_MAT_ID(ffn_moe_down-48): 539 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 385 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 387 us MUL_MAT_ID(ffn_moe_down-50): 215 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 398 us MUL_MAT_ID(ffn_moe_down-51): 214 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 387 us MUL_MAT_ID(ffn_moe_down-52): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 384 us MUL_MAT_ID(ffn_moe_down-53): 180 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 383 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 385 us MUL_MAT_ID(ffn_moe_down-55): 189 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 389 us MUL_MAT_ID(ffn_moe_down-56): 242 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 412 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 419 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 385 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 380 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 377 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 368 us MUL_MAT_ID(ffn_moe_down-62): 207 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 393 us MUL_MAT_ID(ffn_moe_down-63): 304 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 377 us MUL_MAT_ID(ffn_moe_down-64): 186 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 406 us MUL_MAT_ID(ffn_moe_down-65): 527 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 359 us MUL_MAT_ID(ffn_moe_down-66): 335 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 401 us MUL_MAT_ID(ffn_moe_down-67): 196 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 354 us MUL_MAT_ID(ffn_moe_down-68): 271 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 389 us MUL_MAT_ID(ffn_moe_down-69): 190 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 377 us MUL_MAT_ID(ffn_moe_down-70): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 375 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 382 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 423 us MUL_MAT_ID(ffn_moe_down-73): 362 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 386 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 382 us MUL_MAT_ID(ffn_moe_down-75): 210 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 381 us MUL_MAT_ID(ffn_moe_down-76): 182 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 368 us MUL_MAT_ID(ffn_moe_down-77): 211 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 414 us MUL_MAT_ID(ffn_moe_down-78): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 346 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 345 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 370 us MUL_MAT_ID(ffn_moe_down-81): 200 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 356 us MUL_MAT_ID(ffn_moe_down-82): 215 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 375 us MUL_MAT_ID(ffn_moe_down-83): 400 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 402 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 405 us MUL_MAT_ID(ffn_moe_down-85): 226 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 400 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 392 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 382 us MUL_MAT_ID(ffn_moe_down-88): 201 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 386 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 367 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 368 us MUL_MAT_ID(ffn_moe_down-91): 250 us ggml_barrier(...): 54 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 407 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 204 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 378 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 391 us MUL_MAT_ID(ffn_moe_down-27): 221 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 398 us MUL_MAT_ID(ffn_moe_down-28): 210 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 376 us MUL_MAT_ID(ffn_moe_down-29): 217 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 402 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 352 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 385 us MUL_MAT_ID(ffn_moe_down-32): 199 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 367 us MUL_MAT_ID(ffn_moe_down-33): 181 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 389 us MUL_MAT_ID(ffn_moe_down-34): 328 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 365 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 392 us MUL_MAT_ID(ffn_moe_down-36): 244 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 373 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 394 us MUL_MAT_ID(ffn_moe_down-38): 181 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 383 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 394 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 395 us MUL_MAT_ID(ffn_moe_down-41): 205 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 397 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 348 us MUL_MAT_ID(ffn_moe_down-43): 196 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 373 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 376 us MUL_MAT_ID(ffn_moe_down-45): 223 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 375 us MUL_MAT_ID(ffn_moe_down-46): 204 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 398 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 382 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 395 us MUL_MAT_ID(ffn_moe_down-49): 202 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 388 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 361 us MUL_MAT_ID(ffn_moe_down-51): 182 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 398 us MUL_MAT_ID(ffn_moe_down-52): 280 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 359 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 404 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 357 us MUL_MAT_ID(ffn_moe_down-55): 202 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 380 us MUL_MAT_ID(ffn_moe_down-56): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 397 us MUL_MAT_ID(ffn_moe_down-57): 224 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 374 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 364 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 356 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 413 us MUL_MAT_ID(ffn_moe_down-61): 269 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 388 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 367 us MUL_MAT_ID(ffn_moe_down-63): 187 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 400 us MUL_MAT_ID(ffn_moe_down-64): 419 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 397 us MUL_MAT_ID(ffn_moe_down-65): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 407 us MUL_MAT_ID(ffn_moe_down-66): 276 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 388 us MUL_MAT_ID(ffn_moe_down-67): 183 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 390 us MUL_MAT_ID(ffn_moe_down-68): 219 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 386 us MUL_MAT_ID(ffn_moe_down-69): 180 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 403 us MUL_MAT_ID(ffn_moe_down-70): 207 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 381 us MUL_MAT_ID(ffn_moe_down-71): 436 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 372 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 356 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 374 us MUL_MAT_ID(ffn_moe_down-74): 221 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 392 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 394 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 388 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 406 us MUL_MAT_ID(ffn_moe_down-78): 217 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 383 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 364 us MUL_MAT_ID(ffn_moe_down-80): 194 us ggml_barrier(...): 331 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 407 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 375 us MUL_MAT_ID(ffn_moe_down-82): 202 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 460 us MUL_MAT_ID(ffn_moe_down-83): 180 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 422 us MUL_MAT_ID(ffn_moe_down-84): 411 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 405 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 376 us MUL_MAT_ID(ffn_moe_down-86): 195 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 364 us MUL_MAT_ID(ffn_moe_down-87): 354 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 364 us MUL_MAT_ID(ffn_moe_down-88): 297 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 381 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 416 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 408 us MUL_MAT_ID(ffn_moe_down-91): 277 us ggml_barrier(...): 166 us GET_ROWS(inp_embd): 13 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 392 us MUL_MAT_ID(ffn_moe_down-25): 199 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 371 us MUL_MAT_ID(ffn_moe_down-26): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 398 us MUL_MAT_ID(ffn_moe_down-27): 208 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 362 us MUL_MAT_ID(ffn_moe_down-28): 411 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 374 us MUL_MAT_ID(ffn_moe_down-29): 369 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 391 us MUL_MAT_ID(ffn_moe_down-30): 192 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 399 us MUL_MAT_ID(ffn_moe_down-31): 212 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 395 us MUL_MAT_ID(ffn_moe_down-32): 381 us ggml_barrier(...): 298 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 362 us MUL_MAT_ID(ffn_moe_down-33): 214 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 390 us MUL_MAT_ID(ffn_moe_down-34): 206 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 357 us MUL_MAT_ID(ffn_moe_down-35): 174 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 397 us MUL_MAT_ID(ffn_moe_down-36): 211 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 381 us MUL_MAT_ID(ffn_moe_down-37): 298 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 368 us MUL_MAT_ID(ffn_moe_down-38): 186 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 377 us MUL_MAT_ID(ffn_moe_down-39): 229 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 347 us MUL_MAT_ID(ffn_moe_down-40): 245 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 351 us MUL_MAT_ID(ffn_moe_down-41): 386 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 412 us MUL_MAT_ID(ffn_moe_down-42): 298 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 417 us MUL_MAT_ID(ffn_moe_down-43): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 353 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 392 us MUL_MAT_ID(ffn_moe_down-45): 198 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 355 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 420 us MUL_MAT_ID(ffn_moe_down-47): 312 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 352 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 378 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 377 us MUL_MAT_ID(ffn_moe_down-50): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 393 us MUL_MAT_ID(ffn_moe_down-51): 219 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 388 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 392 us MUL_MAT_ID(ffn_moe_down-53): 186 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 348 us MUL_MAT_ID(ffn_moe_down-54): 244 us ggml_barrier(...): 282 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 393 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 408 us MUL_MAT_ID(ffn_moe_down-56): 210 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 372 us MUL_MAT_ID(ffn_moe_down-57): 220 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 372 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 370 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 353 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 381 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 353 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 356 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 349 us MUL_MAT_ID(ffn_moe_down-64): 211 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 376 us MUL_MAT_ID(ffn_moe_down-65): 188 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 355 us MUL_MAT_ID(ffn_moe_down-66): 184 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 395 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 371 us MUL_MAT_ID(ffn_moe_down-68): 526 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 357 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 394 us MUL_MAT_ID(ffn_moe_down-70): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 385 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 393 us MUL_MAT_ID(ffn_moe_down-72): 238 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 372 us MUL_MAT_ID(ffn_moe_down-73): 212 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 359 us MUL_MAT_ID(ffn_moe_down-74): 226 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 381 us MUL_MAT_ID(ffn_moe_down-75): 218 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 374 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 346 us MUL_MAT_ID(ffn_moe_down-77): 357 us ggml_barrier(...): 316 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 379 us MUL_MAT_ID(ffn_moe_down-78): 226 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 368 us MUL_MAT_ID(ffn_moe_down-79): 208 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 379 us MUL_MAT_ID(ffn_moe_down-80): 187 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 372 us MUL_MAT_ID(ffn_moe_down-81): 196 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 392 us MUL_MAT_ID(ffn_moe_down-82): 521 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 386 us MUL_MAT_ID(ffn_moe_down-83): 201 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 409 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 355 us MUL_MAT_ID(ffn_moe_down-85): 235 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 391 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 368 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 354 us MUL_MAT_ID(ffn_moe_down-88): 212 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 368 us MUL_MAT_ID(ffn_moe_down-89): 348 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 360 us MUL_MAT_ID(ffn_moe_down-90): 657 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 409 us MUL_MAT_ID(ffn_moe_down-91): 233 us ggml_barrier(...): 357 us GET_ROWS(inp_embd): 22 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 378 us MUL_MAT_ID(ffn_moe_down-25): 175 us ggml_barrier(...): 335 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 404 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 378 us MUL_MAT_ID(ffn_moe_down-27): 390 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 186 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 364 us MUL_MAT_ID(ffn_moe_down-29): 187 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 419 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 377 us MUL_MAT_ID(ffn_moe_down-31): 208 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 384 us MUL_MAT_ID(ffn_moe_down-32): 260 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 373 us MUL_MAT_ID(ffn_moe_down-33): 396 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 384 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 362 us MUL_MAT_ID(ffn_moe_down-35): 202 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 364 us MUL_MAT_ID(ffn_moe_down-36): 184 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 410 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 376 us MUL_MAT_ID(ffn_moe_down-38): 536 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 391 us MUL_MAT_ID(ffn_moe_down-39): 574 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 422 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 370 us MUL_MAT_ID(ffn_moe_down-41): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 355 us MUL_MAT_ID(ffn_moe_down-42): 206 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 376 us MUL_MAT_ID(ffn_moe_down-43): 232 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 354 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 447 us MUL_MAT_ID(ffn_moe_down-45): 195 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 367 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 373 us MUL_MAT_ID(ffn_moe_down-47): 209 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 371 us MUL_MAT_ID(ffn_moe_down-48): 190 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 368 us MUL_MAT_ID(ffn_moe_down-49): 207 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 382 us MUL_MAT_ID(ffn_moe_down-50): 193 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 365 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 350 us MUL_MAT_ID(ffn_moe_down-52): 239 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 372 us MUL_MAT_ID(ffn_moe_down-53): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 363 us MUL_MAT_ID(ffn_moe_down-54): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 357 us MUL_MAT_ID(ffn_moe_down-55): 184 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 351 us MUL_MAT_ID(ffn_moe_down-56): 357 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 410 us MUL_MAT_ID(ffn_moe_down-57): 203 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 395 us MUL_MAT_ID(ffn_moe_down-58): 185 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 411 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 358 us MUL_MAT_ID(ffn_moe_down-60): 201 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 370 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 363 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 357 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 376 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 359 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 375 us MUL_MAT_ID(ffn_moe_down-67): 385 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 367 us MUL_MAT_ID(ffn_moe_down-68): 257 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 509 us MUL_MAT_ID(ffn_moe_down-69): 391 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 405 us MUL_MAT_ID(ffn_moe_down-70): 374 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 412 us MUL_MAT_ID(ffn_moe_down-71): 212 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 421 us MUL_MAT_ID(ffn_moe_down-72): 340 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 386 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 388 us MUL_MAT_ID(ffn_moe_down-74): 213 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 374 us MUL_MAT_ID(ffn_moe_down-75): 216 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 365 us MUL_MAT_ID(ffn_moe_down-76): 180 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 395 us MUL_MAT_ID(ffn_moe_down-77): 397 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 383 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 398 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 418 us MUL_MAT_ID(ffn_moe_down-80): 344 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 362 us MUL_MAT_ID(ffn_moe_down-81): 181 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 367 us MUL_MAT_ID(ffn_moe_down-82): 190 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 373 us MUL_MAT_ID(ffn_moe_down-83): 394 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 372 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 381 us MUL_MAT_ID(ffn_moe_down-85): 254 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 391 us MUL_MAT_ID(ffn_moe_down-86): 207 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 420 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 386 us MUL_MAT_ID(ffn_moe_down-88): 207 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 418 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 394 us MUL_MAT_ID(ffn_moe_down-90): 246 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 408 us MUL_MAT_ID(ffn_moe_down-91): 229 us ggml_barrier(...): 25 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 417 us MUL_MAT_ID(ffn_moe_down-25): 198 us ggml_barrier(...): 452 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 403 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 396 us MUL_MAT_ID(ffn_moe_down-27): 191 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 354 us MUL_MAT_ID(ffn_moe_down-28): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 387 us MUL_MAT_ID(ffn_moe_down-29): 176 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 421 us MUL_MAT_ID(ffn_moe_down-30): 244 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 378 us MUL_MAT_ID(ffn_moe_down-31): 197 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 385 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 394 us MUL_MAT_ID(ffn_moe_down-33): 212 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 385 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 423 us MUL_MAT_ID(ffn_moe_down-35): 639 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 347 us MUL_MAT_ID(ffn_moe_down-36): 189 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 401 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 381 us MUL_MAT_ID(ffn_moe_down-38): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 352 us MUL_MAT_ID(ffn_moe_down-39): 245 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 356 us MUL_MAT_ID(ffn_moe_down-40): 221 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 360 us MUL_MAT_ID(ffn_moe_down-41): 200 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 376 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 366 us MUL_MAT_ID(ffn_moe_down-43): 197 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 386 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 411 us MUL_MAT_ID(ffn_moe_down-45): 196 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 374 us MUL_MAT_ID(ffn_moe_down-46): 210 us ggml_barrier(...): 134 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 481 us MUL_MAT_ID(ffn_moe_down-47): 208 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 440 us MUL_MAT_ID(ffn_moe_down-48): 270 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 414 us MUL_MAT_ID(ffn_moe_down-49): 216 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 368 us MUL_MAT_ID(ffn_moe_down-50): 189 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 393 us MUL_MAT_ID(ffn_moe_down-51): 191 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 387 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 366 us MUL_MAT_ID(ffn_moe_down-53): 242 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 513 us MUL_MAT_ID(ffn_moe_down-54): 441 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 408 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 386 us MUL_MAT_ID(ffn_moe_down-56): 248 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 393 us MUL_MAT_ID(ffn_moe_down-57): 500 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 367 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 393 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 434 us MUL_MAT_ID(ffn_moe_down-60): 217 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 405 us MUL_MAT_ID(ffn_moe_down-61): 497 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 360 us MUL_MAT_ID(ffn_moe_down-62): 211 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 388 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 379 us MUL_MAT_ID(ffn_moe_down-64): 201 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 570 us MUL_MAT_ID(ffn_moe_down-65): 170 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 400 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 393 us MUL_MAT_ID(ffn_moe_down-67): 183 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 368 us MUL_MAT_ID(ffn_moe_down-68): 524 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 381 us MUL_MAT_ID(ffn_moe_down-69): 194 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 363 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 360 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 353 us MUL_MAT_ID(ffn_moe_down-72): 227 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 199 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 357 us MUL_MAT_ID(ffn_moe_down-74): 196 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 387 us MUL_MAT_ID(ffn_moe_down-75): 188 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 394 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 375 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 353 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 425 us MUL_MAT_ID(ffn_moe_down-79): 197 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 361 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 362 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 388 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 352 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 373 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 368 us MUL_MAT_ID(ffn_moe_down-85): 246 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 379 us MUL_MAT_ID(ffn_moe_down-86): 372 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 442 us MUL_MAT_ID(ffn_moe_down-87): 156 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 408 us MUL_MAT_ID(ffn_moe_down-88): 222 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 392 us MUL_MAT_ID(ffn_moe_down-89): 176 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 358 us MUL_MAT_ID(ffn_moe_down-90): 648 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 415 us MUL_MAT_ID(ffn_moe_down-91): 248 us ggml_barrier(...): 33 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 376 us MUL_MAT_ID(ffn_moe_down-25): 176 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 562 us MUL_MAT_ID(ffn_moe_down-26): 207 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 465 us MUL_MAT_ID(ffn_moe_down-27): 175 us ggml_barrier(...): 506 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 424 us MUL_MAT_ID(ffn_moe_down-28): 404 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 425 us MUL_MAT_ID(ffn_moe_down-29): 375 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 358 us MUL_MAT_ID(ffn_moe_down-30): 267 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 379 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 377 us MUL_MAT_ID(ffn_moe_down-32): 343 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 413 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 388 us MUL_MAT_ID(ffn_moe_down-34): 206 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 430 us MUL_MAT_ID(ffn_moe_down-35): 373 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 367 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 376 us MUL_MAT_ID(ffn_moe_down-37): 382 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 384 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 397 us MUL_MAT_ID(ffn_moe_down-39): 255 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 360 us MUL_MAT_ID(ffn_moe_down-40): 417 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 382 us MUL_MAT_ID(ffn_moe_down-41): 334 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 353 us MUL_MAT_ID(ffn_moe_down-42): 364 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 357 us MUL_MAT_ID(ffn_moe_down-43): 207 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 352 us MUL_MAT_ID(ffn_moe_down-44): 179 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 362 us MUL_MAT_ID(ffn_moe_down-45): 394 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 382 us MUL_MAT_ID(ffn_moe_down-46): 498 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 367 us MUL_MAT_ID(ffn_moe_down-47): 187 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 387 us MUL_MAT_ID(ffn_moe_down-48): 198 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 384 us MUL_MAT_ID(ffn_moe_down-49): 211 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 358 us MUL_MAT_ID(ffn_moe_down-50): 205 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 395 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 396 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 352 us MUL_MAT_ID(ffn_moe_down-53): 178 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 377 us MUL_MAT_ID(ffn_moe_down-54): 519 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 365 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 359 us MUL_MAT_ID(ffn_moe_down-56): 211 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 401 us MUL_MAT_ID(ffn_moe_down-57): 208 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 390 us MUL_MAT_ID(ffn_moe_down-58): 291 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 376 us MUL_MAT_ID(ffn_moe_down-59): 187 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 378 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 395 us MUL_MAT_ID(ffn_moe_down-61): 301 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 344 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 389 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 404 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 399 us MUL_MAT_ID(ffn_moe_down-65): 210 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 395 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 408 us MUL_MAT_ID(ffn_moe_down-67): 279 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 361 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 390 us MUL_MAT_ID(ffn_moe_down-69): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 366 us MUL_MAT_ID(ffn_moe_down-70): 307 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 408 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 359 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 404 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 377 us MUL_MAT_ID(ffn_moe_down-74): 224 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 361 us MUL_MAT_ID(ffn_moe_down-75): 209 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 363 us MUL_MAT_ID(ffn_moe_down-76): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 515 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 393 us MUL_MAT_ID(ffn_moe_down-78): 181 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 365 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 357 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 349 us MUL_MAT_ID(ffn_moe_down-81): 180 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 355 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 387 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 402 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 383 us MUL_MAT_ID(ffn_moe_down-85): 254 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 425 us MUL_MAT_ID(ffn_moe_down-86): 426 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 405 us MUL_MAT_ID(ffn_moe_down-87): 374 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 367 us MUL_MAT_ID(ffn_moe_down-88): 207 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 417 us MUL_MAT_ID(ffn_moe_down-89): 414 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 368 us MUL_MAT_ID(ffn_moe_down-90): 250 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 336 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 69 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 397 us MUL_MAT_ID(ffn_moe_down-25): 200 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 489 us MUL_MAT_ID(ffn_moe_down-26): 177 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 398 us MUL_MAT_ID(ffn_moe_down-27): 399 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 398 us MUL_MAT_ID(ffn_moe_down-29): 210 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 432 us MUL_MAT_ID(ffn_moe_down-30): 191 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 387 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 376 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 400 us MUL_MAT_ID(ffn_moe_down-33): 376 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 415 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 374 us MUL_MAT_ID(ffn_moe_down-35): 468 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 376 us MUL_MAT_ID(ffn_moe_down-36): 531 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 375 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 419 us MUL_MAT_ID(ffn_moe_down-38): 187 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 347 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 396 us MUL_MAT_ID(ffn_moe_down-40): 222 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 376 us MUL_MAT_ID(ffn_moe_down-41): 391 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 372 us MUL_MAT_ID(ffn_moe_down-42): 228 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 387 us MUL_MAT_ID(ffn_moe_down-43): 376 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 397 us MUL_MAT_ID(ffn_moe_down-44): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 401 us MUL_MAT_ID(ffn_moe_down-45): 191 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 407 us MUL_MAT_ID(ffn_moe_down-46): 206 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 407 us MUL_MAT_ID(ffn_moe_down-47): 202 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 359 us MUL_MAT_ID(ffn_moe_down-48): 197 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 386 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 369 us MUL_MAT_ID(ffn_moe_down-50): 218 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 362 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 375 us MUL_MAT_ID(ffn_moe_down-52): 206 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 356 us MUL_MAT_ID(ffn_moe_down-53): 155 us ggml_barrier(...): 367 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 371 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 372 us MUL_MAT_ID(ffn_moe_down-55): 396 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 416 us MUL_MAT_ID(ffn_moe_down-56): 167 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 393 us MUL_MAT_ID(ffn_moe_down-57): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 382 us MUL_MAT_ID(ffn_moe_down-58): 227 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 353 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 376 us MUL_MAT_ID(ffn_moe_down-60): 201 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 415 us MUL_MAT_ID(ffn_moe_down-61): 396 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 364 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 381 us MUL_MAT_ID(ffn_moe_down-63): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 390 us MUL_MAT_ID(ffn_moe_down-64): 216 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 397 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 380 us MUL_MAT_ID(ffn_moe_down-66): 218 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 388 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 388 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 421 us MUL_MAT_ID(ffn_moe_down-69): 410 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 390 us MUL_MAT_ID(ffn_moe_down-70): 429 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 390 us MUL_MAT_ID(ffn_moe_down-71): 244 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 413 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 397 us MUL_MAT_ID(ffn_moe_down-73): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 395 us MUL_MAT_ID(ffn_moe_down-74): 410 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 355 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 390 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 404 us MUL_MAT_ID(ffn_moe_down-77): 187 us ggml_barrier(...): 300 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 382 us MUL_MAT_ID(ffn_moe_down-78): 384 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 443 us MUL_MAT_ID(ffn_moe_down-79): 188 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 347 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 380 us MUL_MAT_ID(ffn_moe_down-81): 200 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 350 us MUL_MAT_ID(ffn_moe_down-82): 359 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 402 us MUL_MAT_ID(ffn_moe_down-83): 190 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 368 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 388 us MUL_MAT_ID(ffn_moe_down-85): 710 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 402 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 381 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 389 us MUL_MAT_ID(ffn_moe_down-88): 179 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 337 us MUL_MAT_ID(ffn_moe_down-89): 196 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 384 us MUL_MAT_ID(ffn_moe_down-90): 672 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 382 us MUL_MAT_ID(ffn_moe_down-91): 248 us ggml_barrier(...): 31 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 382 us MUL_MAT_ID(ffn_moe_down-25): 217 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 383 us MUL_MAT_ID(ffn_moe_down-26): 207 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 425 us MUL_MAT_ID(ffn_moe_down-27): 210 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 403 us MUL_MAT_ID(ffn_moe_down-28): 337 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 393 us MUL_MAT_ID(ffn_moe_down-29): 370 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 386 us MUL_MAT_ID(ffn_moe_down-30): 182 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 377 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 373 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 411 us MUL_MAT_ID(ffn_moe_down-33): 206 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 484 us MUL_MAT_ID(ffn_moe_down-34): 252 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 422 us MUL_MAT_ID(ffn_moe_down-35): 417 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 358 us MUL_MAT_ID(ffn_moe_down-36): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 373 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 360 us MUL_MAT_ID(ffn_moe_down-38): 205 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 346 us MUL_MAT_ID(ffn_moe_down-39): 228 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 382 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 410 us MUL_MAT_ID(ffn_moe_down-41): 374 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 379 us MUL_MAT_ID(ffn_moe_down-42): 186 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 365 us MUL_MAT_ID(ffn_moe_down-43): 352 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 346 us MUL_MAT_ID(ffn_moe_down-44): 183 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 380 us MUL_MAT_ID(ffn_moe_down-45): 212 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 432 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 389 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 382 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 371 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 367 us MUL_MAT_ID(ffn_moe_down-50): 185 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 385 us MUL_MAT_ID(ffn_moe_down-51): 397 us ggml_barrier(...): 449 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 443 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 394 us MUL_MAT_ID(ffn_moe_down-53): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 391 us MUL_MAT_ID(ffn_moe_down-54): 328 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 340 us MUL_MAT_ID(ffn_moe_down-55): 210 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 422 us MUL_MAT_ID(ffn_moe_down-56): 225 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 388 us MUL_MAT_ID(ffn_moe_down-57): 278 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 388 us MUL_MAT_ID(ffn_moe_down-58): 227 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 460 us MUL_MAT_ID(ffn_moe_down-59): 312 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 377 us MUL_MAT_ID(ffn_moe_down-60): 208 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 374 us MUL_MAT_ID(ffn_moe_down-61): 426 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 361 us MUL_MAT_ID(ffn_moe_down-62): 331 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 382 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 394 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 368 us MUL_MAT_ID(ffn_moe_down-65): 204 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 372 us MUL_MAT_ID(ffn_moe_down-66): 201 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 428 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 348 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 403 us MUL_MAT_ID(ffn_moe_down-69): 408 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 403 us MUL_MAT_ID(ffn_moe_down-70): 317 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 407 us MUL_MAT_ID(ffn_moe_down-71): 213 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 377 us MUL_MAT_ID(ffn_moe_down-72): 182 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 391 us MUL_MAT_ID(ffn_moe_down-73): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 374 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 389 us MUL_MAT_ID(ffn_moe_down-75): 272 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 382 us MUL_MAT_ID(ffn_moe_down-76): 204 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 379 us MUL_MAT_ID(ffn_moe_down-77): 215 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 406 us MUL_MAT_ID(ffn_moe_down-78): 216 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 406 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 387 us MUL_MAT_ID(ffn_moe_down-80): 196 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 371 us MUL_MAT_ID(ffn_moe_down-81): 185 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 444 us MUL_MAT_ID(ffn_moe_down-82): 296 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 407 us MUL_MAT_ID(ffn_moe_down-83): 168 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 387 us MUL_MAT_ID(ffn_moe_down-84): 244 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 397 us MUL_MAT_ID(ffn_moe_down-85): 242 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 352 us MUL_MAT_ID(ffn_moe_down-86): 166 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 359 us MUL_MAT_ID(ffn_moe_down-87): 182 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 387 us MUL_MAT_ID(ffn_moe_down-88): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 500 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 361 us MUL_MAT_ID(ffn_moe_down-90): 698 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 370 us MUL_MAT_ID(ffn_moe_down-91): 362 us ggml_barrier(...): 63 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 378 us MUL_MAT_ID(ffn_moe_down-25): 231 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 394 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 361 us MUL_MAT_ID(ffn_moe_down-27): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 399 us MUL_MAT_ID(ffn_moe_down-28): 207 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 416 us MUL_MAT_ID(ffn_moe_down-29): 237 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 420 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 430 us MUL_MAT_ID(ffn_moe_down-31): 194 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 373 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 485 us MUL_MAT_ID(ffn_moe_down-33): 209 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 390 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 400 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 409 us MUL_MAT_ID(ffn_moe_down-36): 384 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 441 us MUL_MAT_ID(ffn_moe_down-37): 602 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 399 us MUL_MAT_ID(ffn_moe_down-38): 340 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 343 us MUL_MAT_ID(ffn_moe_down-39): 493 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 390 us MUL_MAT_ID(ffn_moe_down-40): 248 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 374 us MUL_MAT_ID(ffn_moe_down-41): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 370 us MUL_MAT_ID(ffn_moe_down-42): 288 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 401 us MUL_MAT_ID(ffn_moe_down-43): 204 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 358 us MUL_MAT_ID(ffn_moe_down-44): 380 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 396 us MUL_MAT_ID(ffn_moe_down-45): 339 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 384 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 366 us MUL_MAT_ID(ffn_moe_down-47): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 416 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 372 us MUL_MAT_ID(ffn_moe_down-49): 219 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 404 us MUL_MAT_ID(ffn_moe_down-50): 265 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 360 us MUL_MAT_ID(ffn_moe_down-51): 215 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 364 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 346 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 345 us MUL_MAT_ID(ffn_moe_down-54): 187 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 352 us MUL_MAT_ID(ffn_moe_down-55): 206 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 358 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 394 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 370 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 402 us MUL_MAT_ID(ffn_moe_down-59): 206 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 202 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 354 us MUL_MAT_ID(ffn_moe_down-61): 180 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 339 us MUL_MAT_ID(ffn_moe_down-62): 471 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 579 us MUL_MAT_ID(ffn_moe_down-63): 263 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 356 us MUL_MAT_ID(ffn_moe_down-64): 200 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 389 us MUL_MAT_ID(ffn_moe_down-65): 184 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 378 us MUL_MAT_ID(ffn_moe_down-66): 236 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 406 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 406 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 398 us MUL_MAT_ID(ffn_moe_down-69): 188 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 373 us MUL_MAT_ID(ffn_moe_down-70): 191 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 396 us MUL_MAT_ID(ffn_moe_down-71): 186 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 398 us MUL_MAT_ID(ffn_moe_down-72): 189 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 429 us MUL_MAT_ID(ffn_moe_down-73): 213 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 372 us MUL_MAT_ID(ffn_moe_down-74): 387 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 406 us MUL_MAT_ID(ffn_moe_down-75): 398 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 376 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 391 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 390 us MUL_MAT_ID(ffn_moe_down-78): 208 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 384 us MUL_MAT_ID(ffn_moe_down-79): 178 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 360 us MUL_MAT_ID(ffn_moe_down-80): 387 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 368 us MUL_MAT_ID(ffn_moe_down-81): 311 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 487 us MUL_MAT_ID(ffn_moe_down-82): 270 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 381 us MUL_MAT_ID(ffn_moe_down-83): 266 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 375 us MUL_MAT_ID(ffn_moe_down-84): 246 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 371 us MUL_MAT_ID(ffn_moe_down-85): 409 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 394 us MUL_MAT_ID(ffn_moe_down-86): 202 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 375 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 367 us MUL_MAT_ID(ffn_moe_down-88): 544 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 392 us MUL_MAT_ID(ffn_moe_down-89): 185 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 405 us MUL_MAT_ID(ffn_moe_down-90): 426 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 371 us MUL_MAT_ID(ffn_moe_down-91): 261 us ggml_barrier(...): 50 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 411 us MUL_MAT_ID(ffn_moe_down-25): 186 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 362 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 341 us MUL_MAT_ID(ffn_moe_down-27): 203 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 375 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 402 us MUL_MAT_ID(ffn_moe_down-29): 210 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 365 us MUL_MAT_ID(ffn_moe_down-30): 201 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 409 us MUL_MAT_ID(ffn_moe_down-31): 471 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 397 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 379 us MUL_MAT_ID(ffn_moe_down-33): 191 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 388 us MUL_MAT_ID(ffn_moe_down-34): 207 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 405 us MUL_MAT_ID(ffn_moe_down-35): 303 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 366 us MUL_MAT_ID(ffn_moe_down-36): 214 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 396 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 355 us MUL_MAT_ID(ffn_moe_down-38): 364 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 385 us MUL_MAT_ID(ffn_moe_down-39): 245 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 396 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 370 us MUL_MAT_ID(ffn_moe_down-41): 173 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 354 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 367 us MUL_MAT_ID(ffn_moe_down-43): 199 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 372 us MUL_MAT_ID(ffn_moe_down-44): 200 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 348 us MUL_MAT_ID(ffn_moe_down-45): 201 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 353 us MUL_MAT_ID(ffn_moe_down-46): 188 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 403 us MUL_MAT_ID(ffn_moe_down-47): 189 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 364 us MUL_MAT_ID(ffn_moe_down-48): 191 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 389 us MUL_MAT_ID(ffn_moe_down-49): 201 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 394 us MUL_MAT_ID(ffn_moe_down-50): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 384 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 379 us MUL_MAT_ID(ffn_moe_down-52): 381 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 403 us MUL_MAT_ID(ffn_moe_down-53): 214 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 396 us MUL_MAT_ID(ffn_moe_down-54): 203 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 372 us MUL_MAT_ID(ffn_moe_down-55): 216 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 379 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 355 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 387 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 373 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 384 us MUL_MAT_ID(ffn_moe_down-60): 203 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 387 us MUL_MAT_ID(ffn_moe_down-61): 352 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 269 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 397 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 417 us MUL_MAT_ID(ffn_moe_down-64): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 379 us MUL_MAT_ID(ffn_moe_down-65): 203 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 353 us MUL_MAT_ID(ffn_moe_down-66): 194 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 420 us MUL_MAT_ID(ffn_moe_down-67): 206 us ggml_barrier(...): 365 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 394 us MUL_MAT_ID(ffn_moe_down-68): 544 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 364 us MUL_MAT_ID(ffn_moe_down-69): 215 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 388 us MUL_MAT_ID(ffn_moe_down-70): 396 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 384 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 389 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 381 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 360 us MUL_MAT_ID(ffn_moe_down-74): 503 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 345 us MUL_MAT_ID(ffn_moe_down-75): 325 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 354 us MUL_MAT_ID(ffn_moe_down-76): 354 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 420 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 394 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 406 us MUL_MAT_ID(ffn_moe_down-79): 210 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 391 us MUL_MAT_ID(ffn_moe_down-80): 194 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 424 us MUL_MAT_ID(ffn_moe_down-81): 455 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 405 us MUL_MAT_ID(ffn_moe_down-82): 375 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 406 us MUL_MAT_ID(ffn_moe_down-83): 189 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 388 us MUL_MAT_ID(ffn_moe_down-84): 226 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 398 us MUL_MAT_ID(ffn_moe_down-85): 588 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 412 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 389 us MUL_MAT_ID(ffn_moe_down-87): 536 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 378 us MUL_MAT_ID(ffn_moe_down-88): 205 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 401 us MUL_MAT_ID(ffn_moe_down-89): 181 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 387 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 409 us MUL_MAT_ID(ffn_moe_down-91): 233 us ggml_barrier(...): 35 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 385 us MUL_MAT_ID(ffn_moe_down-25): 430 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 370 us MUL_MAT_ID(ffn_moe_down-26): 303 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 370 us MUL_MAT_ID(ffn_moe_down-27): 180 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 388 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 360 us MUL_MAT_ID(ffn_moe_down-29): 396 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 382 us MUL_MAT_ID(ffn_moe_down-30): 180 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 411 us MUL_MAT_ID(ffn_moe_down-31): 199 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 370 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 347 us MUL_MAT_ID(ffn_moe_down-33): 205 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 408 us MUL_MAT_ID(ffn_moe_down-34): 281 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 414 us MUL_MAT_ID(ffn_moe_down-35): 185 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 358 us MUL_MAT_ID(ffn_moe_down-36): 191 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 388 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 368 us MUL_MAT_ID(ffn_moe_down-38): 199 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 416 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 416 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 401 us MUL_MAT_ID(ffn_moe_down-41): 200 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 348 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 385 us MUL_MAT_ID(ffn_moe_down-43): 336 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 363 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 417 us MUL_MAT_ID(ffn_moe_down-45): 183 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 357 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 392 us MUL_MAT_ID(ffn_moe_down-47): 346 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 410 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 418 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 380 us MUL_MAT_ID(ffn_moe_down-50): 182 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 372 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 422 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 388 us MUL_MAT_ID(ffn_moe_down-53): 195 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 406 us MUL_MAT_ID(ffn_moe_down-54): 239 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 379 us MUL_MAT_ID(ffn_moe_down-55): 295 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 375 us MUL_MAT_ID(ffn_moe_down-56): 201 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 379 us MUL_MAT_ID(ffn_moe_down-57): 208 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 400 us MUL_MAT_ID(ffn_moe_down-58): 188 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 348 us MUL_MAT_ID(ffn_moe_down-59): 329 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 382 us MUL_MAT_ID(ffn_moe_down-60): 224 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 391 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 390 us MUL_MAT_ID(ffn_moe_down-62): 372 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 383 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 366 us MUL_MAT_ID(ffn_moe_down-64): 336 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 385 us MUL_MAT_ID(ffn_moe_down-65): 341 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 388 us MUL_MAT_ID(ffn_moe_down-66): 206 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 397 us MUL_MAT_ID(ffn_moe_down-67): 399 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 391 us MUL_MAT_ID(ffn_moe_down-68): 511 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 365 us MUL_MAT_ID(ffn_moe_down-69): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 365 us MUL_MAT_ID(ffn_moe_down-70): 207 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 367 us MUL_MAT_ID(ffn_moe_down-71): 359 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 362 us MUL_MAT_ID(ffn_moe_down-72): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 364 us MUL_MAT_ID(ffn_moe_down-73): 203 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 385 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 383 us MUL_MAT_ID(ffn_moe_down-75): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 411 us MUL_MAT_ID(ffn_moe_down-76): 690 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 400 us MUL_MAT_ID(ffn_moe_down-77): 184 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 380 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 387 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 362 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 360 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 385 us MUL_MAT_ID(ffn_moe_down-82): 189 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 408 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 355 us MUL_MAT_ID(ffn_moe_down-84): 405 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 386 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 401 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 378 us MUL_MAT_ID(ffn_moe_down-87): 228 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 372 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 384 us MUL_MAT_ID(ffn_moe_down-89): 197 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 385 us MUL_MAT_ID(ffn_moe_down-90): 647 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 396 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 14 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 387 us MUL_MAT_ID(ffn_moe_down-25): 176 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 494 us MUL_MAT_ID(ffn_moe_down-26): 210 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 371 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 409 us MUL_MAT_ID(ffn_moe_down-28): 199 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 364 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 372 us MUL_MAT_ID(ffn_moe_down-30): 188 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 350 us MUL_MAT_ID(ffn_moe_down-31): 409 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 412 us MUL_MAT_ID(ffn_moe_down-32): 252 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 354 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 412 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 374 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 357 us MUL_MAT_ID(ffn_moe_down-36): 226 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 375 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 416 us MUL_MAT_ID(ffn_moe_down-38): 185 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 389 us MUL_MAT_ID(ffn_moe_down-39): 614 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 373 us MUL_MAT_ID(ffn_moe_down-40): 235 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 356 us MUL_MAT_ID(ffn_moe_down-41): 215 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 389 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 394 us MUL_MAT_ID(ffn_moe_down-43): 206 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 402 us MUL_MAT_ID(ffn_moe_down-44): 380 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 400 us MUL_MAT_ID(ffn_moe_down-45): 172 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 396 us MUL_MAT_ID(ffn_moe_down-46): 171 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 390 us MUL_MAT_ID(ffn_moe_down-47): 219 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 368 us MUL_MAT_ID(ffn_moe_down-48): 524 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 395 us MUL_MAT_ID(ffn_moe_down-49): 212 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 382 us MUL_MAT_ID(ffn_moe_down-50): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 375 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 393 us MUL_MAT_ID(ffn_moe_down-52): 201 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 406 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 398 us MUL_MAT_ID(ffn_moe_down-54): 229 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 370 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 370 us MUL_MAT_ID(ffn_moe_down-56): 201 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 377 us MUL_MAT_ID(ffn_moe_down-57): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 420 us MUL_MAT_ID(ffn_moe_down-58): 210 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 365 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 340 us MUL_MAT_ID(ffn_moe_down-60): 174 us ggml_barrier(...): 352 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 375 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 361 us MUL_MAT_ID(ffn_moe_down-62): 208 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 388 us MUL_MAT_ID(ffn_moe_down-64): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 398 us MUL_MAT_ID(ffn_moe_down-65): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 376 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 383 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 353 us MUL_MAT_ID(ffn_moe_down-68): 191 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 382 us MUL_MAT_ID(ffn_moe_down-69): 184 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 362 us MUL_MAT_ID(ffn_moe_down-70): 211 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 359 us MUL_MAT_ID(ffn_moe_down-71): 208 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 376 us MUL_MAT_ID(ffn_moe_down-72): 390 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 376 us MUL_MAT_ID(ffn_moe_down-73): 223 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 480 us MUL_MAT_ID(ffn_moe_down-74): 335 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 368 us MUL_MAT_ID(ffn_moe_down-75): 215 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 487 us MUL_MAT_ID(ffn_moe_down-76): 179 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 395 us MUL_MAT_ID(ffn_moe_down-77): 360 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 397 us MUL_MAT_ID(ffn_moe_down-78): 391 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 364 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 542 us MUL_MAT_ID(ffn_moe_down-80): 205 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 386 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 379 us MUL_MAT_ID(ffn_moe_down-82): 175 us ggml_barrier(...): 352 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 392 us MUL_MAT_ID(ffn_moe_down-83): 188 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 380 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 405 us MUL_MAT_ID(ffn_moe_down-85): 469 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 363 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 512 us MUL_MAT_ID(ffn_moe_down-87): 186 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 388 us MUL_MAT_ID(ffn_moe_down-88): 225 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 394 us MUL_MAT_ID(ffn_moe_down-89): 191 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 370 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 378 us MUL_MAT_ID(ffn_moe_down-91): 254 us ggml_barrier(...): 32 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 370 us MUL_MAT_ID(ffn_moe_down-25): 253 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 450 us MUL_MAT_ID(ffn_moe_down-26): 510 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 386 us MUL_MAT_ID(ffn_moe_down-27): 214 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 384 us MUL_MAT_ID(ffn_moe_down-28): 188 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 381 us MUL_MAT_ID(ffn_moe_down-29): 210 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 485 us MUL_MAT_ID(ffn_moe_down-30): 188 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 412 us MUL_MAT_ID(ffn_moe_down-31): 515 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 395 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 415 us MUL_MAT_ID(ffn_moe_down-33): 219 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 395 us MUL_MAT_ID(ffn_moe_down-34): 189 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 394 us MUL_MAT_ID(ffn_moe_down-35): 202 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 383 us MUL_MAT_ID(ffn_moe_down-36): 205 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 389 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 390 us MUL_MAT_ID(ffn_moe_down-38): 347 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 395 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 429 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 417 us MUL_MAT_ID(ffn_moe_down-41): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 434 us MUL_MAT_ID(ffn_moe_down-42): 512 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 362 us MUL_MAT_ID(ffn_moe_down-43): 171 us ggml_barrier(...): 355 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 362 us MUL_MAT_ID(ffn_moe_down-44): 221 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 422 us MUL_MAT_ID(ffn_moe_down-45): 222 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 384 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 386 us MUL_MAT_ID(ffn_moe_down-47): 182 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 396 us MUL_MAT_ID(ffn_moe_down-48): 221 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 353 us MUL_MAT_ID(ffn_moe_down-49): 185 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 386 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 413 us MUL_MAT_ID(ffn_moe_down-51): 208 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 386 us MUL_MAT_ID(ffn_moe_down-52): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 434 us MUL_MAT_ID(ffn_moe_down-53): 208 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 392 us MUL_MAT_ID(ffn_moe_down-54): 180 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 383 us MUL_MAT_ID(ffn_moe_down-55): 186 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 403 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 395 us MUL_MAT_ID(ffn_moe_down-57): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 407 us MUL_MAT_ID(ffn_moe_down-58): 202 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 503 us MUL_MAT_ID(ffn_moe_down-59): 248 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 407 us MUL_MAT_ID(ffn_moe_down-60): 254 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 375 us MUL_MAT_ID(ffn_moe_down-61): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 410 us MUL_MAT_ID(ffn_moe_down-62): 433 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 349 us MUL_MAT_ID(ffn_moe_down-63): 355 us ggml_barrier(...): 326 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 377 us MUL_MAT_ID(ffn_moe_down-64): 424 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 359 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 354 us MUL_MAT_ID(ffn_moe_down-66): 278 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 412 us MUL_MAT_ID(ffn_moe_down-67): 202 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 371 us MUL_MAT_ID(ffn_moe_down-68): 188 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 343 us MUL_MAT_ID(ffn_moe_down-69): 198 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 396 us MUL_MAT_ID(ffn_moe_down-70): 187 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 452 us MUL_MAT_ID(ffn_moe_down-71): 310 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 386 us MUL_MAT_ID(ffn_moe_down-72): 399 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 379 us MUL_MAT_ID(ffn_moe_down-73): 247 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 385 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 401 us MUL_MAT_ID(ffn_moe_down-75): 199 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 373 us MUL_MAT_ID(ffn_moe_down-76): 215 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 379 us MUL_MAT_ID(ffn_moe_down-77): 216 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 403 us MUL_MAT_ID(ffn_moe_down-78): 375 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 375 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 384 us MUL_MAT_ID(ffn_moe_down-80): 212 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 395 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 402 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 404 us MUL_MAT_ID(ffn_moe_down-83): 193 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 369 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 442 us MUL_MAT_ID(ffn_moe_down-85): 249 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 409 us MUL_MAT_ID(ffn_moe_down-86): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 371 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 376 us MUL_MAT_ID(ffn_moe_down-88): 197 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 396 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 390 us MUL_MAT_ID(ffn_moe_down-90): 421 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 369 us MUL_MAT_ID(ffn_moe_down-91): 243 us ggml_barrier(...): 36 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 429 us MUL_MAT_ID(ffn_moe_down-25): 182 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 383 us MUL_MAT_ID(ffn_moe_down-26): 185 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 395 us MUL_MAT_ID(ffn_moe_down-27): 202 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 402 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 421 us MUL_MAT_ID(ffn_moe_down-29): 190 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 400 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 411 us MUL_MAT_ID(ffn_moe_down-31): 200 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 376 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 387 us MUL_MAT_ID(ffn_moe_down-33): 197 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 397 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 386 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 419 us MUL_MAT_ID(ffn_moe_down-36): 241 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 374 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 377 us MUL_MAT_ID(ffn_moe_down-38): 200 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 373 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 386 us MUL_MAT_ID(ffn_moe_down-40): 242 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 380 us MUL_MAT_ID(ffn_moe_down-41): 208 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 386 us MUL_MAT_ID(ffn_moe_down-42): 393 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 387 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 398 us MUL_MAT_ID(ffn_moe_down-44): 185 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 382 us MUL_MAT_ID(ffn_moe_down-45): 208 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 371 us MUL_MAT_ID(ffn_moe_down-46): 497 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 379 us MUL_MAT_ID(ffn_moe_down-47): 240 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 351 us MUL_MAT_ID(ffn_moe_down-48): 201 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 371 us MUL_MAT_ID(ffn_moe_down-49): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 374 us MUL_MAT_ID(ffn_moe_down-50): 219 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 363 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 379 us MUL_MAT_ID(ffn_moe_down-52): 276 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 387 us MUL_MAT_ID(ffn_moe_down-53): 199 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 377 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 363 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 375 us MUL_MAT_ID(ffn_moe_down-56): 267 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 369 us MUL_MAT_ID(ffn_moe_down-57): 196 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 394 us MUL_MAT_ID(ffn_moe_down-58): 186 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 365 us MUL_MAT_ID(ffn_moe_down-59): 223 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 376 us MUL_MAT_ID(ffn_moe_down-60): 187 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 378 us MUL_MAT_ID(ffn_moe_down-61): 206 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 380 us MUL_MAT_ID(ffn_moe_down-62): 368 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 406 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 372 us MUL_MAT_ID(ffn_moe_down-64): 186 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 478 us MUL_MAT_ID(ffn_moe_down-65): 219 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 415 us MUL_MAT_ID(ffn_moe_down-66): 184 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 369 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 386 us MUL_MAT_ID(ffn_moe_down-68): 484 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 375 us MUL_MAT_ID(ffn_moe_down-69): 454 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 360 us MUL_MAT_ID(ffn_moe_down-70): 241 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 426 us MUL_MAT_ID(ffn_moe_down-71): 269 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 385 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 365 us MUL_MAT_ID(ffn_moe_down-73): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 367 us MUL_MAT_ID(ffn_moe_down-74): 210 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 381 us MUL_MAT_ID(ffn_moe_down-75): 533 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 391 us MUL_MAT_ID(ffn_moe_down-76): 188 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 383 us MUL_MAT_ID(ffn_moe_down-77): 533 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 407 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 383 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 400 us MUL_MAT_ID(ffn_moe_down-80): 202 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 401 us MUL_MAT_ID(ffn_moe_down-81): 392 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 201 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 406 us MUL_MAT_ID(ffn_moe_down-83): 403 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 379 us MUL_MAT_ID(ffn_moe_down-84): 588 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 369 us MUL_MAT_ID(ffn_moe_down-85): 221 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 378 us MUL_MAT_ID(ffn_moe_down-86): 256 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 371 us MUL_MAT_ID(ffn_moe_down-87): 385 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 381 us MUL_MAT_ID(ffn_moe_down-88): 182 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 414 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 348 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 410 us MUL_MAT_ID(ffn_moe_down-91): 270 us ggml_barrier(...): 30 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 412 us MUL_MAT_ID(ffn_moe_down-25): 182 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 384 us MUL_MAT_ID(ffn_moe_down-26): 199 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 381 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 364 us MUL_MAT_ID(ffn_moe_down-28): 208 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 414 us MUL_MAT_ID(ffn_moe_down-29): 188 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 426 us MUL_MAT_ID(ffn_moe_down-30): 346 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 384 us MUL_MAT_ID(ffn_moe_down-31): 166 us ggml_barrier(...): 322 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 369 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 423 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 403 us MUL_MAT_ID(ffn_moe_down-34): 235 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 379 us MUL_MAT_ID(ffn_moe_down-35): 195 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 423 us MUL_MAT_ID(ffn_moe_down-36): 197 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 410 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 381 us MUL_MAT_ID(ffn_moe_down-38): 197 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 407 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 395 us MUL_MAT_ID(ffn_moe_down-40): 642 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 376 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 391 us MUL_MAT_ID(ffn_moe_down-42): 199 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 383 us MUL_MAT_ID(ffn_moe_down-43): 363 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 385 us MUL_MAT_ID(ffn_moe_down-44): 203 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 389 us MUL_MAT_ID(ffn_moe_down-45): 378 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 414 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 385 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 381 us MUL_MAT_ID(ffn_moe_down-48): 208 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 403 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 408 us MUL_MAT_ID(ffn_moe_down-50): 220 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 400 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 385 us MUL_MAT_ID(ffn_moe_down-52): 274 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 353 us MUL_MAT_ID(ffn_moe_down-53): 199 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 360 us MUL_MAT_ID(ffn_moe_down-54): 192 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 394 us MUL_MAT_ID(ffn_moe_down-55): 204 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 368 us MUL_MAT_ID(ffn_moe_down-56): 382 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 418 us MUL_MAT_ID(ffn_moe_down-57): 204 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 462 us MUL_MAT_ID(ffn_moe_down-58): 216 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 407 us MUL_MAT_ID(ffn_moe_down-59): 216 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 394 us MUL_MAT_ID(ffn_moe_down-60): 213 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 364 us MUL_MAT_ID(ffn_moe_down-61): 209 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 382 us MUL_MAT_ID(ffn_moe_down-62): 191 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 385 us MUL_MAT_ID(ffn_moe_down-63): 207 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 387 us MUL_MAT_ID(ffn_moe_down-64): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 378 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 389 us MUL_MAT_ID(ffn_moe_down-66): 367 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 373 us MUL_MAT_ID(ffn_moe_down-67): 206 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 358 us MUL_MAT_ID(ffn_moe_down-68): 205 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 343 us MUL_MAT_ID(ffn_moe_down-69): 228 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 495 us MUL_MAT_ID(ffn_moe_down-70): 172 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 382 us MUL_MAT_ID(ffn_moe_down-71): 439 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 392 us MUL_MAT_ID(ffn_moe_down-73): 222 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 338 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 384 us MUL_MAT_ID(ffn_moe_down-75): 234 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 402 us MUL_MAT_ID(ffn_moe_down-76): 199 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 376 us MUL_MAT_ID(ffn_moe_down-77): 191 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 406 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 397 us MUL_MAT_ID(ffn_moe_down-79): 230 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 383 us MUL_MAT_ID(ffn_moe_down-80): 226 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 404 us MUL_MAT_ID(ffn_moe_down-82): 219 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 377 us MUL_MAT_ID(ffn_moe_down-83): 204 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 385 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 387 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 402 us MUL_MAT_ID(ffn_moe_down-86): 195 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 417 us MUL_MAT_ID(ffn_moe_down-87): 200 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 410 us MUL_MAT_ID(ffn_moe_down-88): 176 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 376 us MUL_MAT_ID(ffn_moe_down-89): 187 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 359 us MUL_MAT_ID(ffn_moe_down-90): 250 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 390 us MUL_MAT_ID(ffn_moe_down-91): 245 us ggml_barrier(...): 17 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 392 us MUL_MAT_ID(ffn_moe_down-25): 207 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 397 us MUL_MAT_ID(ffn_moe_down-26): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 383 us MUL_MAT_ID(ffn_moe_down-27): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 406 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 394 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 376 us MUL_MAT_ID(ffn_moe_down-31): 211 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 387 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 385 us MUL_MAT_ID(ffn_moe_down-33): 297 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 398 us MUL_MAT_ID(ffn_moe_down-34): 339 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 385 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 386 us MUL_MAT_ID(ffn_moe_down-36): 218 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 368 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 403 us MUL_MAT_ID(ffn_moe_down-38): 185 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 394 us MUL_MAT_ID(ffn_moe_down-39): 225 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 379 us MUL_MAT_ID(ffn_moe_down-40): 372 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 503 us MUL_MAT_ID(ffn_moe_down-41): 203 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 524 us MUL_MAT_ID(ffn_moe_down-42): 208 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 405 us MUL_MAT_ID(ffn_moe_down-43): 203 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 413 us MUL_MAT_ID(ffn_moe_down-44): 204 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 394 us MUL_MAT_ID(ffn_moe_down-45): 196 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 386 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 536 us MUL_MAT_ID(ffn_moe_down-47): 208 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 380 us MUL_MAT_ID(ffn_moe_down-48): 369 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 378 us MUL_MAT_ID(ffn_moe_down-49): 269 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 406 us MUL_MAT_ID(ffn_moe_down-50): 184 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 438 us MUL_MAT_ID(ffn_moe_down-51): 206 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 384 us MUL_MAT_ID(ffn_moe_down-52): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 415 us MUL_MAT_ID(ffn_moe_down-53): 207 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 387 us MUL_MAT_ID(ffn_moe_down-54): 192 us ggml_barrier(...): 315 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 383 us MUL_MAT_ID(ffn_moe_down-55): 489 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 421 us MUL_MAT_ID(ffn_moe_down-56): 189 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 362 us MUL_MAT_ID(ffn_moe_down-57): 232 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 385 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 388 us MUL_MAT_ID(ffn_moe_down-59): 188 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 382 us MUL_MAT_ID(ffn_moe_down-60): 207 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 364 us MUL_MAT_ID(ffn_moe_down-61): 188 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 405 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 347 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 395 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 353 us MUL_MAT_ID(ffn_moe_down-65): 200 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 394 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 327 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 356 us MUL_MAT_ID(ffn_moe_down-67): 213 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 362 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 365 us MUL_MAT_ID(ffn_moe_down-69): 205 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 351 us MUL_MAT_ID(ffn_moe_down-70): 374 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 381 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 357 us MUL_MAT_ID(ffn_moe_down-72): 208 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 356 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 364 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 379 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 400 us MUL_MAT_ID(ffn_moe_down-76): 378 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 483 us MUL_MAT_ID(ffn_moe_down-77): 224 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 364 us MUL_MAT_ID(ffn_moe_down-78): 202 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 384 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 382 us MUL_MAT_ID(ffn_moe_down-80): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 372 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 357 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 384 us MUL_MAT_ID(ffn_moe_down-83): 206 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 380 us MUL_MAT_ID(ffn_moe_down-84): 432 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 405 us MUL_MAT_ID(ffn_moe_down-85): 355 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 358 us MUL_MAT_ID(ffn_moe_down-86): 328 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 379 us MUL_MAT_ID(ffn_moe_down-87): 388 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 375 us MUL_MAT_ID(ffn_moe_down-88): 202 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 378 us MUL_MAT_ID(ffn_moe_down-89): 186 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 397 us MUL_MAT_ID(ffn_moe_down-90): 257 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 355 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 55 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 365 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 363 us MUL_MAT_ID(ffn_moe_down-27): 198 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 393 us MUL_MAT_ID(ffn_moe_down-28): 428 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 402 us MUL_MAT_ID(ffn_moe_down-29): 214 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 389 us MUL_MAT_ID(ffn_moe_down-30): 164 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 376 us MUL_MAT_ID(ffn_moe_down-31): 220 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 409 us MUL_MAT_ID(ffn_moe_down-32): 254 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 351 us MUL_MAT_ID(ffn_moe_down-33): 196 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 400 us MUL_MAT_ID(ffn_moe_down-34): 320 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 407 us MUL_MAT_ID(ffn_moe_down-35): 212 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 366 us MUL_MAT_ID(ffn_moe_down-36): 202 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 388 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 382 us MUL_MAT_ID(ffn_moe_down-38): 394 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 411 us MUL_MAT_ID(ffn_moe_down-39): 333 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 366 us MUL_MAT_ID(ffn_moe_down-40): 439 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 425 us MUL_MAT_ID(ffn_moe_down-41): 238 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 366 us MUL_MAT_ID(ffn_moe_down-42): 172 us ggml_barrier(...): 343 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 365 us MUL_MAT_ID(ffn_moe_down-43): 204 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 392 us MUL_MAT_ID(ffn_moe_down-44): 201 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 344 us MUL_MAT_ID(ffn_moe_down-45): 214 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 386 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 364 us MUL_MAT_ID(ffn_moe_down-47): 488 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 403 us MUL_MAT_ID(ffn_moe_down-48): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 367 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 396 us MUL_MAT_ID(ffn_moe_down-50): 208 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 361 us MUL_MAT_ID(ffn_moe_down-51): 501 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 379 us MUL_MAT_ID(ffn_moe_down-52): 206 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 353 us MUL_MAT_ID(ffn_moe_down-53): 200 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 409 us MUL_MAT_ID(ffn_moe_down-54): 293 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 370 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 386 us MUL_MAT_ID(ffn_moe_down-57): 185 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 355 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 358 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 378 us MUL_MAT_ID(ffn_moe_down-60): 369 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 379 us MUL_MAT_ID(ffn_moe_down-61): 194 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 402 us MUL_MAT_ID(ffn_moe_down-62): 205 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 372 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 358 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 415 us MUL_MAT_ID(ffn_moe_down-65): 208 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 394 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 353 us MUL_MAT_ID(ffn_moe_down-67): 255 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 394 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 352 us MUL_MAT_ID(ffn_moe_down-69): 197 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 356 us MUL_MAT_ID(ffn_moe_down-70): 189 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 416 us MUL_MAT_ID(ffn_moe_down-71): 191 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 357 us MUL_MAT_ID(ffn_moe_down-72): 320 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 201 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 345 us MUL_MAT_ID(ffn_moe_down-74): 205 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 391 us MUL_MAT_ID(ffn_moe_down-75): 186 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 383 us MUL_MAT_ID(ffn_moe_down-76): 216 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 357 us MUL_MAT_ID(ffn_moe_down-77): 196 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 382 us MUL_MAT_ID(ffn_moe_down-78): 214 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 407 us MUL_MAT_ID(ffn_moe_down-79): 182 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 379 us MUL_MAT_ID(ffn_moe_down-80): 209 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 395 us MUL_MAT_ID(ffn_moe_down-81): 383 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 358 us MUL_MAT_ID(ffn_moe_down-82): 483 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 379 us MUL_MAT_ID(ffn_moe_down-83): 252 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 375 us MUL_MAT_ID(ffn_moe_down-84): 225 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 348 us MUL_MAT_ID(ffn_moe_down-85): 758 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 384 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 379 us MUL_MAT_ID(ffn_moe_down-87): 199 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 471 us MUL_MAT_ID(ffn_moe_down-88): 309 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 384 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 359 us MUL_MAT_ID(ffn_moe_down-90): 238 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 410 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 12 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 392 us MUL_MAT_ID(ffn_moe_down-25): 181 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 350 us MUL_MAT_ID(ffn_moe_down-26): 191 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 364 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 365 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 335 us MUL_MAT_ID(ffn_moe_down-29): 534 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 353 us MUL_MAT_ID(ffn_moe_down-30): 215 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 404 us MUL_MAT_ID(ffn_moe_down-31): 183 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 416 us MUL_MAT_ID(ffn_moe_down-32): 346 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 421 us MUL_MAT_ID(ffn_moe_down-33): 208 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 376 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 423 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 374 us MUL_MAT_ID(ffn_moe_down-36): 184 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 401 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 376 us MUL_MAT_ID(ffn_moe_down-38): 204 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 400 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 404 us MUL_MAT_ID(ffn_moe_down-40): 237 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 345 us MUL_MAT_ID(ffn_moe_down-41): 362 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 397 us MUL_MAT_ID(ffn_moe_down-42): 206 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 382 us MUL_MAT_ID(ffn_moe_down-43): 184 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 393 us MUL_MAT_ID(ffn_moe_down-44): 186 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 395 us MUL_MAT_ID(ffn_moe_down-45): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 365 us MUL_MAT_ID(ffn_moe_down-46): 251 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 358 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 534 us MUL_MAT_ID(ffn_moe_down-48): 315 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 366 us MUL_MAT_ID(ffn_moe_down-49): 201 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 377 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 393 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 363 us MUL_MAT_ID(ffn_moe_down-52): 383 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 372 us MUL_MAT_ID(ffn_moe_down-53): 230 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 356 us MUL_MAT_ID(ffn_moe_down-54): 198 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 366 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 493 us MUL_MAT_ID(ffn_moe_down-56): 406 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 392 us MUL_MAT_ID(ffn_moe_down-57): 180 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 399 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 416 us MUL_MAT_ID(ffn_moe_down-59): 344 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 380 us MUL_MAT_ID(ffn_moe_down-60): 189 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 412 us MUL_MAT_ID(ffn_moe_down-61): 249 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 396 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 390 us MUL_MAT_ID(ffn_moe_down-63): 205 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 406 us MUL_MAT_ID(ffn_moe_down-64): 196 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 367 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 410 us MUL_MAT_ID(ffn_moe_down-66): 198 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 396 us MUL_MAT_ID(ffn_moe_down-67): 170 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 440 us MUL_MAT_ID(ffn_moe_down-68): 293 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 380 us MUL_MAT_ID(ffn_moe_down-69): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 382 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 377 us MUL_MAT_ID(ffn_moe_down-71): 232 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 532 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 373 us MUL_MAT_ID(ffn_moe_down-73): 211 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 402 us MUL_MAT_ID(ffn_moe_down-74): 348 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 405 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 372 us MUL_MAT_ID(ffn_moe_down-76): 312 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 399 us MUL_MAT_ID(ffn_moe_down-77): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 370 us MUL_MAT_ID(ffn_moe_down-78): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 381 us MUL_MAT_ID(ffn_moe_down-79): 212 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 389 us MUL_MAT_ID(ffn_moe_down-80): 208 us ggml_barrier(...): 251 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 401 us MUL_MAT_ID(ffn_moe_down-81): 495 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 506 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 392 us MUL_MAT_ID(ffn_moe_down-83): 193 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 383 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 366 us MUL_MAT_ID(ffn_moe_down-85): 723 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 401 us MUL_MAT_ID(ffn_moe_down-86): 216 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 405 us MUL_MAT_ID(ffn_moe_down-87): 197 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 367 us MUL_MAT_ID(ffn_moe_down-88): 201 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 370 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 364 us MUL_MAT_ID(ffn_moe_down-90): 658 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 378 us MUL_MAT_ID(ffn_moe_down-91): 284 us ggml_barrier(...): 47 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 415 us MUL_MAT_ID(ffn_moe_down-25): 181 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 382 us MUL_MAT_ID(ffn_moe_down-26): 206 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 407 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 359 us MUL_MAT_ID(ffn_moe_down-28): 193 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 393 us MUL_MAT_ID(ffn_moe_down-29): 197 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 422 us MUL_MAT_ID(ffn_moe_down-30): 184 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 419 us MUL_MAT_ID(ffn_moe_down-31): 322 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 372 us MUL_MAT_ID(ffn_moe_down-32): 206 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 372 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 392 us MUL_MAT_ID(ffn_moe_down-34): 363 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 362 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 426 us MUL_MAT_ID(ffn_moe_down-36): 374 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 380 us MUL_MAT_ID(ffn_moe_down-37): 258 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 377 us MUL_MAT_ID(ffn_moe_down-38): 431 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 399 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 358 us MUL_MAT_ID(ffn_moe_down-40): 247 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 398 us MUL_MAT_ID(ffn_moe_down-41): 209 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 370 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 429 us MUL_MAT_ID(ffn_moe_down-43): 206 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 402 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 365 us MUL_MAT_ID(ffn_moe_down-45): 203 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 369 us MUL_MAT_ID(ffn_moe_down-46): 211 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 383 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 383 us MUL_MAT_ID(ffn_moe_down-48): 215 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 412 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 371 us MUL_MAT_ID(ffn_moe_down-50): 373 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 412 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 2 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 372 us MUL_MAT_ID(ffn_moe_down-52): 223 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 386 us MUL_MAT_ID(ffn_moe_down-53): 417 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 359 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 405 us MUL_MAT_ID(ffn_moe_down-55): 199 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 408 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 391 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 379 us MUL_MAT_ID(ffn_moe_down-57): 182 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 399 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 358 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 373 us MUL_MAT_ID(ffn_moe_down-60): 193 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 383 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 355 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 350 us MUL_MAT_ID(ffn_moe_down-63): 195 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 398 us MUL_MAT_ID(ffn_moe_down-64): 369 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 349 us MUL_MAT_ID(ffn_moe_down-65): 237 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 381 us MUL_MAT_ID(ffn_moe_down-66): 352 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 372 us MUL_MAT_ID(ffn_moe_down-67): 403 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 365 us MUL_MAT_ID(ffn_moe_down-68): 247 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 416 us MUL_MAT_ID(ffn_moe_down-69): 203 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 371 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 384 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 403 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 398 us MUL_MAT_ID(ffn_moe_down-73): 188 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 384 us MUL_MAT_ID(ffn_moe_down-74): 191 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 361 us MUL_MAT_ID(ffn_moe_down-75): 219 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 373 us MUL_MAT_ID(ffn_moe_down-76): 203 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 406 us MUL_MAT_ID(ffn_moe_down-77): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 408 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 376 us MUL_MAT_ID(ffn_moe_down-79): 372 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 392 us MUL_MAT_ID(ffn_moe_down-80): 227 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 394 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 412 us MUL_MAT_ID(ffn_moe_down-82): 189 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 411 us MUL_MAT_ID(ffn_moe_down-83): 184 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 397 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 382 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 419 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 374 us MUL_MAT_ID(ffn_moe_down-87): 204 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 385 us MUL_MAT_ID(ffn_moe_down-88): 387 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 341 us MUL_MAT_ID(ffn_moe_down-89): 394 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 389 us MUL_MAT_ID(ffn_moe_down-90): 248 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 354 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 437 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 398 us MUL_MAT_ID(ffn_moe_down-26): 183 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 391 us MUL_MAT_ID(ffn_moe_down-27): 211 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 365 us MUL_MAT_ID(ffn_moe_down-28): 199 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 409 us MUL_MAT_ID(ffn_moe_down-29): 315 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 401 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 417 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 356 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 371 us MUL_MAT_ID(ffn_moe_down-33): 266 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 385 us MUL_MAT_ID(ffn_moe_down-34): 187 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 409 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 375 us MUL_MAT_ID(ffn_moe_down-36): 218 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 360 us MUL_MAT_ID(ffn_moe_down-37): 215 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 359 us MUL_MAT_ID(ffn_moe_down-38): 184 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 375 us MUL_MAT_ID(ffn_moe_down-39): 658 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 360 us MUL_MAT_ID(ffn_moe_down-40): 238 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 385 us MUL_MAT_ID(ffn_moe_down-41): 203 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 416 us MUL_MAT_ID(ffn_moe_down-42): 413 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 370 us MUL_MAT_ID(ffn_moe_down-43): 484 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 374 us MUL_MAT_ID(ffn_moe_down-44): 325 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 430 us MUL_MAT_ID(ffn_moe_down-45): 210 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 382 us MUL_MAT_ID(ffn_moe_down-46): 208 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 389 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 380 us MUL_MAT_ID(ffn_moe_down-48): 234 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 371 us MUL_MAT_ID(ffn_moe_down-49): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 358 us MUL_MAT_ID(ffn_moe_down-50): 186 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 412 us MUL_MAT_ID(ffn_moe_down-51): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 393 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 392 us MUL_MAT_ID(ffn_moe_down-53): 192 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 373 us MUL_MAT_ID(ffn_moe_down-54): 234 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 402 us MUL_MAT_ID(ffn_moe_down-55): 279 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 403 us MUL_MAT_ID(ffn_moe_down-56): 191 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 390 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 374 us MUL_MAT_ID(ffn_moe_down-58): 187 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 379 us MUL_MAT_ID(ffn_moe_down-59): 187 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 400 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 371 us MUL_MAT_ID(ffn_moe_down-61): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 364 us MUL_MAT_ID(ffn_moe_down-62): 176 us ggml_barrier(...): 322 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 387 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 375 us MUL_MAT_ID(ffn_moe_down-64): 332 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 539 us MUL_MAT_ID(ffn_moe_down-65): 262 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 357 us MUL_MAT_ID(ffn_moe_down-66): 182 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 386 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 413 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 362 us MUL_MAT_ID(ffn_moe_down-69): 207 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 401 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 373 us MUL_MAT_ID(ffn_moe_down-71): 368 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 397 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 366 us MUL_MAT_ID(ffn_moe_down-73): 178 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 369 us MUL_MAT_ID(ffn_moe_down-74): 192 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 384 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 360 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 371 us MUL_MAT_ID(ffn_moe_down-77): 209 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 394 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 395 us MUL_MAT_ID(ffn_moe_down-79): 187 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 370 us MUL_MAT_ID(ffn_moe_down-80): 214 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 381 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 378 us MUL_MAT_ID(ffn_moe_down-82): 205 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 393 us MUL_MAT_ID(ffn_moe_down-83): 186 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 386 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 366 us MUL_MAT_ID(ffn_moe_down-85): 428 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 398 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 362 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 352 us MUL_MAT_ID(ffn_moe_down-88): 201 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 390 us MUL_MAT_ID(ffn_moe_down-89): 203 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 382 us MUL_MAT_ID(ffn_moe_down-90): 229 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 391 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 41 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 410 us MUL_MAT_ID(ffn_moe_down-25): 199 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 375 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 500 us MUL_MAT_ID(ffn_moe_down-27): 200 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 353 us MUL_MAT_ID(ffn_moe_down-28): 179 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 398 us MUL_MAT_ID(ffn_moe_down-29): 220 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 465 us MUL_MAT_ID(ffn_moe_down-30): 187 us ggml_barrier(...): 188 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 375 us MUL_MAT_ID(ffn_moe_down-31): 207 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 360 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 356 us MUL_MAT_ID(ffn_moe_down-33): 208 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 394 us MUL_MAT_ID(ffn_moe_down-34): 205 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 371 us MUL_MAT_ID(ffn_moe_down-35): 189 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 373 us MUL_MAT_ID(ffn_moe_down-36): 211 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 380 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 376 us MUL_MAT_ID(ffn_moe_down-38): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 385 us MUL_MAT_ID(ffn_moe_down-39): 221 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 390 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 353 us MUL_MAT_ID(ffn_moe_down-41): 353 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 389 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 390 us MUL_MAT_ID(ffn_moe_down-43): 216 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 387 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 352 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 377 us MUL_MAT_ID(ffn_moe_down-46): 182 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 393 us MUL_MAT_ID(ffn_moe_down-47): 202 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 394 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 387 us MUL_MAT_ID(ffn_moe_down-49): 453 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 368 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 366 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 377 us MUL_MAT_ID(ffn_moe_down-52): 336 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 352 us MUL_MAT_ID(ffn_moe_down-53): 209 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 348 us MUL_MAT_ID(ffn_moe_down-54): 226 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 406 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 364 us MUL_MAT_ID(ffn_moe_down-56): 271 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 363 us MUL_MAT_ID(ffn_moe_down-57): 187 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 391 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 367 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 396 us MUL_MAT_ID(ffn_moe_down-60): 197 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 380 us MUL_MAT_ID(ffn_moe_down-61): 187 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 377 us MUL_MAT_ID(ffn_moe_down-62): 378 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 412 us MUL_MAT_ID(ffn_moe_down-63): 205 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 381 us MUL_MAT_ID(ffn_moe_down-64): 203 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 375 us MUL_MAT_ID(ffn_moe_down-65): 187 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 376 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 355 us MUL_MAT_ID(ffn_moe_down-67): 221 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 384 us MUL_MAT_ID(ffn_moe_down-68): 355 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 411 us MUL_MAT_ID(ffn_moe_down-69): 376 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 391 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 413 us MUL_MAT_ID(ffn_moe_down-71): 382 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 395 us MUL_MAT_ID(ffn_moe_down-72): 188 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 399 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 378 us MUL_MAT_ID(ffn_moe_down-74): 206 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 394 us MUL_MAT_ID(ffn_moe_down-75): 525 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 366 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 394 us MUL_MAT_ID(ffn_moe_down-77): 196 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 418 us MUL_MAT_ID(ffn_moe_down-78): 188 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 361 us MUL_MAT_ID(ffn_moe_down-79): 202 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 376 us MUL_MAT_ID(ffn_moe_down-80): 200 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 501 us MUL_MAT_ID(ffn_moe_down-81): 181 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 379 us MUL_MAT_ID(ffn_moe_down-82): 191 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 396 us MUL_MAT_ID(ffn_moe_down-83): 460 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 405 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 402 us MUL_MAT_ID(ffn_moe_down-85): 227 us ggml_barrier(...): 94 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 361 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 382 us MUL_MAT_ID(ffn_moe_down-87): 215 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 380 us MUL_MAT_ID(ffn_moe_down-88): 193 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 366 us MUL_MAT_ID(ffn_moe_down-89): 191 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 394 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 385 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 50 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 405 us MUL_MAT_ID(ffn_moe_down-25): 219 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 385 us MUL_MAT_ID(ffn_moe_down-26): 197 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 388 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 370 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 349 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 376 us MUL_MAT_ID(ffn_moe_down-29): 244 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 362 us MUL_MAT_ID(ffn_moe_down-30): 406 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 383 us MUL_MAT_ID(ffn_moe_down-31): 216 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 387 us MUL_MAT_ID(ffn_moe_down-32): 201 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 393 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 387 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 403 us MUL_MAT_ID(ffn_moe_down-35): 197 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 496 us MUL_MAT_ID(ffn_moe_down-36): 296 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 432 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 374 us MUL_MAT_ID(ffn_moe_down-38): 212 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 415 us MUL_MAT_ID(ffn_moe_down-39): 641 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 369 us MUL_MAT_ID(ffn_moe_down-40): 257 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 394 us MUL_MAT_ID(ffn_moe_down-41): 208 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 399 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 386 us MUL_MAT_ID(ffn_moe_down-43): 216 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 374 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 360 us MUL_MAT_ID(ffn_moe_down-45): 200 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 366 us MUL_MAT_ID(ffn_moe_down-46): 209 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 387 us MUL_MAT_ID(ffn_moe_down-47): 379 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 351 us MUL_MAT_ID(ffn_moe_down-48): 345 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 367 us MUL_MAT_ID(ffn_moe_down-49): 209 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 400 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 360 us MUL_MAT_ID(ffn_moe_down-51): 389 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 361 us MUL_MAT_ID(ffn_moe_down-52): 190 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 405 us MUL_MAT_ID(ffn_moe_down-53): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 356 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 413 us MUL_MAT_ID(ffn_moe_down-55): 194 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 375 us MUL_MAT_ID(ffn_moe_down-56): 530 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 360 us MUL_MAT_ID(ffn_moe_down-57): 213 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 363 us MUL_MAT_ID(ffn_moe_down-58): 204 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 366 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 406 us MUL_MAT_ID(ffn_moe_down-60): 222 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 397 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 382 us MUL_MAT_ID(ffn_moe_down-62): 537 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 407 us MUL_MAT_ID(ffn_moe_down-63): 215 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 384 us MUL_MAT_ID(ffn_moe_down-64): 180 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 388 us MUL_MAT_ID(ffn_moe_down-65): 207 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 407 us MUL_MAT_ID(ffn_moe_down-66): 297 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 352 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 361 us MUL_MAT_ID(ffn_moe_down-68): 192 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 385 us MUL_MAT_ID(ffn_moe_down-69): 241 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 376 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 345 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 356 us MUL_MAT_ID(ffn_moe_down-72): 218 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 371 us MUL_MAT_ID(ffn_moe_down-73): 193 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 385 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 376 us MUL_MAT_ID(ffn_moe_down-75): 228 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 413 us MUL_MAT_ID(ffn_moe_down-76): 257 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 362 us MUL_MAT_ID(ffn_moe_down-77): 200 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 406 us MUL_MAT_ID(ffn_moe_down-78): 206 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 451 us MUL_MAT_ID(ffn_moe_down-79): 164 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 389 us MUL_MAT_ID(ffn_moe_down-80): 192 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 394 us MUL_MAT_ID(ffn_moe_down-81): 184 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 390 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 369 us MUL_MAT_ID(ffn_moe_down-83): 284 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 363 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 405 us MUL_MAT_ID(ffn_moe_down-85): 698 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 379 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 406 us MUL_MAT_ID(ffn_moe_down-87): 211 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 382 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 399 us MUL_MAT_ID(ffn_moe_down-89): 211 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 392 us MUL_MAT_ID(ffn_moe_down-90): 666 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 393 us MUL_MAT_ID(ffn_moe_down-91): 248 us ggml_barrier(...): 36 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 352 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 383 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 363 us MUL_MAT_ID(ffn_moe_down-27): 209 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 355 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 389 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 422 us MUL_MAT_ID(ffn_moe_down-30): 180 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 379 us MUL_MAT_ID(ffn_moe_down-31): 227 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 372 us MUL_MAT_ID(ffn_moe_down-32): 186 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 399 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 379 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 405 us MUL_MAT_ID(ffn_moe_down-35): 326 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 357 us MUL_MAT_ID(ffn_moe_down-36): 222 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 376 us MUL_MAT_ID(ffn_moe_down-37): 320 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 221 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 396 us MUL_MAT_ID(ffn_moe_down-39): 682 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 443 us MUL_MAT_ID(ffn_moe_down-40): 224 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 375 us MUL_MAT_ID(ffn_moe_down-41): 411 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 363 us MUL_MAT_ID(ffn_moe_down-42): 259 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 377 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 354 us MUL_MAT_ID(ffn_moe_down-44): 364 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 357 us MUL_MAT_ID(ffn_moe_down-45): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 386 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 395 us MUL_MAT_ID(ffn_moe_down-47): 221 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 378 us MUL_MAT_ID(ffn_moe_down-48): 204 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 363 us MUL_MAT_ID(ffn_moe_down-49): 182 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 389 us MUL_MAT_ID(ffn_moe_down-50): 188 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 382 us MUL_MAT_ID(ffn_moe_down-51): 199 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 390 us MUL_MAT_ID(ffn_moe_down-52): 220 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 423 us MUL_MAT_ID(ffn_moe_down-53): 168 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 350 us MUL_MAT_ID(ffn_moe_down-54): 220 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 393 us MUL_MAT_ID(ffn_moe_down-55): 370 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 363 us MUL_MAT_ID(ffn_moe_down-56): 188 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 378 us MUL_MAT_ID(ffn_moe_down-57): 210 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 392 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 471 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 393 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 387 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 359 us MUL_MAT_ID(ffn_moe_down-62): 376 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 377 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 392 us MUL_MAT_ID(ffn_moe_down-64): 189 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 340 us MUL_MAT_ID(ffn_moe_down-65): 321 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 347 us MUL_MAT_ID(ffn_moe_down-66): 199 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 399 us MUL_MAT_ID(ffn_moe_down-67): 209 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 395 us MUL_MAT_ID(ffn_moe_down-68): 374 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 386 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 376 us MUL_MAT_ID(ffn_moe_down-70): 186 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 404 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 395 us MUL_MAT_ID(ffn_moe_down-72): 561 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 373 us MUL_MAT_ID(ffn_moe_down-73): 202 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 384 us MUL_MAT_ID(ffn_moe_down-74): 228 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 373 us MUL_MAT_ID(ffn_moe_down-75): 230 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 368 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 408 us MUL_MAT_ID(ffn_moe_down-77): 197 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 453 us MUL_MAT_ID(ffn_moe_down-78): 309 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 382 us MUL_MAT_ID(ffn_moe_down-79): 318 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 373 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 379 us MUL_MAT_ID(ffn_moe_down-81): 354 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 385 us MUL_MAT_ID(ffn_moe_down-82): 524 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 383 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 393 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 376 us MUL_MAT_ID(ffn_moe_down-85): 239 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 370 us MUL_MAT_ID(ffn_moe_down-86): 170 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 414 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 425 us MUL_MAT_ID(ffn_moe_down-88): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 390 us MUL_MAT_ID(ffn_moe_down-89): 191 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 372 us MUL_MAT_ID(ffn_moe_down-90): 243 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 357 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 61 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 413 us MUL_MAT_ID(ffn_moe_down-25): 178 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 391 us MUL_MAT_ID(ffn_moe_down-26): 201 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 393 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 401 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 384 us MUL_MAT_ID(ffn_moe_down-29): 184 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 372 us MUL_MAT_ID(ffn_moe_down-30): 190 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 366 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 403 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 413 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 412 us MUL_MAT_ID(ffn_moe_down-34): 283 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 344 us MUL_MAT_ID(ffn_moe_down-35): 481 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 367 us MUL_MAT_ID(ffn_moe_down-36): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 387 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 373 us MUL_MAT_ID(ffn_moe_down-38): 208 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 365 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 354 us MUL_MAT_ID(ffn_moe_down-40): 271 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 353 us MUL_MAT_ID(ffn_moe_down-41): 204 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 361 us MUL_MAT_ID(ffn_moe_down-42): 202 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 386 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 364 us MUL_MAT_ID(ffn_moe_down-44): 270 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 375 us MUL_MAT_ID(ffn_moe_down-45): 221 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 405 us MUL_MAT_ID(ffn_moe_down-46): 219 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 400 us MUL_MAT_ID(ffn_moe_down-47): 198 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 379 us MUL_MAT_ID(ffn_moe_down-48): 351 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 385 us MUL_MAT_ID(ffn_moe_down-49): 187 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 350 us MUL_MAT_ID(ffn_moe_down-50): 210 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 424 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 374 us MUL_MAT_ID(ffn_moe_down-52): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 400 us MUL_MAT_ID(ffn_moe_down-53): 185 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 391 us MUL_MAT_ID(ffn_moe_down-54): 209 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 394 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 374 us MUL_MAT_ID(ffn_moe_down-56): 211 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 377 us MUL_MAT_ID(ffn_moe_down-57): 185 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 389 us MUL_MAT_ID(ffn_moe_down-58): 322 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 383 us MUL_MAT_ID(ffn_moe_down-59): 220 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 395 us MUL_MAT_ID(ffn_moe_down-60): 248 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 404 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 389 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 353 us MUL_MAT_ID(ffn_moe_down-63): 192 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 383 us MUL_MAT_ID(ffn_moe_down-64): 257 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 362 us MUL_MAT_ID(ffn_moe_down-65): 186 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 384 us MUL_MAT_ID(ffn_moe_down-66): 212 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 395 us MUL_MAT_ID(ffn_moe_down-67): 268 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 402 us MUL_MAT_ID(ffn_moe_down-68): 200 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 354 us MUL_MAT_ID(ffn_moe_down-69): 203 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 420 us MUL_MAT_ID(ffn_moe_down-70): 196 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 340 us MUL_MAT_ID(ffn_moe_down-71): 201 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 386 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 362 us MUL_MAT_ID(ffn_moe_down-73): 527 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 375 us MUL_MAT_ID(ffn_moe_down-74): 354 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 403 us MUL_MAT_ID(ffn_moe_down-75): 429 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 399 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 366 us MUL_MAT_ID(ffn_moe_down-77): 217 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 392 us MUL_MAT_ID(ffn_moe_down-78): 216 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 354 us MUL_MAT_ID(ffn_moe_down-79): 461 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 409 us MUL_MAT_ID(ffn_moe_down-80): 220 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 359 us MUL_MAT_ID(ffn_moe_down-81): 383 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 178 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 356 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 386 us MUL_MAT_ID(ffn_moe_down-84): 226 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 359 us MUL_MAT_ID(ffn_moe_down-85): 241 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 354 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 374 us MUL_MAT_ID(ffn_moe_down-87): 209 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 395 us MUL_MAT_ID(ffn_moe_down-88): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 365 us MUL_MAT_ID(ffn_moe_down-89): 187 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 369 us MUL_MAT_ID(ffn_moe_down-90): 561 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 356 us MUL_MAT_ID(ffn_moe_down-91): 247 us ggml_barrier(...): 67 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 383 us MUL_MAT_ID(ffn_moe_down-25): 182 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 489 us MUL_MAT_ID(ffn_moe_down-26): 177 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 381 us MUL_MAT_ID(ffn_moe_down-27): 366 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 385 us MUL_MAT_ID(ffn_moe_down-28): 315 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 385 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 460 us MUL_MAT_ID(ffn_moe_down-30): 165 us ggml_barrier(...): 205 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 370 us MUL_MAT_ID(ffn_moe_down-31): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 471 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 396 us MUL_MAT_ID(ffn_moe_down-33): 540 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 404 us MUL_MAT_ID(ffn_moe_down-34): 244 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 377 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 370 us MUL_MAT_ID(ffn_moe_down-36): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 390 us MUL_MAT_ID(ffn_moe_down-37): 221 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 385 us MUL_MAT_ID(ffn_moe_down-38): 208 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 389 us MUL_MAT_ID(ffn_moe_down-39): 196 us ggml_barrier(...): 295 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 387 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 477 us MUL_MAT_ID(ffn_moe_down-41): 166 us ggml_barrier(...): 192 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 431 us MUL_MAT_ID(ffn_moe_down-42): 503 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 401 us MUL_MAT_ID(ffn_moe_down-43): 208 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 411 us MUL_MAT_ID(ffn_moe_down-44): 248 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 352 us MUL_MAT_ID(ffn_moe_down-45): 218 us ggml_barrier(...): 260 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 412 us MUL_MAT_ID(ffn_moe_down-46): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 398 us MUL_MAT_ID(ffn_moe_down-47): 203 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 367 us MUL_MAT_ID(ffn_moe_down-48): 197 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 382 us MUL_MAT_ID(ffn_moe_down-49): 274 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 380 us MUL_MAT_ID(ffn_moe_down-50): 206 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 368 us MUL_MAT_ID(ffn_moe_down-51): 220 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 403 us MUL_MAT_ID(ffn_moe_down-52): 307 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 387 us MUL_MAT_ID(ffn_moe_down-53): 178 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 383 us MUL_MAT_ID(ffn_moe_down-54): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 374 us MUL_MAT_ID(ffn_moe_down-55): 186 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 388 us MUL_MAT_ID(ffn_moe_down-56): 263 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 394 us MUL_MAT_ID(ffn_moe_down-57): 188 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 408 us MUL_MAT_ID(ffn_moe_down-58): 215 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 402 us MUL_MAT_ID(ffn_moe_down-59): 414 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 352 us MUL_MAT_ID(ffn_moe_down-60): 252 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 375 us MUL_MAT_ID(ffn_moe_down-61): 188 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 355 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 436 us MUL_MAT_ID(ffn_moe_down-63): 254 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 390 us MUL_MAT_ID(ffn_moe_down-64): 186 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 437 us MUL_MAT_ID(ffn_moe_down-65): 238 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 421 us MUL_MAT_ID(ffn_moe_down-66): 289 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 370 us MUL_MAT_ID(ffn_moe_down-67): 210 us ggml_barrier(...): 264 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 348 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 379 us MUL_MAT_ID(ffn_moe_down-69): 192 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 400 us MUL_MAT_ID(ffn_moe_down-70): 393 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 479 us MUL_MAT_ID(ffn_moe_down-71): 187 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 373 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 384 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 360 us MUL_MAT_ID(ffn_moe_down-74): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 437 us MUL_MAT_ID(ffn_moe_down-75): 304 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 388 us MUL_MAT_ID(ffn_moe_down-76): 243 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 372 us MUL_MAT_ID(ffn_moe_down-77): 218 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 368 us MUL_MAT_ID(ffn_moe_down-78): 189 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 506 us MUL_MAT_ID(ffn_moe_down-79): 212 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 347 us MUL_MAT_ID(ffn_moe_down-80): 368 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 394 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 478 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 395 us MUL_MAT_ID(ffn_moe_down-83): 185 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 382 us MUL_MAT_ID(ffn_moe_down-84): 217 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 361 us MUL_MAT_ID(ffn_moe_down-85): 295 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 366 us MUL_MAT_ID(ffn_moe_down-86): 170 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 365 us MUL_MAT_ID(ffn_moe_down-87): 177 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 357 us MUL_MAT_ID(ffn_moe_down-88): 429 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 365 us MUL_MAT_ID(ffn_moe_down-89): 199 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 347 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 400 us MUL_MAT_ID(ffn_moe_down-91): 284 us ggml_barrier(...): 15 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 400 us MUL_MAT_ID(ffn_moe_down-25): 199 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 373 us MUL_MAT_ID(ffn_moe_down-26): 186 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 391 us MUL_MAT_ID(ffn_moe_down-27): 191 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 386 us MUL_MAT_ID(ffn_moe_down-28): 188 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 388 us MUL_MAT_ID(ffn_moe_down-29): 198 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 377 us MUL_MAT_ID(ffn_moe_down-30): 214 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 388 us MUL_MAT_ID(ffn_moe_down-31): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 392 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 401 us MUL_MAT_ID(ffn_moe_down-33): 225 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 407 us MUL_MAT_ID(ffn_moe_down-34): 206 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 349 us MUL_MAT_ID(ffn_moe_down-35): 550 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 402 us MUL_MAT_ID(ffn_moe_down-36): 395 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 396 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 418 us MUL_MAT_ID(ffn_moe_down-38): 186 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 347 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 394 us MUL_MAT_ID(ffn_moe_down-40): 228 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 369 us MUL_MAT_ID(ffn_moe_down-41): 360 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 516 us MUL_MAT_ID(ffn_moe_down-42): 220 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 365 us MUL_MAT_ID(ffn_moe_down-43): 213 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 370 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 402 us MUL_MAT_ID(ffn_moe_down-45): 221 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 346 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 388 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 383 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 369 us MUL_MAT_ID(ffn_moe_down-49): 486 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 354 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 408 us MUL_MAT_ID(ffn_moe_down-51): 204 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 394 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 376 us MUL_MAT_ID(ffn_moe_down-53): 325 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 365 us MUL_MAT_ID(ffn_moe_down-54): 218 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 357 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 381 us MUL_MAT_ID(ffn_moe_down-56): 187 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 352 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 369 us MUL_MAT_ID(ffn_moe_down-58): 565 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 365 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 348 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 385 us MUL_MAT_ID(ffn_moe_down-61): 186 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 399 us MUL_MAT_ID(ffn_moe_down-62): 204 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 367 us MUL_MAT_ID(ffn_moe_down-63): 200 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 356 us MUL_MAT_ID(ffn_moe_down-64): 201 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 407 us MUL_MAT_ID(ffn_moe_down-65): 392 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 381 us MUL_MAT_ID(ffn_moe_down-66): 201 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 379 us MUL_MAT_ID(ffn_moe_down-67): 210 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 415 us MUL_MAT_ID(ffn_moe_down-68): 196 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 357 us MUL_MAT_ID(ffn_moe_down-69): 209 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 374 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 387 us MUL_MAT_ID(ffn_moe_down-71): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 348 us MUL_MAT_ID(ffn_moe_down-72): 192 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 420 us MUL_MAT_ID(ffn_moe_down-73): 260 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 384 us MUL_MAT_ID(ffn_moe_down-74): 210 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 378 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 400 us MUL_MAT_ID(ffn_moe_down-76): 229 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 358 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 362 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 398 us MUL_MAT_ID(ffn_moe_down-79): 197 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 380 us MUL_MAT_ID(ffn_moe_down-80): 202 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 394 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 284 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 383 us MUL_MAT_ID(ffn_moe_down-83): 206 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 386 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 391 us MUL_MAT_ID(ffn_moe_down-85): 226 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 397 us MUL_MAT_ID(ffn_moe_down-86): 213 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 371 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 373 us MUL_MAT_ID(ffn_moe_down-88): 230 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 393 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 383 us MUL_MAT_ID(ffn_moe_down-90): 237 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 362 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 51 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 382 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 403 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 355 us MUL_MAT_ID(ffn_moe_down-27): 199 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 379 us MUL_MAT_ID(ffn_moe_down-28): 195 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 368 us MUL_MAT_ID(ffn_moe_down-29): 187 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 361 us MUL_MAT_ID(ffn_moe_down-30): 381 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 388 us MUL_MAT_ID(ffn_moe_down-31): 201 us ggml_barrier(...): 411 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 356 us MUL_MAT_ID(ffn_moe_down-32): 387 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 390 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 357 us MUL_MAT_ID(ffn_moe_down-34): 491 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 364 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 417 us MUL_MAT_ID(ffn_moe_down-36): 323 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 376 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 398 us MUL_MAT_ID(ffn_moe_down-38): 361 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 384 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 342 us MUL_MAT_ID(ffn_moe_down-40): 218 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 375 us MUL_MAT_ID(ffn_moe_down-41): 215 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 380 us MUL_MAT_ID(ffn_moe_down-42): 200 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 337 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 396 us MUL_MAT_ID(ffn_moe_down-44): 193 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 397 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 417 us MUL_MAT_ID(ffn_moe_down-46): 217 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 374 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 408 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 419 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 392 us MUL_MAT_ID(ffn_moe_down-50): 493 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 515 us MUL_MAT_ID(ffn_moe_down-51): 181 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 386 us MUL_MAT_ID(ffn_moe_down-52): 271 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 402 us MUL_MAT_ID(ffn_moe_down-53): 247 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 400 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 359 us MUL_MAT_ID(ffn_moe_down-55): 226 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 408 us MUL_MAT_ID(ffn_moe_down-56): 204 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 425 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 387 us MUL_MAT_ID(ffn_moe_down-58): 266 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 411 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 406 us MUL_MAT_ID(ffn_moe_down-60): 197 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 390 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 371 us MUL_MAT_ID(ffn_moe_down-62): 210 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 369 us MUL_MAT_ID(ffn_moe_down-63): 182 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 352 us MUL_MAT_ID(ffn_moe_down-64): 192 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 387 us MUL_MAT_ID(ffn_moe_down-65): 204 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 396 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 422 us MUL_MAT_ID(ffn_moe_down-67): 391 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 401 us MUL_MAT_ID(ffn_moe_down-68): 185 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 394 us MUL_MAT_ID(ffn_moe_down-69): 169 us ggml_barrier(...): 368 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 372 us MUL_MAT_ID(ffn_moe_down-70): 271 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 395 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 400 us MUL_MAT_ID(ffn_moe_down-72): 521 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 360 us MUL_MAT_ID(ffn_moe_down-73): 253 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 376 us MUL_MAT_ID(ffn_moe_down-74): 368 us ggml_barrier(...): 313 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 344 us MUL_MAT_ID(ffn_moe_down-75): 240 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 385 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 380 us MUL_MAT_ID(ffn_moe_down-77): 188 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 378 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 364 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 378 us MUL_MAT_ID(ffn_moe_down-80): 287 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 401 us MUL_MAT_ID(ffn_moe_down-81): 203 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 341 us MUL_MAT_ID(ffn_moe_down-82): 180 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 387 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 407 us MUL_MAT_ID(ffn_moe_down-84): 224 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 360 us MUL_MAT_ID(ffn_moe_down-85): 227 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 372 us MUL_MAT_ID(ffn_moe_down-86): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 380 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 400 us MUL_MAT_ID(ffn_moe_down-88): 182 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 397 us MUL_MAT_ID(ffn_moe_down-89): 388 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 397 us MUL_MAT_ID(ffn_moe_down-90): 240 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 466 us MUL_MAT_ID(ffn_moe_down-91): 258 us ggml_barrier(...): 192 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 395 us MUL_MAT_ID(ffn_moe_down-25): 184 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 411 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 402 us MUL_MAT_ID(ffn_moe_down-27): 212 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 366 us MUL_MAT_ID(ffn_moe_down-28): 187 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 398 us MUL_MAT_ID(ffn_moe_down-29): 213 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 385 us MUL_MAT_ID(ffn_moe_down-30): 203 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 369 us MUL_MAT_ID(ffn_moe_down-31): 183 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 392 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 213 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 393 us MUL_MAT_ID(ffn_moe_down-34): 209 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 413 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 350 us MUL_MAT_ID(ffn_moe_down-36): 187 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 376 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 364 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 380 us MUL_MAT_ID(ffn_moe_down-39): 226 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 350 us MUL_MAT_ID(ffn_moe_down-40): 229 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 373 us MUL_MAT_ID(ffn_moe_down-41): 204 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 407 us MUL_MAT_ID(ffn_moe_down-42): 388 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 374 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 361 us MUL_MAT_ID(ffn_moe_down-44): 277 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 376 us MUL_MAT_ID(ffn_moe_down-45): 389 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 353 us MUL_MAT_ID(ffn_moe_down-46): 182 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 390 us MUL_MAT_ID(ffn_moe_down-47): 217 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 390 us MUL_MAT_ID(ffn_moe_down-48): 205 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 476 us MUL_MAT_ID(ffn_moe_down-49): 234 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 378 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 375 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 479 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 370 us MUL_MAT_ID(ffn_moe_down-52): 413 us ggml_barrier(...): 269 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 365 us MUL_MAT_ID(ffn_moe_down-53): 196 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 386 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 414 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 374 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 358 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 401 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 401 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 375 us MUL_MAT_ID(ffn_moe_down-60): 202 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 392 us MUL_MAT_ID(ffn_moe_down-61): 187 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 357 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 378 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 368 us MUL_MAT_ID(ffn_moe_down-64): 179 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 390 us MUL_MAT_ID(ffn_moe_down-65): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 388 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 354 us MUL_MAT_ID(ffn_moe_down-67): 227 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 372 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 372 us MUL_MAT_ID(ffn_moe_down-69): 203 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 380 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 393 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 357 us MUL_MAT_ID(ffn_moe_down-72): 325 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 361 us MUL_MAT_ID(ffn_moe_down-73): 187 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 371 us MUL_MAT_ID(ffn_moe_down-74): 369 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 356 us MUL_MAT_ID(ffn_moe_down-75): 516 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 416 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 371 us MUL_MAT_ID(ffn_moe_down-77): 204 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 363 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 393 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 370 us MUL_MAT_ID(ffn_moe_down-80): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 397 us MUL_MAT_ID(ffn_moe_down-81): 217 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 354 us MUL_MAT_ID(ffn_moe_down-82): 214 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 361 us MUL_MAT_ID(ffn_moe_down-83): 312 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 395 us MUL_MAT_ID(ffn_moe_down-84): 215 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 377 us MUL_MAT_ID(ffn_moe_down-85): 561 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 416 us MUL_MAT_ID(ffn_moe_down-86): 184 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 386 us MUL_MAT_ID(ffn_moe_down-87): 514 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 402 us MUL_MAT_ID(ffn_moe_down-88): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 350 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 366 us MUL_MAT_ID(ffn_moe_down-90): 249 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 371 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 35 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 378 us MUL_MAT_ID(ffn_moe_down-25): 184 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 379 us MUL_MAT_ID(ffn_moe_down-26): 186 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 393 us MUL_MAT_ID(ffn_moe_down-27): 328 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 394 us MUL_MAT_ID(ffn_moe_down-28): 196 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 400 us MUL_MAT_ID(ffn_moe_down-29): 309 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 410 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 363 us MUL_MAT_ID(ffn_moe_down-31): 364 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 359 us MUL_MAT_ID(ffn_moe_down-32): 209 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 377 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 399 us MUL_MAT_ID(ffn_moe_down-34): 190 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 383 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 382 us MUL_MAT_ID(ffn_moe_down-36): 336 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 388 us MUL_MAT_ID(ffn_moe_down-37): 238 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 410 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 401 us MUL_MAT_ID(ffn_moe_down-39): 627 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 385 us MUL_MAT_ID(ffn_moe_down-40): 702 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 369 us MUL_MAT_ID(ffn_moe_down-41): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 359 us MUL_MAT_ID(ffn_moe_down-42): 206 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 374 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 382 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 365 us MUL_MAT_ID(ffn_moe_down-45): 189 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 365 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 356 us MUL_MAT_ID(ffn_moe_down-47): 398 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 421 us MUL_MAT_ID(ffn_moe_down-48): 378 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 379 us MUL_MAT_ID(ffn_moe_down-49): 305 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 383 us MUL_MAT_ID(ffn_moe_down-50): 201 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 398 us MUL_MAT_ID(ffn_moe_down-51): 192 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 389 us MUL_MAT_ID(ffn_moe_down-52): 465 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 369 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 362 us MUL_MAT_ID(ffn_moe_down-54): 205 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 398 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 394 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 388 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 400 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 410 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 368 us MUL_MAT_ID(ffn_moe_down-60): 207 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 400 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 352 us MUL_MAT_ID(ffn_moe_down-62): 312 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 398 us MUL_MAT_ID(ffn_moe_down-63): 352 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 414 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 423 us MUL_MAT_ID(ffn_moe_down-65): 259 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 363 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 466 us MUL_MAT_ID(ffn_moe_down-67): 185 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 362 us MUL_MAT_ID(ffn_moe_down-68): 519 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 354 us MUL_MAT_ID(ffn_moe_down-69): 184 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 351 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 355 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 367 us MUL_MAT_ID(ffn_moe_down-72): 204 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 389 us MUL_MAT_ID(ffn_moe_down-73): 200 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 359 us MUL_MAT_ID(ffn_moe_down-74): 223 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 348 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 381 us MUL_MAT_ID(ffn_moe_down-76): 204 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 368 us MUL_MAT_ID(ffn_moe_down-77): 204 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 391 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 440 us MUL_MAT_ID(ffn_moe_down-79): 203 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 380 us MUL_MAT_ID(ffn_moe_down-80): 194 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 363 us MUL_MAT_ID(ffn_moe_down-81): 188 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 376 us MUL_MAT_ID(ffn_moe_down-82): 204 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 380 us MUL_MAT_ID(ffn_moe_down-83): 190 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 361 us MUL_MAT_ID(ffn_moe_down-84): 225 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 381 us MUL_MAT_ID(ffn_moe_down-85): 695 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 352 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 365 us MUL_MAT_ID(ffn_moe_down-87): 320 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 390 us MUL_MAT_ID(ffn_moe_down-88): 187 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 349 us MUL_MAT_ID(ffn_moe_down-89): 445 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 362 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 378 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 294 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 363 us MUL_MAT_ID(ffn_moe_down-26): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 396 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 405 us MUL_MAT_ID(ffn_moe_down-28): 359 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 398 us MUL_MAT_ID(ffn_moe_down-29): 394 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 382 us MUL_MAT_ID(ffn_moe_down-30): 188 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 385 us MUL_MAT_ID(ffn_moe_down-31): 183 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 409 us MUL_MAT_ID(ffn_moe_down-32): 227 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 413 us MUL_MAT_ID(ffn_moe_down-33): 210 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 406 us MUL_MAT_ID(ffn_moe_down-34): 210 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 378 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 402 us MUL_MAT_ID(ffn_moe_down-36): 362 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 361 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 390 us MUL_MAT_ID(ffn_moe_down-38): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 409 us MUL_MAT_ID(ffn_moe_down-39): 248 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 403 us MUL_MAT_ID(ffn_moe_down-40): 237 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 348 us MUL_MAT_ID(ffn_moe_down-41): 200 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 408 us MUL_MAT_ID(ffn_moe_down-42): 528 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 350 us MUL_MAT_ID(ffn_moe_down-43): 192 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 378 us MUL_MAT_ID(ffn_moe_down-44): 203 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 394 us MUL_MAT_ID(ffn_moe_down-45): 240 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 378 us MUL_MAT_ID(ffn_moe_down-46): 209 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 390 us MUL_MAT_ID(ffn_moe_down-47): 216 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 393 us MUL_MAT_ID(ffn_moe_down-48): 203 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 472 us MUL_MAT_ID(ffn_moe_down-49): 233 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 381 us MUL_MAT_ID(ffn_moe_down-50): 367 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 375 us MUL_MAT_ID(ffn_moe_down-51): 224 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 392 us MUL_MAT_ID(ffn_moe_down-52): 373 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 379 us MUL_MAT_ID(ffn_moe_down-53): 429 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 378 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 385 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 368 us MUL_MAT_ID(ffn_moe_down-56): 206 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 355 us MUL_MAT_ID(ffn_moe_down-57): 198 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 385 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 483 us MUL_MAT_ID(ffn_moe_down-59): 173 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 385 us MUL_MAT_ID(ffn_moe_down-60): 205 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 365 us MUL_MAT_ID(ffn_moe_down-61): 171 us ggml_barrier(...): 386 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 365 us MUL_MAT_ID(ffn_moe_down-62): 212 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 352 us MUL_MAT_ID(ffn_moe_down-63): 363 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 378 us MUL_MAT_ID(ffn_moe_down-64): 331 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 355 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 397 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 388 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 378 us MUL_MAT_ID(ffn_moe_down-68): 201 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 377 us MUL_MAT_ID(ffn_moe_down-69): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 356 us MUL_MAT_ID(ffn_moe_down-70): 213 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 393 us MUL_MAT_ID(ffn_moe_down-71): 219 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 382 us MUL_MAT_ID(ffn_moe_down-72): 204 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 363 us MUL_MAT_ID(ffn_moe_down-73): 202 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 395 us MUL_MAT_ID(ffn_moe_down-74): 210 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 536 us MUL_MAT_ID(ffn_moe_down-75): 177 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 431 us MUL_MAT_ID(ffn_moe_down-76): 250 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 391 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 424 us MUL_MAT_ID(ffn_moe_down-78): 227 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 372 us MUL_MAT_ID(ffn_moe_down-79): 385 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 365 us MUL_MAT_ID(ffn_moe_down-80): 222 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 374 us MUL_MAT_ID(ffn_moe_down-81): 216 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 378 us MUL_MAT_ID(ffn_moe_down-82): 559 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 387 us MUL_MAT_ID(ffn_moe_down-83): 201 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 372 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 369 us MUL_MAT_ID(ffn_moe_down-85): 222 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 367 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 345 us MUL_MAT_ID(ffn_moe_down-87): 189 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 399 us MUL_MAT_ID(ffn_moe_down-88): 188 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 340 us MUL_MAT_ID(ffn_moe_down-89): 167 us ggml_barrier(...): 390 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 385 us MUL_MAT_ID(ffn_moe_down-90): 384 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 400 us MUL_MAT_ID(ffn_moe_down-91): 548 us ggml_barrier(...): 84 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 389 us MUL_MAT_ID(ffn_moe_down-25): 184 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 346 us MUL_MAT_ID(ffn_moe_down-26): 198 us ggml_barrier(...): 361 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 387 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 386 us MUL_MAT_ID(ffn_moe_down-28): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 368 us MUL_MAT_ID(ffn_moe_down-29): 205 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 349 us MUL_MAT_ID(ffn_moe_down-30): 192 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 377 us MUL_MAT_ID(ffn_moe_down-31): 188 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 372 us MUL_MAT_ID(ffn_moe_down-32): 178 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 395 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 391 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 406 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 368 us MUL_MAT_ID(ffn_moe_down-36): 184 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 388 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 358 us MUL_MAT_ID(ffn_moe_down-38): 212 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 365 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 389 us MUL_MAT_ID(ffn_moe_down-40): 246 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 373 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 386 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 374 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 391 us MUL_MAT_ID(ffn_moe_down-44): 214 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 348 us MUL_MAT_ID(ffn_moe_down-45): 197 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 347 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 433 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 377 us MUL_MAT_ID(ffn_moe_down-48): 201 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 392 us MUL_MAT_ID(ffn_moe_down-49): 362 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 364 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 389 us MUL_MAT_ID(ffn_moe_down-51): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 376 us MUL_MAT_ID(ffn_moe_down-52): 361 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 356 us MUL_MAT_ID(ffn_moe_down-53): 219 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 397 us MUL_MAT_ID(ffn_moe_down-54): 372 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 395 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 393 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 392 us MUL_MAT_ID(ffn_moe_down-57): 209 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 374 us MUL_MAT_ID(ffn_moe_down-58): 173 us ggml_barrier(...): 481 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 369 us MUL_MAT_ID(ffn_moe_down-59): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 431 us MUL_MAT_ID(ffn_moe_down-60): 207 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 401 us MUL_MAT_ID(ffn_moe_down-61): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 353 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 353 us MUL_MAT_ID(ffn_moe_down-63): 196 us ggml_barrier(...): 358 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 402 us MUL_MAT_ID(ffn_moe_down-64): 307 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 397 us MUL_MAT_ID(ffn_moe_down-65): 268 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 419 us MUL_MAT_ID(ffn_moe_down-66): 213 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 367 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 375 us MUL_MAT_ID(ffn_moe_down-68): 213 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 375 us MUL_MAT_ID(ffn_moe_down-69): 371 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 361 us MUL_MAT_ID(ffn_moe_down-70): 395 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 372 us MUL_MAT_ID(ffn_moe_down-71): 203 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 372 us MUL_MAT_ID(ffn_moe_down-72): 420 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 359 us MUL_MAT_ID(ffn_moe_down-73): 220 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 386 us MUL_MAT_ID(ffn_moe_down-74): 516 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 382 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 501 us MUL_MAT_ID(ffn_moe_down-76): 182 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 414 us MUL_MAT_ID(ffn_moe_down-77): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 412 us MUL_MAT_ID(ffn_moe_down-78): 243 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 377 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 393 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 396 us MUL_MAT_ID(ffn_moe_down-81): 203 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 466 us MUL_MAT_ID(ffn_moe_down-82): 274 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 398 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 348 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 381 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 351 us MUL_MAT_ID(ffn_moe_down-86): 207 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 375 us MUL_MAT_ID(ffn_moe_down-87): 203 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 378 us MUL_MAT_ID(ffn_moe_down-88): 350 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 379 us MUL_MAT_ID(ffn_moe_down-89): 245 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 388 us MUL_MAT_ID(ffn_moe_down-90): 227 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 382 us MUL_MAT_ID(ffn_moe_down-91): 264 us ggml_barrier(...): 27 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 416 us MUL_MAT_ID(ffn_moe_down-25): 183 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 370 us MUL_MAT_ID(ffn_moe_down-26): 216 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 376 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 386 us MUL_MAT_ID(ffn_moe_down-28): 389 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 216 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 379 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 384 us MUL_MAT_ID(ffn_moe_down-31): 477 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 392 us MUL_MAT_ID(ffn_moe_down-32): 211 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 402 us MUL_MAT_ID(ffn_moe_down-33): 218 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 380 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 394 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 367 us MUL_MAT_ID(ffn_moe_down-36): 191 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 375 us MUL_MAT_ID(ffn_moe_down-37): 219 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 424 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 361 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 349 us MUL_MAT_ID(ffn_moe_down-40): 228 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 400 us MUL_MAT_ID(ffn_moe_down-41): 234 us ggml_barrier(...): 245 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 380 us MUL_MAT_ID(ffn_moe_down-42): 189 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 379 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 365 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 363 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 355 us MUL_MAT_ID(ffn_moe_down-45): 224 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 422 us MUL_MAT_ID(ffn_moe_down-46): 253 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 386 us MUL_MAT_ID(ffn_moe_down-47): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 363 us MUL_MAT_ID(ffn_moe_down-48): 190 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 402 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 364 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 380 us MUL_MAT_ID(ffn_moe_down-51): 350 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 355 us MUL_MAT_ID(ffn_moe_down-52): 201 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 351 us MUL_MAT_ID(ffn_moe_down-53): 348 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 403 us MUL_MAT_ID(ffn_moe_down-54): 227 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 409 us MUL_MAT_ID(ffn_moe_down-55): 187 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 371 us MUL_MAT_ID(ffn_moe_down-56): 276 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 396 us MUL_MAT_ID(ffn_moe_down-57): 188 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 413 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 357 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 394 us MUL_MAT_ID(ffn_moe_down-60): 206 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 401 us MUL_MAT_ID(ffn_moe_down-61): 361 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 361 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 369 us MUL_MAT_ID(ffn_moe_down-63): 201 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 435 us MUL_MAT_ID(ffn_moe_down-64): 357 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 363 us MUL_MAT_ID(ffn_moe_down-65): 189 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 373 us MUL_MAT_ID(ffn_moe_down-66): 226 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 402 us MUL_MAT_ID(ffn_moe_down-67): 233 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 369 us MUL_MAT_ID(ffn_moe_down-68): 188 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 348 us MUL_MAT_ID(ffn_moe_down-69): 307 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 357 us MUL_MAT_ID(ffn_moe_down-70): 187 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 396 us MUL_MAT_ID(ffn_moe_down-71): 207 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 393 us MUL_MAT_ID(ffn_moe_down-72): 338 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 367 us MUL_MAT_ID(ffn_moe_down-74): 243 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 379 us MUL_MAT_ID(ffn_moe_down-75): 199 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 366 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 374 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 421 us MUL_MAT_ID(ffn_moe_down-78): 206 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 374 us MUL_MAT_ID(ffn_moe_down-79): 189 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 437 us MUL_MAT_ID(ffn_moe_down-80): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 354 us MUL_MAT_ID(ffn_moe_down-81): 210 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 365 us MUL_MAT_ID(ffn_moe_down-82): 212 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 379 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 353 us MUL_MAT_ID(ffn_moe_down-84): 399 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 374 us MUL_MAT_ID(ffn_moe_down-85): 223 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 372 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 373 us MUL_MAT_ID(ffn_moe_down-87): 220 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 426 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 401 us MUL_MAT_ID(ffn_moe_down-89): 206 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 387 us MUL_MAT_ID(ffn_moe_down-90): 433 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 383 us MUL_MAT_ID(ffn_moe_down-91): 340 us ggml_barrier(...): 46 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 183 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 455 us MUL_MAT_ID(ffn_moe_down-26): 296 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 392 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 362 us MUL_MAT_ID(ffn_moe_down-28): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 361 us MUL_MAT_ID(ffn_moe_down-29): 182 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 393 us MUL_MAT_ID(ffn_moe_down-30): 209 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 384 us MUL_MAT_ID(ffn_moe_down-31): 188 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 380 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 411 us MUL_MAT_ID(ffn_moe_down-33): 185 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 409 us MUL_MAT_ID(ffn_moe_down-34): 215 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 357 us MUL_MAT_ID(ffn_moe_down-35): 177 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 368 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 381 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 391 us MUL_MAT_ID(ffn_moe_down-38): 215 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 377 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 372 us MUL_MAT_ID(ffn_moe_down-40): 235 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 366 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 395 us MUL_MAT_ID(ffn_moe_down-42): 220 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 409 us MUL_MAT_ID(ffn_moe_down-43): 199 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 370 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 384 us MUL_MAT_ID(ffn_moe_down-45): 227 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 438 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 233 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 383 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 385 us MUL_MAT_ID(ffn_moe_down-48): 411 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 398 us MUL_MAT_ID(ffn_moe_down-49): 276 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 385 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 415 us MUL_MAT_ID(ffn_moe_down-51): 442 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 394 us MUL_MAT_ID(ffn_moe_down-52): 212 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 368 us MUL_MAT_ID(ffn_moe_down-53): 196 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 355 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 430 us MUL_MAT_ID(ffn_moe_down-55): 290 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 375 us MUL_MAT_ID(ffn_moe_down-56): 186 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 400 us MUL_MAT_ID(ffn_moe_down-57): 204 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 348 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 357 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 398 us MUL_MAT_ID(ffn_moe_down-60): 321 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 414 us MUL_MAT_ID(ffn_moe_down-61): 182 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 402 us MUL_MAT_ID(ffn_moe_down-62): 203 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 392 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 427 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 394 us MUL_MAT_ID(ffn_moe_down-65): 221 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 382 us MUL_MAT_ID(ffn_moe_down-66): 272 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 368 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 388 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 393 us MUL_MAT_ID(ffn_moe_down-69): 211 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 395 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 395 us MUL_MAT_ID(ffn_moe_down-71): 201 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 365 us MUL_MAT_ID(ffn_moe_down-72): 310 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 383 us MUL_MAT_ID(ffn_moe_down-73): 207 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 412 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 393 us MUL_MAT_ID(ffn_moe_down-75): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 444 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 392 us MUL_MAT_ID(ffn_moe_down-77): 197 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 420 us MUL_MAT_ID(ffn_moe_down-78): 207 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 386 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 364 us MUL_MAT_ID(ffn_moe_down-80): 193 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 369 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 368 us MUL_MAT_ID(ffn_moe_down-82): 202 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 394 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 386 us MUL_MAT_ID(ffn_moe_down-84): 242 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 365 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 398 us MUL_MAT_ID(ffn_moe_down-86): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 362 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 430 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 405 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 366 us MUL_MAT_ID(ffn_moe_down-90): 318 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 355 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 51 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 394 us MUL_MAT_ID(ffn_moe_down-25): 204 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 357 us MUL_MAT_ID(ffn_moe_down-26): 314 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 381 us MUL_MAT_ID(ffn_moe_down-27): 185 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 372 us MUL_MAT_ID(ffn_moe_down-28): 199 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 465 us MUL_MAT_ID(ffn_moe_down-29): 490 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 411 us MUL_MAT_ID(ffn_moe_down-30): 184 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 415 us MUL_MAT_ID(ffn_moe_down-31): 326 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 376 us MUL_MAT_ID(ffn_moe_down-32): 209 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 416 us MUL_MAT_ID(ffn_moe_down-33): 530 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 433 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 375 us MUL_MAT_ID(ffn_moe_down-35): 177 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 380 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 373 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 360 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 350 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 406 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 374 us MUL_MAT_ID(ffn_moe_down-41): 237 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 364 us MUL_MAT_ID(ffn_moe_down-42): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 401 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 389 us MUL_MAT_ID(ffn_moe_down-44): 299 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 376 us MUL_MAT_ID(ffn_moe_down-45): 229 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 396 us MUL_MAT_ID(ffn_moe_down-46): 185 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 366 us MUL_MAT_ID(ffn_moe_down-47): 376 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 402 us MUL_MAT_ID(ffn_moe_down-48): 438 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 383 us MUL_MAT_ID(ffn_moe_down-49): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 370 us MUL_MAT_ID(ffn_moe_down-50): 210 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 384 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 435 us MUL_MAT_ID(ffn_moe_down-52): 178 us ggml_barrier(...): 263 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 413 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 381 us MUL_MAT_ID(ffn_moe_down-54): 195 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 381 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 378 us MUL_MAT_ID(ffn_moe_down-56): 210 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 379 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 402 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 411 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 384 us MUL_MAT_ID(ffn_moe_down-60): 228 us ggml_barrier(...): 390 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 338 us MUL_MAT_ID(ffn_moe_down-61): 206 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 398 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 351 us MUL_MAT_ID(ffn_moe_down-63): 203 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 383 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 358 us MUL_MAT_ID(ffn_moe_down-65): 208 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 348 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 380 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 388 us MUL_MAT_ID(ffn_moe_down-68): 188 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 402 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 369 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 516 us MUL_MAT_ID(ffn_moe_down-71): 197 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 374 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 384 us MUL_MAT_ID(ffn_moe_down-73): 202 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 380 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 347 us MUL_MAT_ID(ffn_moe_down-75): 194 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 359 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 378 us MUL_MAT_ID(ffn_moe_down-77): 202 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 366 us MUL_MAT_ID(ffn_moe_down-78): 394 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 356 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 391 us MUL_MAT_ID(ffn_moe_down-80): 225 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 369 us MUL_MAT_ID(ffn_moe_down-81): 189 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 383 us MUL_MAT_ID(ffn_moe_down-82): 203 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 396 us MUL_MAT_ID(ffn_moe_down-83): 565 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 373 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 346 us MUL_MAT_ID(ffn_moe_down-85): 244 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 393 us MUL_MAT_ID(ffn_moe_down-86): 232 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 340 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 389 us MUL_MAT_ID(ffn_moe_down-88): 214 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 403 us MUL_MAT_ID(ffn_moe_down-89): 224 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 358 us MUL_MAT_ID(ffn_moe_down-90): 245 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 373 us MUL_MAT_ID(ffn_moe_down-91): 236 us ggml_barrier(...): 35 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 396 us MUL_MAT_ID(ffn_moe_down-25): 271 us ggml_barrier(...): 259 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 356 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 385 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 399 us MUL_MAT_ID(ffn_moe_down-28): 216 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 356 us MUL_MAT_ID(ffn_moe_down-29): 209 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 393 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 379 us MUL_MAT_ID(ffn_moe_down-31): 191 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 426 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 377 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 406 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 363 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 366 us MUL_MAT_ID(ffn_moe_down-36): 205 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 363 us MUL_MAT_ID(ffn_moe_down-37): 307 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 379 us MUL_MAT_ID(ffn_moe_down-38): 194 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 352 us MUL_MAT_ID(ffn_moe_down-39): 654 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 397 us MUL_MAT_ID(ffn_moe_down-40): 256 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 374 us MUL_MAT_ID(ffn_moe_down-41): 361 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 397 us MUL_MAT_ID(ffn_moe_down-42): 202 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 374 us MUL_MAT_ID(ffn_moe_down-43): 208 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 365 us MUL_MAT_ID(ffn_moe_down-44): 203 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 372 us MUL_MAT_ID(ffn_moe_down-45): 190 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 377 us MUL_MAT_ID(ffn_moe_down-46): 369 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 398 us MUL_MAT_ID(ffn_moe_down-47): 411 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 361 us MUL_MAT_ID(ffn_moe_down-48): 399 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 270 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 360 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 512 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 373 us MUL_MAT_ID(ffn_moe_down-52): 395 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 368 us MUL_MAT_ID(ffn_moe_down-53): 228 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 358 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 413 us MUL_MAT_ID(ffn_moe_down-55): 189 us ggml_barrier(...): 317 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 380 us MUL_MAT_ID(ffn_moe_down-56): 217 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 348 us MUL_MAT_ID(ffn_moe_down-57): 200 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 393 us MUL_MAT_ID(ffn_moe_down-58): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 404 us MUL_MAT_ID(ffn_moe_down-59): 437 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 361 us MUL_MAT_ID(ffn_moe_down-60): 212 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 386 us MUL_MAT_ID(ffn_moe_down-61): 343 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 412 us MUL_MAT_ID(ffn_moe_down-62): 206 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 199 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 397 us MUL_MAT_ID(ffn_moe_down-64): 393 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 352 us MUL_MAT_ID(ffn_moe_down-65): 376 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 374 us MUL_MAT_ID(ffn_moe_down-66): 414 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 398 us MUL_MAT_ID(ffn_moe_down-67): 242 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 383 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 360 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 403 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 377 us MUL_MAT_ID(ffn_moe_down-71): 343 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 368 us MUL_MAT_ID(ffn_moe_down-72): 159 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 394 us MUL_MAT_ID(ffn_moe_down-73): 207 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 358 us MUL_MAT_ID(ffn_moe_down-74): 222 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 374 us MUL_MAT_ID(ffn_moe_down-75): 180 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 396 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 257 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 357 us MUL_MAT_ID(ffn_moe_down-78): 430 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 412 us MUL_MAT_ID(ffn_moe_down-79): 187 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 365 us MUL_MAT_ID(ffn_moe_down-80): 387 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 350 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 377 us MUL_MAT_ID(ffn_moe_down-82): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 373 us MUL_MAT_ID(ffn_moe_down-83): 189 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 374 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 372 us MUL_MAT_ID(ffn_moe_down-85): 222 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 388 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 392 us MUL_MAT_ID(ffn_moe_down-87): 363 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 409 us MUL_MAT_ID(ffn_moe_down-88): 301 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 391 us MUL_MAT_ID(ffn_moe_down-89): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 364 us MUL_MAT_ID(ffn_moe_down-90): 675 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 393 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 50 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 599 us MUL_MAT_ID(ffn_moe_down-25): 239 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 376 us MUL_MAT_ID(ffn_moe_down-26): 218 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 367 us MUL_MAT_ID(ffn_moe_down-27): 338 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 400 us MUL_MAT_ID(ffn_moe_down-28): 193 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 380 us MUL_MAT_ID(ffn_moe_down-29): 203 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 363 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 377 us MUL_MAT_ID(ffn_moe_down-31): 224 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 379 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 397 us MUL_MAT_ID(ffn_moe_down-33): 198 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 400 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 402 us MUL_MAT_ID(ffn_moe_down-35): 196 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 367 us MUL_MAT_ID(ffn_moe_down-36): 209 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 389 us MUL_MAT_ID(ffn_moe_down-37): 269 us ggml_barrier(...): 234 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 380 us MUL_MAT_ID(ffn_moe_down-38): 227 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 386 us MUL_MAT_ID(ffn_moe_down-39): 263 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 402 us MUL_MAT_ID(ffn_moe_down-40): 190 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 413 us MUL_MAT_ID(ffn_moe_down-41): 202 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 404 us MUL_MAT_ID(ffn_moe_down-42): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 390 us MUL_MAT_ID(ffn_moe_down-43): 399 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 358 us MUL_MAT_ID(ffn_moe_down-44): 202 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 410 us MUL_MAT_ID(ffn_moe_down-45): 230 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 385 us MUL_MAT_ID(ffn_moe_down-46): 424 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 363 us MUL_MAT_ID(ffn_moe_down-47): 182 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 379 us MUL_MAT_ID(ffn_moe_down-48): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 368 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 363 us MUL_MAT_ID(ffn_moe_down-50): 215 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 388 us MUL_MAT_ID(ffn_moe_down-51): 186 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 409 us MUL_MAT_ID(ffn_moe_down-52): 216 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 385 us MUL_MAT_ID(ffn_moe_down-53): 191 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 359 us MUL_MAT_ID(ffn_moe_down-54): 192 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 368 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 375 us MUL_MAT_ID(ffn_moe_down-56): 212 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 385 us MUL_MAT_ID(ffn_moe_down-57): 206 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 366 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 351 us MUL_MAT_ID(ffn_moe_down-59): 187 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 356 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 414 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 376 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 386 us MUL_MAT_ID(ffn_moe_down-63): 202 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 415 us MUL_MAT_ID(ffn_moe_down-64): 186 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 407 us MUL_MAT_ID(ffn_moe_down-65): 235 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 377 us MUL_MAT_ID(ffn_moe_down-66): 222 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 395 us MUL_MAT_ID(ffn_moe_down-67): 224 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 350 us MUL_MAT_ID(ffn_moe_down-68): 288 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 403 us MUL_MAT_ID(ffn_moe_down-69): 186 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 394 us MUL_MAT_ID(ffn_moe_down-70): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 415 us MUL_MAT_ID(ffn_moe_down-71): 207 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 392 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 385 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 347 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 399 us MUL_MAT_ID(ffn_moe_down-75): 357 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 394 us MUL_MAT_ID(ffn_moe_down-76): 185 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 376 us MUL_MAT_ID(ffn_moe_down-77): 193 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 381 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 377 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 366 us MUL_MAT_ID(ffn_moe_down-80): 207 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 374 us MUL_MAT_ID(ffn_moe_down-81): 297 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 361 us MUL_MAT_ID(ffn_moe_down-82): 206 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 392 us MUL_MAT_ID(ffn_moe_down-83): 191 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 364 us MUL_MAT_ID(ffn_moe_down-84): 240 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 378 us MUL_MAT_ID(ffn_moe_down-85): 657 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 373 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 402 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 392 us MUL_MAT_ID(ffn_moe_down-88): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 382 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 362 us MUL_MAT_ID(ffn_moe_down-90): 329 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 366 us MUL_MAT_ID(ffn_moe_down-91): 233 us ggml_barrier(...): 70 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 423 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 380 us MUL_MAT_ID(ffn_moe_down-27): 186 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 352 us MUL_MAT_ID(ffn_moe_down-28): 212 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 402 us MUL_MAT_ID(ffn_moe_down-29): 182 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 399 us MUL_MAT_ID(ffn_moe_down-30): 189 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 373 us MUL_MAT_ID(ffn_moe_down-31): 391 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 368 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 412 us MUL_MAT_ID(ffn_moe_down-33): 268 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 403 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 391 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 345 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 374 us MUL_MAT_ID(ffn_moe_down-37): 245 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 379 us MUL_MAT_ID(ffn_moe_down-38): 511 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 402 us MUL_MAT_ID(ffn_moe_down-39): 243 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 365 us MUL_MAT_ID(ffn_moe_down-40): 430 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 391 us MUL_MAT_ID(ffn_moe_down-41): 210 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 394 us MUL_MAT_ID(ffn_moe_down-42): 195 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 396 us MUL_MAT_ID(ffn_moe_down-43): 203 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 358 us MUL_MAT_ID(ffn_moe_down-44): 280 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 371 us MUL_MAT_ID(ffn_moe_down-45): 210 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 394 us MUL_MAT_ID(ffn_moe_down-46): 500 us ggml_barrier(...): 392 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 390 us MUL_MAT_ID(ffn_moe_down-47): 267 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 396 us MUL_MAT_ID(ffn_moe_down-48): 205 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 378 us MUL_MAT_ID(ffn_moe_down-49): 227 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 376 us MUL_MAT_ID(ffn_moe_down-50): 230 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 381 us MUL_MAT_ID(ffn_moe_down-51): 186 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 401 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 388 us MUL_MAT_ID(ffn_moe_down-53): 354 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 382 us MUL_MAT_ID(ffn_moe_down-54): 189 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 363 us MUL_MAT_ID(ffn_moe_down-55): 359 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 384 us MUL_MAT_ID(ffn_moe_down-56): 207 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 396 us MUL_MAT_ID(ffn_moe_down-57): 235 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 370 us MUL_MAT_ID(ffn_moe_down-58): 444 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 370 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 397 us MUL_MAT_ID(ffn_moe_down-60): 208 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 404 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 363 us MUL_MAT_ID(ffn_moe_down-62): 356 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 390 us MUL_MAT_ID(ffn_moe_down-63): 181 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 384 us MUL_MAT_ID(ffn_moe_down-64): 210 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 370 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 393 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 420 us MUL_MAT_ID(ffn_moe_down-67): 206 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 374 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 410 us MUL_MAT_ID(ffn_moe_down-69): 204 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 358 us MUL_MAT_ID(ffn_moe_down-70): 395 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 383 us MUL_MAT_ID(ffn_moe_down-71): 222 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 413 us MUL_MAT_ID(ffn_moe_down-72): 219 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 371 us MUL_MAT_ID(ffn_moe_down-73): 234 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 376 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 385 us MUL_MAT_ID(ffn_moe_down-75): 216 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 393 us MUL_MAT_ID(ffn_moe_down-76): 268 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 353 us MUL_MAT_ID(ffn_moe_down-77): 303 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 381 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 374 us MUL_MAT_ID(ffn_moe_down-79): 216 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 385 us MUL_MAT_ID(ffn_moe_down-80): 206 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 388 us MUL_MAT_ID(ffn_moe_down-82): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 391 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 405 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 392 us MUL_MAT_ID(ffn_moe_down-85): 256 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 379 us MUL_MAT_ID(ffn_moe_down-86): 223 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 396 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 368 us MUL_MAT_ID(ffn_moe_down-88): 211 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 387 us MUL_MAT_ID(ffn_moe_down-89): 204 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 382 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 414 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 17 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 406 us MUL_MAT_ID(ffn_moe_down-25): 186 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 389 us MUL_MAT_ID(ffn_moe_down-26): 197 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 403 us MUL_MAT_ID(ffn_moe_down-27): 182 us ggml_barrier(...): 292 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 402 us MUL_MAT_ID(ffn_moe_down-28): 375 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 362 us MUL_MAT_ID(ffn_moe_down-29): 181 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 408 us MUL_MAT_ID(ffn_moe_down-30): 192 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 376 us MUL_MAT_ID(ffn_moe_down-31): 207 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 417 us MUL_MAT_ID(ffn_moe_down-32): 226 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 353 us MUL_MAT_ID(ffn_moe_down-33): 186 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 450 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 353 us MUL_MAT_ID(ffn_moe_down-35): 291 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 356 us MUL_MAT_ID(ffn_moe_down-36): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 361 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 385 us MUL_MAT_ID(ffn_moe_down-38): 200 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 390 us MUL_MAT_ID(ffn_moe_down-39): 231 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 357 us MUL_MAT_ID(ffn_moe_down-40): 416 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 341 us MUL_MAT_ID(ffn_moe_down-41): 202 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 409 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 433 us MUL_MAT_ID(ffn_moe_down-43): 219 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 398 us MUL_MAT_ID(ffn_moe_down-44): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 490 us MUL_MAT_ID(ffn_moe_down-45): 200 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 407 us MUL_MAT_ID(ffn_moe_down-46): 215 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 402 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 360 us MUL_MAT_ID(ffn_moe_down-48): 198 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 365 us MUL_MAT_ID(ffn_moe_down-49): 432 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 379 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 387 us MUL_MAT_ID(ffn_moe_down-51): 351 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 387 us MUL_MAT_ID(ffn_moe_down-52): 375 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 456 us MUL_MAT_ID(ffn_moe_down-53): 164 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 378 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 407 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 373 us MUL_MAT_ID(ffn_moe_down-56): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 364 us MUL_MAT_ID(ffn_moe_down-57): 371 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 388 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 406 us MUL_MAT_ID(ffn_moe_down-59): 438 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 365 us MUL_MAT_ID(ffn_moe_down-60): 286 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 383 us MUL_MAT_ID(ffn_moe_down-61): 371 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 352 us MUL_MAT_ID(ffn_moe_down-62): 192 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 223 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 421 us MUL_MAT_ID(ffn_moe_down-64): 171 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 428 us MUL_MAT_ID(ffn_moe_down-65): 288 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 359 us MUL_MAT_ID(ffn_moe_down-66): 184 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 383 us MUL_MAT_ID(ffn_moe_down-67): 409 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 404 us MUL_MAT_ID(ffn_moe_down-68): 200 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 379 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 398 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 409 us MUL_MAT_ID(ffn_moe_down-71): 246 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 360 us MUL_MAT_ID(ffn_moe_down-72): 192 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 354 us MUL_MAT_ID(ffn_moe_down-73): 223 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 373 us MUL_MAT_ID(ffn_moe_down-74): 170 us ggml_barrier(...): 284 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 351 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 384 us MUL_MAT_ID(ffn_moe_down-76): 188 us ggml_barrier(...): 310 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 379 us MUL_MAT_ID(ffn_moe_down-77): 219 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 369 us MUL_MAT_ID(ffn_moe_down-78): 191 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 391 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 394 us MUL_MAT_ID(ffn_moe_down-80): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 399 us MUL_MAT_ID(ffn_moe_down-81): 226 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 386 us MUL_MAT_ID(ffn_moe_down-82): 202 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 408 us MUL_MAT_ID(ffn_moe_down-83): 411 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 369 us MUL_MAT_ID(ffn_moe_down-84): 242 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 362 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 376 us MUL_MAT_ID(ffn_moe_down-86): 211 us ggml_barrier(...): 285 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 366 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 408 us MUL_MAT_ID(ffn_moe_down-88): 490 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 405 us MUL_MAT_ID(ffn_moe_down-89): 200 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 364 us MUL_MAT_ID(ffn_moe_down-90): 242 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 367 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 68 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 188 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 395 us MUL_MAT_ID(ffn_moe_down-26): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 366 us MUL_MAT_ID(ffn_moe_down-27): 210 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 379 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 430 us MUL_MAT_ID(ffn_moe_down-29): 197 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 359 us MUL_MAT_ID(ffn_moe_down-30): 209 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 405 us MUL_MAT_ID(ffn_moe_down-31): 200 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 405 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 370 us MUL_MAT_ID(ffn_moe_down-33): 209 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 391 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 418 us MUL_MAT_ID(ffn_moe_down-35): 206 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 385 us MUL_MAT_ID(ffn_moe_down-36): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 376 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 401 us MUL_MAT_ID(ffn_moe_down-38): 187 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 372 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 391 us MUL_MAT_ID(ffn_moe_down-40): 235 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 383 us MUL_MAT_ID(ffn_moe_down-41): 536 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 404 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 390 us MUL_MAT_ID(ffn_moe_down-43): 554 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 354 us MUL_MAT_ID(ffn_moe_down-44): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 390 us MUL_MAT_ID(ffn_moe_down-45): 216 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 377 us MUL_MAT_ID(ffn_moe_down-46): 203 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 371 us MUL_MAT_ID(ffn_moe_down-47): 217 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 344 us MUL_MAT_ID(ffn_moe_down-48): 188 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 373 us MUL_MAT_ID(ffn_moe_down-49): 206 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 362 us MUL_MAT_ID(ffn_moe_down-50): 201 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 380 us MUL_MAT_ID(ffn_moe_down-51): 188 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 406 us MUL_MAT_ID(ffn_moe_down-52): 195 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 354 us MUL_MAT_ID(ffn_moe_down-53): 198 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 381 us MUL_MAT_ID(ffn_moe_down-54): 189 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 383 us MUL_MAT_ID(ffn_moe_down-55): 241 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 405 us MUL_MAT_ID(ffn_moe_down-56): 197 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 356 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 390 us MUL_MAT_ID(ffn_moe_down-58): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 411 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 370 us MUL_MAT_ID(ffn_moe_down-60): 419 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 386 us MUL_MAT_ID(ffn_moe_down-61): 370 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 372 us MUL_MAT_ID(ffn_moe_down-62): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 389 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 381 us MUL_MAT_ID(ffn_moe_down-64): 285 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 368 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 401 us MUL_MAT_ID(ffn_moe_down-66): 208 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 400 us MUL_MAT_ID(ffn_moe_down-67): 189 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 360 us MUL_MAT_ID(ffn_moe_down-68): 201 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 378 us MUL_MAT_ID(ffn_moe_down-69): 387 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 354 us MUL_MAT_ID(ffn_moe_down-70): 406 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 370 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 418 us MUL_MAT_ID(ffn_moe_down-72): 186 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 378 us MUL_MAT_ID(ffn_moe_down-73): 201 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 392 us MUL_MAT_ID(ffn_moe_down-74): 217 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 387 us MUL_MAT_ID(ffn_moe_down-75): 181 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 373 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 359 us MUL_MAT_ID(ffn_moe_down-77): 209 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 385 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 374 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 397 us MUL_MAT_ID(ffn_moe_down-80): 216 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 378 us MUL_MAT_ID(ffn_moe_down-81): 366 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 389 us MUL_MAT_ID(ffn_moe_down-82): 203 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 397 us MUL_MAT_ID(ffn_moe_down-83): 379 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 366 us MUL_MAT_ID(ffn_moe_down-84): 264 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 376 us MUL_MAT_ID(ffn_moe_down-85): 248 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 378 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 489 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 215 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 399 us MUL_MAT_ID(ffn_moe_down-88): 367 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 376 us MUL_MAT_ID(ffn_moe_down-89): 417 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 372 us MUL_MAT_ID(ffn_moe_down-90): 664 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 398 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 31 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 183 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 365 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 389 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 389 us MUL_MAT_ID(ffn_moe_down-28): 206 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 385 us MUL_MAT_ID(ffn_moe_down-29): 194 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 369 us MUL_MAT_ID(ffn_moe_down-30): 223 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 403 us MUL_MAT_ID(ffn_moe_down-31): 233 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 402 us MUL_MAT_ID(ffn_moe_down-32): 205 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 410 us MUL_MAT_ID(ffn_moe_down-33): 190 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 389 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 417 us MUL_MAT_ID(ffn_moe_down-35): 273 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 387 us MUL_MAT_ID(ffn_moe_down-36): 219 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 361 us MUL_MAT_ID(ffn_moe_down-37): 299 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 361 us MUL_MAT_ID(ffn_moe_down-38): 185 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 357 us MUL_MAT_ID(ffn_moe_down-39): 228 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 376 us MUL_MAT_ID(ffn_moe_down-40): 241 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 371 us MUL_MAT_ID(ffn_moe_down-41): 387 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 397 us MUL_MAT_ID(ffn_moe_down-42): 186 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 371 us MUL_MAT_ID(ffn_moe_down-43): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 348 us MUL_MAT_ID(ffn_moe_down-44): 185 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 370 us MUL_MAT_ID(ffn_moe_down-45): 362 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 374 us MUL_MAT_ID(ffn_moe_down-46): 330 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 402 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 354 us MUL_MAT_ID(ffn_moe_down-48): 190 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 350 us MUL_MAT_ID(ffn_moe_down-49): 218 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 378 us MUL_MAT_ID(ffn_moe_down-50): 351 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 365 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 386 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 369 us MUL_MAT_ID(ffn_moe_down-53): 201 us ggml_barrier(...): 333 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 373 us MUL_MAT_ID(ffn_moe_down-54): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 405 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 387 us MUL_MAT_ID(ffn_moe_down-56): 306 us ggml_barrier(...): 302 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 366 us MUL_MAT_ID(ffn_moe_down-57): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 383 us MUL_MAT_ID(ffn_moe_down-58): 398 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 394 us MUL_MAT_ID(ffn_moe_down-59): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 423 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 381 us MUL_MAT_ID(ffn_moe_down-61): 183 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 352 us MUL_MAT_ID(ffn_moe_down-62): 214 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 360 us MUL_MAT_ID(ffn_moe_down-63): 187 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 408 us MUL_MAT_ID(ffn_moe_down-64): 263 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 370 us MUL_MAT_ID(ffn_moe_down-65): 210 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 415 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 411 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 374 us MUL_MAT_ID(ffn_moe_down-68): 207 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 384 us MUL_MAT_ID(ffn_moe_down-69): 207 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 456 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 404 us MUL_MAT_ID(ffn_moe_down-71): 188 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 356 us MUL_MAT_ID(ffn_moe_down-72): 547 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 209 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 383 us MUL_MAT_ID(ffn_moe_down-74): 221 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 342 us MUL_MAT_ID(ffn_moe_down-75): 198 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 385 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 375 us MUL_MAT_ID(ffn_moe_down-77): 233 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 358 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 404 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 398 us MUL_MAT_ID(ffn_moe_down-80): 223 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 385 us MUL_MAT_ID(ffn_moe_down-81): 186 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 363 us MUL_MAT_ID(ffn_moe_down-82): 180 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 362 us MUL_MAT_ID(ffn_moe_down-83): 193 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 353 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 354 us MUL_MAT_ID(ffn_moe_down-85): 221 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 405 us MUL_MAT_ID(ffn_moe_down-86): 364 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 374 us MUL_MAT_ID(ffn_moe_down-87): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 397 us MUL_MAT_ID(ffn_moe_down-88): 181 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 394 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 368 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 346 us MUL_MAT_ID(ffn_moe_down-91): 694 us ggml_barrier(...): 80 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 385 us MUL_MAT_ID(ffn_moe_down-25): 179 us ggml_barrier(...): 214 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 487 us MUL_MAT_ID(ffn_moe_down-26): 184 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 412 us MUL_MAT_ID(ffn_moe_down-27): 284 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 360 us MUL_MAT_ID(ffn_moe_down-28): 407 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 370 us MUL_MAT_ID(ffn_moe_down-29): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 396 us MUL_MAT_ID(ffn_moe_down-30): 401 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 377 us MUL_MAT_ID(ffn_moe_down-31): 395 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 395 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 377 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 348 us MUL_MAT_ID(ffn_moe_down-34): 388 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 388 us MUL_MAT_ID(ffn_moe_down-35): 184 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 378 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 367 us MUL_MAT_ID(ffn_moe_down-37): 641 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 390 us MUL_MAT_ID(ffn_moe_down-38): 222 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 345 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 403 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 367 us MUL_MAT_ID(ffn_moe_down-41): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 361 us MUL_MAT_ID(ffn_moe_down-42): 224 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 364 us MUL_MAT_ID(ffn_moe_down-43): 201 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 366 us MUL_MAT_ID(ffn_moe_down-44): 202 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 379 us MUL_MAT_ID(ffn_moe_down-45): 193 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 411 us MUL_MAT_ID(ffn_moe_down-46): 202 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 357 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 400 us MUL_MAT_ID(ffn_moe_down-48): 222 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 365 us MUL_MAT_ID(ffn_moe_down-49): 353 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 364 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 353 us MUL_MAT_ID(ffn_moe_down-51): 193 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 412 us MUL_MAT_ID(ffn_moe_down-52): 249 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 371 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 385 us MUL_MAT_ID(ffn_moe_down-54): 217 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 363 us MUL_MAT_ID(ffn_moe_down-55): 194 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 380 us MUL_MAT_ID(ffn_moe_down-56): 406 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 368 us MUL_MAT_ID(ffn_moe_down-57): 299 us ggml_barrier(...): 201 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 381 us MUL_MAT_ID(ffn_moe_down-58): 262 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 400 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 407 us MUL_MAT_ID(ffn_moe_down-60): 268 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 352 us MUL_MAT_ID(ffn_moe_down-61): 374 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 352 us MUL_MAT_ID(ffn_moe_down-62): 310 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 381 us MUL_MAT_ID(ffn_moe_down-63): 390 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 412 us MUL_MAT_ID(ffn_moe_down-64): 200 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 363 us MUL_MAT_ID(ffn_moe_down-65): 221 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 392 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 395 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 346 us MUL_MAT_ID(ffn_moe_down-68): 201 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 391 us MUL_MAT_ID(ffn_moe_down-69): 379 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 379 us MUL_MAT_ID(ffn_moe_down-70): 194 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 386 us MUL_MAT_ID(ffn_moe_down-71): 201 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 370 us MUL_MAT_ID(ffn_moe_down-72): 186 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 385 us MUL_MAT_ID(ffn_moe_down-73): 460 us ggml_barrier(...): 327 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 394 us MUL_MAT_ID(ffn_moe_down-74): 306 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 336 us MUL_MAT_ID(ffn_moe_down-75): 189 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 412 us MUL_MAT_ID(ffn_moe_down-76): 286 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 381 us MUL_MAT_ID(ffn_moe_down-77): 200 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 400 us MUL_MAT_ID(ffn_moe_down-78): 226 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 359 us MUL_MAT_ID(ffn_moe_down-79): 261 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 404 us MUL_MAT_ID(ffn_moe_down-80): 212 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 400 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 370 us MUL_MAT_ID(ffn_moe_down-82): 428 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 423 us MUL_MAT_ID(ffn_moe_down-83): 194 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 412 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 413 us MUL_MAT_ID(ffn_moe_down-85): 252 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 381 us MUL_MAT_ID(ffn_moe_down-86): 187 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 406 us MUL_MAT_ID(ffn_moe_down-87): 256 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 392 us MUL_MAT_ID(ffn_moe_down-88): 210 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 403 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 376 us MUL_MAT_ID(ffn_moe_down-90): 208 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 416 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 14 us GET_ROWS(inp_embd): 18 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 387 us MUL_MAT_ID(ffn_moe_down-25): 206 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 384 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 400 us MUL_MAT_ID(ffn_moe_down-27): 220 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 372 us MUL_MAT_ID(ffn_moe_down-28): 209 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 396 us MUL_MAT_ID(ffn_moe_down-29): 273 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 408 us MUL_MAT_ID(ffn_moe_down-30): 208 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 356 us MUL_MAT_ID(ffn_moe_down-31): 436 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 383 us MUL_MAT_ID(ffn_moe_down-32): 208 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 348 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 403 us MUL_MAT_ID(ffn_moe_down-34): 203 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 368 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 381 us MUL_MAT_ID(ffn_moe_down-36): 195 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 390 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 373 us MUL_MAT_ID(ffn_moe_down-39): 233 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 365 us MUL_MAT_ID(ffn_moe_down-40): 242 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 359 us MUL_MAT_ID(ffn_moe_down-41): 345 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 367 us MUL_MAT_ID(ffn_moe_down-42): 207 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 386 us MUL_MAT_ID(ffn_moe_down-43): 368 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 356 us MUL_MAT_ID(ffn_moe_down-44): 203 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 375 us MUL_MAT_ID(ffn_moe_down-45): 224 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 403 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 395 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 395 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 400 us MUL_MAT_ID(ffn_moe_down-49): 198 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 379 us MUL_MAT_ID(ffn_moe_down-50): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 400 us MUL_MAT_ID(ffn_moe_down-51): 358 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 392 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 376 us MUL_MAT_ID(ffn_moe_down-53): 244 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 376 us MUL_MAT_ID(ffn_moe_down-54): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 207 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 411 us MUL_MAT_ID(ffn_moe_down-56): 380 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 403 us MUL_MAT_ID(ffn_moe_down-57): 184 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 403 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 375 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 382 us MUL_MAT_ID(ffn_moe_down-60): 274 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 384 us MUL_MAT_ID(ffn_moe_down-61): 204 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 209 us ggml_barrier(...): 387 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 393 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 412 us MUL_MAT_ID(ffn_moe_down-64): 206 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 363 us MUL_MAT_ID(ffn_moe_down-65): 193 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 377 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 380 us MUL_MAT_ID(ffn_moe_down-67): 190 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 414 us MUL_MAT_ID(ffn_moe_down-68): 225 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 382 us MUL_MAT_ID(ffn_moe_down-69): 192 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 350 us MUL_MAT_ID(ffn_moe_down-70): 226 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 381 us MUL_MAT_ID(ffn_moe_down-71): 390 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 389 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 378 us MUL_MAT_ID(ffn_moe_down-73): 197 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 404 us MUL_MAT_ID(ffn_moe_down-74): 255 us ggml_barrier(...): 276 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 358 us MUL_MAT_ID(ffn_moe_down-75): 189 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 364 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 401 us MUL_MAT_ID(ffn_moe_down-77): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 422 us MUL_MAT_ID(ffn_moe_down-78): 187 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 361 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 357 us MUL_MAT_ID(ffn_moe_down-80): 190 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 365 us MUL_MAT_ID(ffn_moe_down-81): 219 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 381 us MUL_MAT_ID(ffn_moe_down-82): 211 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 366 us MUL_MAT_ID(ffn_moe_down-83): 209 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 353 us MUL_MAT_ID(ffn_moe_down-84): 225 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 391 us MUL_MAT_ID(ffn_moe_down-85): 228 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 380 us MUL_MAT_ID(ffn_moe_down-86): 366 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 375 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 396 us MUL_MAT_ID(ffn_moe_down-88): 183 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 388 us MUL_MAT_ID(ffn_moe_down-89): 349 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 436 us MUL_MAT_ID(ffn_moe_down-90): 661 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 342 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 68 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 376 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 423 us MUL_MAT_ID(ffn_moe_down-26): 184 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 351 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 377 us MUL_MAT_ID(ffn_moe_down-28): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 369 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 410 us MUL_MAT_ID(ffn_moe_down-30): 197 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 403 us MUL_MAT_ID(ffn_moe_down-31): 551 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 399 us MUL_MAT_ID(ffn_moe_down-32): 199 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 401 us MUL_MAT_ID(ffn_moe_down-33): 208 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 400 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 412 us MUL_MAT_ID(ffn_moe_down-35): 188 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 362 us MUL_MAT_ID(ffn_moe_down-36): 189 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 366 us MUL_MAT_ID(ffn_moe_down-37): 239 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 378 us MUL_MAT_ID(ffn_moe_down-38): 212 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 378 us MUL_MAT_ID(ffn_moe_down-39): 605 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 407 us MUL_MAT_ID(ffn_moe_down-40): 239 us ggml_barrier(...): 222 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 383 us MUL_MAT_ID(ffn_moe_down-41): 348 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 396 us MUL_MAT_ID(ffn_moe_down-42): 220 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 376 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 402 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 388 us MUL_MAT_ID(ffn_moe_down-45): 215 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 368 us MUL_MAT_ID(ffn_moe_down-46): 181 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 417 us MUL_MAT_ID(ffn_moe_down-47): 202 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 380 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 403 us MUL_MAT_ID(ffn_moe_down-49): 201 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 389 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 372 us MUL_MAT_ID(ffn_moe_down-51): 215 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 367 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 396 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 366 us MUL_MAT_ID(ffn_moe_down-54): 219 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 389 us MUL_MAT_ID(ffn_moe_down-55): 189 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 392 us MUL_MAT_ID(ffn_moe_down-56): 187 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 350 us MUL_MAT_ID(ffn_moe_down-57): 187 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 395 us MUL_MAT_ID(ffn_moe_down-58): 186 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 411 us MUL_MAT_ID(ffn_moe_down-59): 279 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 386 us MUL_MAT_ID(ffn_moe_down-60): 197 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 398 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 375 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 384 us MUL_MAT_ID(ffn_moe_down-63): 212 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 374 us MUL_MAT_ID(ffn_moe_down-64): 180 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 385 us MUL_MAT_ID(ffn_moe_down-65): 410 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 395 us MUL_MAT_ID(ffn_moe_down-66): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 409 us MUL_MAT_ID(ffn_moe_down-67): 200 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 356 us MUL_MAT_ID(ffn_moe_down-68): 203 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 386 us MUL_MAT_ID(ffn_moe_down-69): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 390 us MUL_MAT_ID(ffn_moe_down-70): 214 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 395 us MUL_MAT_ID(ffn_moe_down-71): 187 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 354 us MUL_MAT_ID(ffn_moe_down-72): 181 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 400 us MUL_MAT_ID(ffn_moe_down-73): 202 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 393 us MUL_MAT_ID(ffn_moe_down-74): 188 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 356 us MUL_MAT_ID(ffn_moe_down-75): 369 us ggml_barrier(...): 320 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 406 us MUL_MAT_ID(ffn_moe_down-76): 322 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 384 us MUL_MAT_ID(ffn_moe_down-77): 196 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 365 us MUL_MAT_ID(ffn_moe_down-78): 381 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 367 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 361 us MUL_MAT_ID(ffn_moe_down-80): 186 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 384 us MUL_MAT_ID(ffn_moe_down-81): 396 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 383 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 356 us MUL_MAT_ID(ffn_moe_down-83): 186 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 394 us MUL_MAT_ID(ffn_moe_down-84): 216 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 370 us MUL_MAT_ID(ffn_moe_down-85): 228 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 354 us MUL_MAT_ID(ffn_moe_down-86): 364 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 390 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 398 us MUL_MAT_ID(ffn_moe_down-88): 387 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 392 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 388 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 401 us MUL_MAT_ID(ffn_moe_down-91): 232 us ggml_barrier(...): 31 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 390 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 385 us MUL_MAT_ID(ffn_moe_down-26): 319 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 491 us MUL_MAT_ID(ffn_moe_down-27): 162 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 384 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 387 us MUL_MAT_ID(ffn_moe_down-29): 183 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 347 us MUL_MAT_ID(ffn_moe_down-30): 209 us ggml_barrier(...): 523 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 384 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 379 us MUL_MAT_ID(ffn_moe_down-32): 215 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 376 us MUL_MAT_ID(ffn_moe_down-33): 211 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 399 us MUL_MAT_ID(ffn_moe_down-34): 391 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 369 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 397 us MUL_MAT_ID(ffn_moe_down-36): 188 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 351 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 419 us MUL_MAT_ID(ffn_moe_down-38): 200 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 369 us MUL_MAT_ID(ffn_moe_down-39): 225 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 360 us MUL_MAT_ID(ffn_moe_down-40): 240 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 426 us MUL_MAT_ID(ffn_moe_down-41): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 397 us MUL_MAT_ID(ffn_moe_down-42): 205 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 404 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 386 us MUL_MAT_ID(ffn_moe_down-44): 185 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 378 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 372 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 366 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 408 us MUL_MAT_ID(ffn_moe_down-48): 219 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 340 us MUL_MAT_ID(ffn_moe_down-49): 202 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 373 us MUL_MAT_ID(ffn_moe_down-50): 251 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 377 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 350 us MUL_MAT_ID(ffn_moe_down-52): 182 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 397 us MUL_MAT_ID(ffn_moe_down-53): 184 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 417 us MUL_MAT_ID(ffn_moe_down-54): 190 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 353 us MUL_MAT_ID(ffn_moe_down-55): 533 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 422 us MUL_MAT_ID(ffn_moe_down-56): 381 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 423 us MUL_MAT_ID(ffn_moe_down-57): 356 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 414 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 379 us MUL_MAT_ID(ffn_moe_down-59): 190 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 386 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 392 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 398 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 359 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 393 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 369 us MUL_MAT_ID(ffn_moe_down-65): 333 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 426 us MUL_MAT_ID(ffn_moe_down-66): 360 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 364 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 355 us MUL_MAT_ID(ffn_moe_down-68): 372 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 397 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 377 us MUL_MAT_ID(ffn_moe_down-70): 212 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 350 us MUL_MAT_ID(ffn_moe_down-71): 216 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 381 us MUL_MAT_ID(ffn_moe_down-72): 192 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 374 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 402 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 356 us MUL_MAT_ID(ffn_moe_down-75): 203 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 355 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 369 us MUL_MAT_ID(ffn_moe_down-77): 204 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 402 us MUL_MAT_ID(ffn_moe_down-78): 201 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 404 us MUL_MAT_ID(ffn_moe_down-79): 195 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 379 us MUL_MAT_ID(ffn_moe_down-80): 185 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 384 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 390 us MUL_MAT_ID(ffn_moe_down-82): 187 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 382 us MUL_MAT_ID(ffn_moe_down-83): 184 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 360 us MUL_MAT_ID(ffn_moe_down-84): 230 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 399 us MUL_MAT_ID(ffn_moe_down-85): 223 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 386 us MUL_MAT_ID(ffn_moe_down-86): 220 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 370 us MUL_MAT_ID(ffn_moe_down-87): 208 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 364 us MUL_MAT_ID(ffn_moe_down-88): 254 us ggml_barrier(...): 252 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 359 us MUL_MAT_ID(ffn_moe_down-89): 218 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 412 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 394 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 24 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 462 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 362 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 402 us MUL_MAT_ID(ffn_moe_down-27): 537 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 365 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 406 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 355 us MUL_MAT_ID(ffn_moe_down-30): 351 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 488 us MUL_MAT_ID(ffn_moe_down-31): 204 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 371 us MUL_MAT_ID(ffn_moe_down-32): 342 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 427 us MUL_MAT_ID(ffn_moe_down-33): 225 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 392 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 397 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 418 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 336 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 408 us MUL_MAT_ID(ffn_moe_down-38): 184 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 375 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 362 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 376 us MUL_MAT_ID(ffn_moe_down-41): 372 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 348 us MUL_MAT_ID(ffn_moe_down-42): 210 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 378 us MUL_MAT_ID(ffn_moe_down-43): 245 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 361 us MUL_MAT_ID(ffn_moe_down-44): 383 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 368 us MUL_MAT_ID(ffn_moe_down-45): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 411 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 385 us MUL_MAT_ID(ffn_moe_down-47): 356 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 343 us MUL_MAT_ID(ffn_moe_down-48): 191 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 390 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 366 us MUL_MAT_ID(ffn_moe_down-50): 194 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 371 us MUL_MAT_ID(ffn_moe_down-51): 208 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 383 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 368 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 358 us MUL_MAT_ID(ffn_moe_down-54): 201 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 377 us MUL_MAT_ID(ffn_moe_down-55): 204 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 428 us MUL_MAT_ID(ffn_moe_down-56): 252 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 362 us MUL_MAT_ID(ffn_moe_down-57): 181 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 368 us MUL_MAT_ID(ffn_moe_down-58): 163 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 416 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 389 us MUL_MAT_ID(ffn_moe_down-60): 205 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 378 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 375 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 363 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 403 us MUL_MAT_ID(ffn_moe_down-64): 189 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 371 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 381 us MUL_MAT_ID(ffn_moe_down-66): 209 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 402 us MUL_MAT_ID(ffn_moe_down-67): 210 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 370 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 350 us MUL_MAT_ID(ffn_moe_down-69): 201 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 390 us MUL_MAT_ID(ffn_moe_down-70): 218 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 401 us MUL_MAT_ID(ffn_moe_down-71): 212 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 378 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 377 us MUL_MAT_ID(ffn_moe_down-73): 211 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 359 us MUL_MAT_ID(ffn_moe_down-74): 406 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 344 us MUL_MAT_ID(ffn_moe_down-75): 215 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 362 us MUL_MAT_ID(ffn_moe_down-76): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 387 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 385 us MUL_MAT_ID(ffn_moe_down-78): 182 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 362 us MUL_MAT_ID(ffn_moe_down-79): 200 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 371 us MUL_MAT_ID(ffn_moe_down-80): 194 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 368 us MUL_MAT_ID(ffn_moe_down-81): 184 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 387 us MUL_MAT_ID(ffn_moe_down-82): 191 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 394 us MUL_MAT_ID(ffn_moe_down-83): 207 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 384 us MUL_MAT_ID(ffn_moe_down-84): 217 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 388 us MUL_MAT_ID(ffn_moe_down-85): 245 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 406 us MUL_MAT_ID(ffn_moe_down-86): 209 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 379 us MUL_MAT_ID(ffn_moe_down-87): 565 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 393 us MUL_MAT_ID(ffn_moe_down-88): 478 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 394 us MUL_MAT_ID(ffn_moe_down-89): 390 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 389 us MUL_MAT_ID(ffn_moe_down-90): 237 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 362 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 399 us MUL_MAT_ID(ffn_moe_down-25): 197 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 391 us MUL_MAT_ID(ffn_moe_down-26): 191 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 374 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 403 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 374 us MUL_MAT_ID(ffn_moe_down-29): 218 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 402 us MUL_MAT_ID(ffn_moe_down-30): 200 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 405 us MUL_MAT_ID(ffn_moe_down-31): 418 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 401 us MUL_MAT_ID(ffn_moe_down-32): 209 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 360 us MUL_MAT_ID(ffn_moe_down-33): 189 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 385 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 370 us MUL_MAT_ID(ffn_moe_down-35): 201 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 408 us MUL_MAT_ID(ffn_moe_down-36): 218 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 372 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 412 us MUL_MAT_ID(ffn_moe_down-38): 337 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 400 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 414 us MUL_MAT_ID(ffn_moe_down-40): 260 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 407 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 406 us MUL_MAT_ID(ffn_moe_down-42): 216 us ggml_barrier(...): 332 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 384 us MUL_MAT_ID(ffn_moe_down-43): 220 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 372 us MUL_MAT_ID(ffn_moe_down-44): 199 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 398 us MUL_MAT_ID(ffn_moe_down-45): 204 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 371 us MUL_MAT_ID(ffn_moe_down-46): 185 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 357 us MUL_MAT_ID(ffn_moe_down-47): 211 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 359 us MUL_MAT_ID(ffn_moe_down-48): 184 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 364 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 373 us MUL_MAT_ID(ffn_moe_down-50): 207 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 387 us MUL_MAT_ID(ffn_moe_down-51): 380 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 346 us MUL_MAT_ID(ffn_moe_down-52): 181 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 368 us MUL_MAT_ID(ffn_moe_down-53): 298 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 366 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 399 us MUL_MAT_ID(ffn_moe_down-56): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 369 us MUL_MAT_ID(ffn_moe_down-57): 188 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 397 us MUL_MAT_ID(ffn_moe_down-58): 188 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 356 us MUL_MAT_ID(ffn_moe_down-59): 240 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 367 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 407 us MUL_MAT_ID(ffn_moe_down-61): 176 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 376 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 364 us MUL_MAT_ID(ffn_moe_down-63): 195 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 402 us MUL_MAT_ID(ffn_moe_down-64): 193 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 404 us MUL_MAT_ID(ffn_moe_down-65): 286 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 413 us MUL_MAT_ID(ffn_moe_down-66): 398 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 361 us MUL_MAT_ID(ffn_moe_down-67): 222 us ggml_barrier(...): 253 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 360 us MUL_MAT_ID(ffn_moe_down-68): 290 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 390 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 375 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 417 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 358 us MUL_MAT_ID(ffn_moe_down-72): 222 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 356 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 374 us MUL_MAT_ID(ffn_moe_down-74): 208 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 383 us MUL_MAT_ID(ffn_moe_down-75): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 377 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 368 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 370 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 383 us MUL_MAT_ID(ffn_moe_down-79): 207 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 372 us MUL_MAT_ID(ffn_moe_down-80): 253 us ggml_barrier(...): 219 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 395 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 365 us MUL_MAT_ID(ffn_moe_down-82): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 492 us MUL_MAT_ID(ffn_moe_down-83): 274 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 401 us MUL_MAT_ID(ffn_moe_down-84): 271 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 415 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 355 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 397 us MUL_MAT_ID(ffn_moe_down-87): 182 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 396 us MUL_MAT_ID(ffn_moe_down-88): 365 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 387 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 388 us MUL_MAT_ID(ffn_moe_down-90): 229 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 385 us MUL_MAT_ID(ffn_moe_down-91): 227 us ggml_barrier(...): 29 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 426 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 375 us MUL_MAT_ID(ffn_moe_down-26): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 376 us MUL_MAT_ID(ffn_moe_down-27): 211 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 375 us MUL_MAT_ID(ffn_moe_down-28): 261 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 561 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 369 us MUL_MAT_ID(ffn_moe_down-30): 206 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 401 us MUL_MAT_ID(ffn_moe_down-31): 185 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 366 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 394 us MUL_MAT_ID(ffn_moe_down-33): 219 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 417 us MUL_MAT_ID(ffn_moe_down-34): 205 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 378 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 217 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 387 us MUL_MAT_ID(ffn_moe_down-37): 251 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 395 us MUL_MAT_ID(ffn_moe_down-38): 204 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 422 us MUL_MAT_ID(ffn_moe_down-39): 244 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 392 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 418 us MUL_MAT_ID(ffn_moe_down-41): 208 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 357 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 367 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 373 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 387 us MUL_MAT_ID(ffn_moe_down-45): 218 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 377 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 410 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 520 us MUL_MAT_ID(ffn_moe_down-48): 227 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 356 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 359 us MUL_MAT_ID(ffn_moe_down-50): 239 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 403 us MUL_MAT_ID(ffn_moe_down-51): 203 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 357 us MUL_MAT_ID(ffn_moe_down-52): 207 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 366 us MUL_MAT_ID(ffn_moe_down-53): 536 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 389 us MUL_MAT_ID(ffn_moe_down-54): 224 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 368 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 380 us MUL_MAT_ID(ffn_moe_down-56): 203 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 373 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 349 us MUL_MAT_ID(ffn_moe_down-58): 171 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 412 us MUL_MAT_ID(ffn_moe_down-59): 200 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 367 us MUL_MAT_ID(ffn_moe_down-60): 182 us ggml_barrier(...): 303 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 399 us MUL_MAT_ID(ffn_moe_down-61): 184 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 373 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 376 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 373 us MUL_MAT_ID(ffn_moe_down-64): 188 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 406 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 369 us MUL_MAT_ID(ffn_moe_down-66): 179 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 360 us MUL_MAT_ID(ffn_moe_down-67): 205 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 391 us MUL_MAT_ID(ffn_moe_down-68): 207 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 374 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 368 us MUL_MAT_ID(ffn_moe_down-70): 200 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 379 us MUL_MAT_ID(ffn_moe_down-71): 208 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 373 us MUL_MAT_ID(ffn_moe_down-72): 189 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 409 us MUL_MAT_ID(ffn_moe_down-73): 380 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 353 us MUL_MAT_ID(ffn_moe_down-74): 192 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 361 us MUL_MAT_ID(ffn_moe_down-75): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 410 us MUL_MAT_ID(ffn_moe_down-76): 361 us ggml_barrier(...): 305 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 382 us MUL_MAT_ID(ffn_moe_down-77): 188 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 356 us MUL_MAT_ID(ffn_moe_down-78): 294 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 462 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 227 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 370 us MUL_MAT_ID(ffn_moe_down-80): 203 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 370 us MUL_MAT_ID(ffn_moe_down-81): 284 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 374 us MUL_MAT_ID(ffn_moe_down-82): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 375 us MUL_MAT_ID(ffn_moe_down-83): 184 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 396 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 360 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 358 us MUL_MAT_ID(ffn_moe_down-86): 208 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 388 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 396 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 371 us MUL_MAT_ID(ffn_moe_down-89): 206 us ggml_barrier(...): 361 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 374 us MUL_MAT_ID(ffn_moe_down-90): 244 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 399 us MUL_MAT_ID(ffn_moe_down-91): 537 us ggml_barrier(...): 42 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 413 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 459 us MUL_MAT_ID(ffn_moe_down-26): 302 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 499 us MUL_MAT_ID(ffn_moe_down-27): 203 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 380 us MUL_MAT_ID(ffn_moe_down-28): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 363 us MUL_MAT_ID(ffn_moe_down-29): 214 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 385 us MUL_MAT_ID(ffn_moe_down-30): 205 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 360 us MUL_MAT_ID(ffn_moe_down-31): 376 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 365 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 393 us MUL_MAT_ID(ffn_moe_down-33): 235 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 400 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 389 us MUL_MAT_ID(ffn_moe_down-35): 202 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 375 us MUL_MAT_ID(ffn_moe_down-36): 384 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 384 us MUL_MAT_ID(ffn_moe_down-37): 235 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 352 us MUL_MAT_ID(ffn_moe_down-38): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 345 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 384 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 364 us MUL_MAT_ID(ffn_moe_down-41): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 369 us MUL_MAT_ID(ffn_moe_down-42): 248 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 408 us MUL_MAT_ID(ffn_moe_down-43): 201 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 359 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 363 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 355 us MUL_MAT_ID(ffn_moe_down-46): 182 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 393 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 398 us MUL_MAT_ID(ffn_moe_down-48): 618 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 372 us MUL_MAT_ID(ffn_moe_down-49): 217 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 411 us MUL_MAT_ID(ffn_moe_down-50): 226 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 392 us MUL_MAT_ID(ffn_moe_down-51): 193 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 421 us MUL_MAT_ID(ffn_moe_down-52): 304 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 381 us MUL_MAT_ID(ffn_moe_down-53): 216 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 408 us MUL_MAT_ID(ffn_moe_down-54): 205 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 402 us MUL_MAT_ID(ffn_moe_down-55): 331 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 362 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 392 us MUL_MAT_ID(ffn_moe_down-57): 198 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 388 us MUL_MAT_ID(ffn_moe_down-58): 255 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 381 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 403 us MUL_MAT_ID(ffn_moe_down-60): 191 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 373 us MUL_MAT_ID(ffn_moe_down-61): 187 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 395 us MUL_MAT_ID(ffn_moe_down-62): 235 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 373 us MUL_MAT_ID(ffn_moe_down-63): 284 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 364 us MUL_MAT_ID(ffn_moe_down-64): 175 us ggml_barrier(...): 334 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 368 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 379 us MUL_MAT_ID(ffn_moe_down-66): 196 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 401 us MUL_MAT_ID(ffn_moe_down-67): 181 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 420 us MUL_MAT_ID(ffn_moe_down-68): 192 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 408 us MUL_MAT_ID(ffn_moe_down-69): 212 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 372 us MUL_MAT_ID(ffn_moe_down-70): 188 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 385 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 388 us MUL_MAT_ID(ffn_moe_down-72): 193 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 370 us MUL_MAT_ID(ffn_moe_down-73): 189 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 399 us MUL_MAT_ID(ffn_moe_down-74): 208 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 361 us MUL_MAT_ID(ffn_moe_down-75): 186 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 395 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 402 us MUL_MAT_ID(ffn_moe_down-77): 189 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 365 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 376 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 375 us MUL_MAT_ID(ffn_moe_down-80): 185 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 396 us MUL_MAT_ID(ffn_moe_down-81): 215 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 368 us MUL_MAT_ID(ffn_moe_down-82): 490 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 392 us MUL_MAT_ID(ffn_moe_down-83): 202 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 352 us MUL_MAT_ID(ffn_moe_down-84): 246 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 386 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 417 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 404 us MUL_MAT_ID(ffn_moe_down-87): 217 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 433 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 365 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 391 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 366 us MUL_MAT_ID(ffn_moe_down-91): 427 us ggml_barrier(...): 56 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 355 us MUL_MAT_ID(ffn_moe_down-25): 473 us ggml_barrier(...): 196 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 365 us MUL_MAT_ID(ffn_moe_down-26): 229 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 346 us MUL_MAT_ID(ffn_moe_down-27): 369 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 370 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 409 us MUL_MAT_ID(ffn_moe_down-29): 220 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 372 us MUL_MAT_ID(ffn_moe_down-30): 189 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 391 us MUL_MAT_ID(ffn_moe_down-31): 201 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 390 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 367 us MUL_MAT_ID(ffn_moe_down-33): 209 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 382 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 350 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 357 us MUL_MAT_ID(ffn_moe_down-36): 211 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 360 us MUL_MAT_ID(ffn_moe_down-37): 222 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 354 us MUL_MAT_ID(ffn_moe_down-38): 192 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 391 us MUL_MAT_ID(ffn_moe_down-39): 223 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 350 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 366 us MUL_MAT_ID(ffn_moe_down-41): 369 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 404 us MUL_MAT_ID(ffn_moe_down-42): 230 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 389 us MUL_MAT_ID(ffn_moe_down-43): 200 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 376 us MUL_MAT_ID(ffn_moe_down-44): 304 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 385 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 384 us MUL_MAT_ID(ffn_moe_down-46): 210 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 374 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 402 us MUL_MAT_ID(ffn_moe_down-48): 546 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 383 us MUL_MAT_ID(ffn_moe_down-49): 225 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 192 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 370 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 115 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 369 us MUL_MAT_ID(ffn_moe_down-52): 358 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 366 us MUL_MAT_ID(ffn_moe_down-53): 187 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 373 us MUL_MAT_ID(ffn_moe_down-54): 207 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 368 us MUL_MAT_ID(ffn_moe_down-55): 181 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 370 us MUL_MAT_ID(ffn_moe_down-56): 344 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 368 us MUL_MAT_ID(ffn_moe_down-57): 188 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 380 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 393 us MUL_MAT_ID(ffn_moe_down-59): 250 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 355 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 390 us MUL_MAT_ID(ffn_moe_down-61): 196 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 358 us MUL_MAT_ID(ffn_moe_down-62): 204 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 367 us MUL_MAT_ID(ffn_moe_down-63): 187 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 403 us MUL_MAT_ID(ffn_moe_down-64): 333 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 372 us MUL_MAT_ID(ffn_moe_down-65): 217 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 392 us MUL_MAT_ID(ffn_moe_down-66): 203 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 386 us MUL_MAT_ID(ffn_moe_down-67): 186 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 369 us MUL_MAT_ID(ffn_moe_down-68): 422 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 337 us MUL_MAT_ID(ffn_moe_down-69): 191 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 359 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 421 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 403 us MUL_MAT_ID(ffn_moe_down-72): 404 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 348 us MUL_MAT_ID(ffn_moe_down-73): 203 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 470 us MUL_MAT_ID(ffn_moe_down-74): 214 us ggml_barrier(...): 200 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 382 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 410 us MUL_MAT_ID(ffn_moe_down-76): 198 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 382 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 385 us MUL_MAT_ID(ffn_moe_down-78): 222 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 393 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 375 us MUL_MAT_ID(ffn_moe_down-80): 342 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 349 us MUL_MAT_ID(ffn_moe_down-81): 202 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 375 us MUL_MAT_ID(ffn_moe_down-82): 208 us ggml_barrier(...): 297 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 391 us MUL_MAT_ID(ffn_moe_down-83): 253 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 372 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 351 us MUL_MAT_ID(ffn_moe_down-85): 228 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 348 us MUL_MAT_ID(ffn_moe_down-86): 197 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 403 us MUL_MAT_ID(ffn_moe_down-87): 207 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 388 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 359 us MUL_MAT_ID(ffn_moe_down-89): 349 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 347 us MUL_MAT_ID(ffn_moe_down-90): 286 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 351 us MUL_MAT_ID(ffn_moe_down-91): 241 us ggml_barrier(...): 60 us GET_ROWS(inp_embd): 24 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 495 us MUL_MAT_ID(ffn_moe_down-25): 193 us ggml_barrier(...): 440 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 410 us MUL_MAT_ID(ffn_moe_down-26): 351 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 394 us MUL_MAT_ID(ffn_moe_down-27): 204 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 184 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 417 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 422 us MUL_MAT_ID(ffn_moe_down-30): 295 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 374 us MUL_MAT_ID(ffn_moe_down-31): 400 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 457 us MUL_MAT_ID(ffn_moe_down-32): 484 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 411 us MUL_MAT_ID(ffn_moe_down-33): 222 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 405 us MUL_MAT_ID(ffn_moe_down-34): 261 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 346 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 396 us MUL_MAT_ID(ffn_moe_down-36): 229 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 385 us MUL_MAT_ID(ffn_moe_down-37): 242 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 371 us MUL_MAT_ID(ffn_moe_down-39): 267 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 380 us MUL_MAT_ID(ffn_moe_down-40): 248 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 395 us MUL_MAT_ID(ffn_moe_down-41): 211 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 410 us MUL_MAT_ID(ffn_moe_down-42): 205 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 424 us MUL_MAT_ID(ffn_moe_down-43): 192 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 370 us MUL_MAT_ID(ffn_moe_down-44): 198 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 382 us MUL_MAT_ID(ffn_moe_down-45): 199 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 368 us MUL_MAT_ID(ffn_moe_down-46): 185 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 401 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 385 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 404 us MUL_MAT_ID(ffn_moe_down-49): 210 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 383 us MUL_MAT_ID(ffn_moe_down-50): 192 us ggml_barrier(...): 129 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 371 us MUL_MAT_ID(ffn_moe_down-51): 194 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 366 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 111 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 384 us MUL_MAT_ID(ffn_moe_down-53): 303 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 354 us MUL_MAT_ID(ffn_moe_down-54): 179 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 405 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 409 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 366 us MUL_MAT_ID(ffn_moe_down-57): 510 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 381 us MUL_MAT_ID(ffn_moe_down-58): 331 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 367 us MUL_MAT_ID(ffn_moe_down-59): 188 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 355 us MUL_MAT_ID(ffn_moe_down-60): 198 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 377 us MUL_MAT_ID(ffn_moe_down-61): 181 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 394 us MUL_MAT_ID(ffn_moe_down-62): 365 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 377 us MUL_MAT_ID(ffn_moe_down-63): 187 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 381 us MUL_MAT_ID(ffn_moe_down-64): 500 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 410 us MUL_MAT_ID(ffn_moe_down-65): 221 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 363 us MUL_MAT_ID(ffn_moe_down-66): 208 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 439 us MUL_MAT_ID(ffn_moe_down-67): 205 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 380 us MUL_MAT_ID(ffn_moe_down-68): 188 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 377 us MUL_MAT_ID(ffn_moe_down-69): 210 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 407 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 391 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 357 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 385 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 378 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 391 us MUL_MAT_ID(ffn_moe_down-75): 347 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 426 us MUL_MAT_ID(ffn_moe_down-76): 188 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 381 us MUL_MAT_ID(ffn_moe_down-77): 350 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 398 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 369 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 376 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 354 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 368 us MUL_MAT_ID(ffn_moe_down-82): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 398 us MUL_MAT_ID(ffn_moe_down-83): 221 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 421 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 411 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 398 us MUL_MAT_ID(ffn_moe_down-86): 208 us ggml_barrier(...): 389 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 369 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 372 us MUL_MAT_ID(ffn_moe_down-88): 218 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 375 us MUL_MAT_ID(ffn_moe_down-89): 510 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 419 us MUL_MAT_ID(ffn_moe_down-90): 649 us ggml_barrier(...): 293 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 418 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 57 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 408 us MUL_MAT_ID(ffn_moe_down-25): 209 us ggml_barrier(...): 125 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 414 us MUL_MAT_ID(ffn_moe_down-26): 233 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 378 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 360 us MUL_MAT_ID(ffn_moe_down-28): 209 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 409 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 381 us MUL_MAT_ID(ffn_moe_down-30): 182 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 383 us MUL_MAT_ID(ffn_moe_down-31): 425 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 395 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 376 us MUL_MAT_ID(ffn_moe_down-33): 554 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 355 us MUL_MAT_ID(ffn_moe_down-34): 207 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 400 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 392 us MUL_MAT_ID(ffn_moe_down-36): 227 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 365 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 371 us MUL_MAT_ID(ffn_moe_down-38): 229 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 372 us MUL_MAT_ID(ffn_moe_down-39): 689 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 393 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 401 us MUL_MAT_ID(ffn_moe_down-41): 192 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 354 us MUL_MAT_ID(ffn_moe_down-42): 192 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 353 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 364 us MUL_MAT_ID(ffn_moe_down-44): 231 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 385 us MUL_MAT_ID(ffn_moe_down-45): 200 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 357 us MUL_MAT_ID(ffn_moe_down-46): 237 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 335 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 354 us MUL_MAT_ID(ffn_moe_down-48): 185 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 388 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 398 us MUL_MAT_ID(ffn_moe_down-50): 193 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 353 us MUL_MAT_ID(ffn_moe_down-51): 206 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 382 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 356 us MUL_MAT_ID(ffn_moe_down-53): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 341 us MUL_MAT_ID(ffn_moe_down-54): 230 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 402 us MUL_MAT_ID(ffn_moe_down-55): 358 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 403 us MUL_MAT_ID(ffn_moe_down-56): 185 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 374 us MUL_MAT_ID(ffn_moe_down-57): 202 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 372 us MUL_MAT_ID(ffn_moe_down-58): 292 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 382 us MUL_MAT_ID(ffn_moe_down-59): 245 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 391 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 372 us MUL_MAT_ID(ffn_moe_down-61): 500 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 369 us MUL_MAT_ID(ffn_moe_down-62): 342 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 355 us MUL_MAT_ID(ffn_moe_down-63): 216 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 387 us MUL_MAT_ID(ffn_moe_down-64): 194 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 386 us MUL_MAT_ID(ffn_moe_down-65): 189 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 375 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 469 us MUL_MAT_ID(ffn_moe_down-67): 305 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 389 us MUL_MAT_ID(ffn_moe_down-68): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 375 us MUL_MAT_ID(ffn_moe_down-69): 169 us ggml_barrier(...): 322 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 374 us MUL_MAT_ID(ffn_moe_down-70): 188 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 398 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 438 us MUL_MAT_ID(ffn_moe_down-72): 166 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 398 us MUL_MAT_ID(ffn_moe_down-73): 431 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 414 us MUL_MAT_ID(ffn_moe_down-74): 233 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 359 us MUL_MAT_ID(ffn_moe_down-75): 300 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 350 us MUL_MAT_ID(ffn_moe_down-76): 207 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 376 us MUL_MAT_ID(ffn_moe_down-77): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 496 us MUL_MAT_ID(ffn_moe_down-78): 212 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 409 us MUL_MAT_ID(ffn_moe_down-79): 249 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 398 us MUL_MAT_ID(ffn_moe_down-80): 207 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 403 us MUL_MAT_ID(ffn_moe_down-81): 229 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 396 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 405 us MUL_MAT_ID(ffn_moe_down-83): 212 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 391 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 409 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 398 us MUL_MAT_ID(ffn_moe_down-86): 206 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 351 us MUL_MAT_ID(ffn_moe_down-87): 325 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 353 us MUL_MAT_ID(ffn_moe_down-88): 369 us ggml_barrier(...): 296 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 375 us MUL_MAT_ID(ffn_moe_down-89): 189 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 389 us MUL_MAT_ID(ffn_moe_down-90): 670 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 394 us MUL_MAT_ID(ffn_moe_down-91): 333 us ggml_barrier(...): 28 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 368 us MUL_MAT_ID(ffn_moe_down-25): 179 us ggml_barrier(...): 505 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 408 us MUL_MAT_ID(ffn_moe_down-26): 274 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 387 us MUL_MAT_ID(ffn_moe_down-27): 212 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 367 us MUL_MAT_ID(ffn_moe_down-28): 178 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 371 us MUL_MAT_ID(ffn_moe_down-29): 227 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 375 us MUL_MAT_ID(ffn_moe_down-30): 186 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 388 us MUL_MAT_ID(ffn_moe_down-31): 515 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 401 us MUL_MAT_ID(ffn_moe_down-32): 366 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 402 us MUL_MAT_ID(ffn_moe_down-33): 190 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 456 us MUL_MAT_ID(ffn_moe_down-34): 223 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 427 us MUL_MAT_ID(ffn_moe_down-35): 371 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 357 us MUL_MAT_ID(ffn_moe_down-36): 385 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 372 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 377 us MUL_MAT_ID(ffn_moe_down-38): 191 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 386 us MUL_MAT_ID(ffn_moe_down-39): 758 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 362 us MUL_MAT_ID(ffn_moe_down-40): 228 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 402 us MUL_MAT_ID(ffn_moe_down-41): 215 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 341 us MUL_MAT_ID(ffn_moe_down-42): 171 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 388 us MUL_MAT_ID(ffn_moe_down-43): 186 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 371 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 395 us MUL_MAT_ID(ffn_moe_down-45): 252 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 401 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 369 us MUL_MAT_ID(ffn_moe_down-47): 204 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 399 us MUL_MAT_ID(ffn_moe_down-48): 211 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 351 us MUL_MAT_ID(ffn_moe_down-49): 166 us ggml_barrier(...): 299 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 396 us MUL_MAT_ID(ffn_moe_down-50): 180 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 354 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 401 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 391 us MUL_MAT_ID(ffn_moe_down-53): 421 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 376 us MUL_MAT_ID(ffn_moe_down-54): 368 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 213 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 395 us MUL_MAT_ID(ffn_moe_down-56): 191 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 387 us MUL_MAT_ID(ffn_moe_down-57): 204 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 404 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 401 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 397 us MUL_MAT_ID(ffn_moe_down-60): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 399 us MUL_MAT_ID(ffn_moe_down-61): 200 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 401 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 402 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 407 us MUL_MAT_ID(ffn_moe_down-64): 217 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 377 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 404 us MUL_MAT_ID(ffn_moe_down-66): 386 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 407 us MUL_MAT_ID(ffn_moe_down-67): 224 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 364 us MUL_MAT_ID(ffn_moe_down-68): 204 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 409 us MUL_MAT_ID(ffn_moe_down-69): 228 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 460 us MUL_MAT_ID(ffn_moe_down-70): 161 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 408 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 385 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 393 us MUL_MAT_ID(ffn_moe_down-73): 251 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 346 us MUL_MAT_ID(ffn_moe_down-74): 210 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 358 us MUL_MAT_ID(ffn_moe_down-75): 205 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 359 us MUL_MAT_ID(ffn_moe_down-76): 181 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 424 us MUL_MAT_ID(ffn_moe_down-77): 204 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 393 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 479 us MUL_MAT_ID(ffn_moe_down-79): 325 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 405 us MUL_MAT_ID(ffn_moe_down-80): 379 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 366 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 406 us MUL_MAT_ID(ffn_moe_down-82): 213 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 398 us MUL_MAT_ID(ffn_moe_down-83): 200 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 393 us MUL_MAT_ID(ffn_moe_down-84): 329 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 393 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 365 us MUL_MAT_ID(ffn_moe_down-86): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 381 us MUL_MAT_ID(ffn_moe_down-87): 201 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 360 us MUL_MAT_ID(ffn_moe_down-88): 391 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 375 us MUL_MAT_ID(ffn_moe_down-89): 195 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 420 us MUL_MAT_ID(ffn_moe_down-90): 639 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 371 us MUL_MAT_ID(ffn_moe_down-91): 250 us ggml_barrier(...): 39 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 388 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 377 us MUL_MAT_ID(ffn_moe_down-26): 178 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 385 us MUL_MAT_ID(ffn_moe_down-27): 546 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 412 us MUL_MAT_ID(ffn_moe_down-28): 235 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 386 us MUL_MAT_ID(ffn_moe_down-29): 200 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 362 us MUL_MAT_ID(ffn_moe_down-30): 237 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 411 us MUL_MAT_ID(ffn_moe_down-31): 212 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 418 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 349 us MUL_MAT_ID(ffn_moe_down-33): 211 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 400 us MUL_MAT_ID(ffn_moe_down-34): 538 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 384 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 360 us MUL_MAT_ID(ffn_moe_down-36): 190 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 371 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 356 us MUL_MAT_ID(ffn_moe_down-38): 186 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 373 us MUL_MAT_ID(ffn_moe_down-39): 231 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 392 us MUL_MAT_ID(ffn_moe_down-40): 233 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 367 us MUL_MAT_ID(ffn_moe_down-41): 339 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 407 us MUL_MAT_ID(ffn_moe_down-42): 223 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 376 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 402 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 388 us MUL_MAT_ID(ffn_moe_down-45): 404 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 394 us MUL_MAT_ID(ffn_moe_down-46): 277 us ggml_barrier(...): 168 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 400 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 384 us MUL_MAT_ID(ffn_moe_down-48): 214 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 211 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 387 us MUL_MAT_ID(ffn_moe_down-50): 512 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 403 us MUL_MAT_ID(ffn_moe_down-51): 208 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 388 us MUL_MAT_ID(ffn_moe_down-52): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 370 us MUL_MAT_ID(ffn_moe_down-53): 191 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 383 us MUL_MAT_ID(ffn_moe_down-54): 206 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 403 us MUL_MAT_ID(ffn_moe_down-55): 183 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 356 us MUL_MAT_ID(ffn_moe_down-56): 190 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 403 us MUL_MAT_ID(ffn_moe_down-57): 213 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 394 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 361 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 379 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 391 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 370 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 362 us MUL_MAT_ID(ffn_moe_down-63): 206 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 383 us MUL_MAT_ID(ffn_moe_down-64): 226 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 382 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 360 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 382 us MUL_MAT_ID(ffn_moe_down-67): 196 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 410 us MUL_MAT_ID(ffn_moe_down-68): 196 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 375 us MUL_MAT_ID(ffn_moe_down-69): 203 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 398 us MUL_MAT_ID(ffn_moe_down-70): 230 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 388 us MUL_MAT_ID(ffn_moe_down-71): 203 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 364 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 377 us MUL_MAT_ID(ffn_moe_down-73): 201 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 376 us MUL_MAT_ID(ffn_moe_down-74): 217 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 385 us MUL_MAT_ID(ffn_moe_down-75): 310 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 400 us MUL_MAT_ID(ffn_moe_down-76): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 398 us MUL_MAT_ID(ffn_moe_down-77): 207 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 394 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 393 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 377 us MUL_MAT_ID(ffn_moe_down-80): 361 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 380 us MUL_MAT_ID(ffn_moe_down-81): 219 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 378 us MUL_MAT_ID(ffn_moe_down-82): 198 us ggml_barrier(...): 343 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 385 us MUL_MAT_ID(ffn_moe_down-83): 191 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 430 us MUL_MAT_ID(ffn_moe_down-84): 251 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 361 us MUL_MAT_ID(ffn_moe_down-85): 249 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 376 us MUL_MAT_ID(ffn_moe_down-86): 213 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 374 us MUL_MAT_ID(ffn_moe_down-87): 221 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 395 us MUL_MAT_ID(ffn_moe_down-88): 250 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 375 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 362 us MUL_MAT_ID(ffn_moe_down-90): 222 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 404 us MUL_MAT_ID(ffn_moe_down-91): 546 us ggml_barrier(...): 24 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 403 us MUL_MAT_ID(ffn_moe_down-25): 193 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 383 us MUL_MAT_ID(ffn_moe_down-26): 181 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 355 us MUL_MAT_ID(ffn_moe_down-27): 187 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 361 us MUL_MAT_ID(ffn_moe_down-28): 386 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 423 us MUL_MAT_ID(ffn_moe_down-29): 213 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 418 us MUL_MAT_ID(ffn_moe_down-30): 191 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 402 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 410 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 368 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 476 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 373 us MUL_MAT_ID(ffn_moe_down-35): 302 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 369 us MUL_MAT_ID(ffn_moe_down-36): 185 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 421 us MUL_MAT_ID(ffn_moe_down-37): 225 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 408 us MUL_MAT_ID(ffn_moe_down-38): 204 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 401 us MUL_MAT_ID(ffn_moe_down-39): 245 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 390 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 375 us MUL_MAT_ID(ffn_moe_down-41): 312 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 389 us MUL_MAT_ID(ffn_moe_down-42): 193 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 405 us MUL_MAT_ID(ffn_moe_down-43): 212 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 392 us MUL_MAT_ID(ffn_moe_down-44): 344 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 374 us MUL_MAT_ID(ffn_moe_down-45): 206 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 353 us MUL_MAT_ID(ffn_moe_down-46): 528 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 363 us MUL_MAT_ID(ffn_moe_down-47): 184 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 380 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 369 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 363 us MUL_MAT_ID(ffn_moe_down-50): 212 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 383 us MUL_MAT_ID(ffn_moe_down-51): 229 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 355 us MUL_MAT_ID(ffn_moe_down-52): 179 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 349 us MUL_MAT_ID(ffn_moe_down-53): 208 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 379 us MUL_MAT_ID(ffn_moe_down-54): 490 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 405 us MUL_MAT_ID(ffn_moe_down-56): 305 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 370 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 409 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 495 us MUL_MAT_ID(ffn_moe_down-59): 221 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 410 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 351 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 378 us MUL_MAT_ID(ffn_moe_down-62): 184 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 388 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 417 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 378 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 386 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 373 us MUL_MAT_ID(ffn_moe_down-67): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 355 us MUL_MAT_ID(ffn_moe_down-68): 528 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 403 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 418 us MUL_MAT_ID(ffn_moe_down-70): 183 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 385 us MUL_MAT_ID(ffn_moe_down-71): 239 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 369 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 368 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 354 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 376 us MUL_MAT_ID(ffn_moe_down-75): 191 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 401 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 380 us MUL_MAT_ID(ffn_moe_down-77): 187 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 372 us MUL_MAT_ID(ffn_moe_down-78): 263 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 534 us MUL_MAT_ID(ffn_moe_down-79): 168 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 383 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 360 us MUL_MAT_ID(ffn_moe_down-81): 186 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 375 us MUL_MAT_ID(ffn_moe_down-82): 223 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 427 us MUL_MAT_ID(ffn_moe_down-83): 537 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 348 us MUL_MAT_ID(ffn_moe_down-84): 232 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 355 us MUL_MAT_ID(ffn_moe_down-85): 247 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 385 us MUL_MAT_ID(ffn_moe_down-86): 534 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 369 us MUL_MAT_ID(ffn_moe_down-87): 207 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 394 us MUL_MAT_ID(ffn_moe_down-88): 202 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 361 us MUL_MAT_ID(ffn_moe_down-89): 208 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 375 us MUL_MAT_ID(ffn_moe_down-90): 222 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 348 us MUL_MAT_ID(ffn_moe_down-91): 226 us ggml_barrier(...): 67 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 403 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 390 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 376 us MUL_MAT_ID(ffn_moe_down-27): 380 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 367 us MUL_MAT_ID(ffn_moe_down-28): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 370 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 375 us MUL_MAT_ID(ffn_moe_down-30): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 392 us MUL_MAT_ID(ffn_moe_down-31): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 393 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 366 us MUL_MAT_ID(ffn_moe_down-33): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 401 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 403 us MUL_MAT_ID(ffn_moe_down-35): 186 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 367 us MUL_MAT_ID(ffn_moe_down-36): 357 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 394 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 381 us MUL_MAT_ID(ffn_moe_down-38): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 361 us MUL_MAT_ID(ffn_moe_down-39): 241 us ggml_barrier(...): 470 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 500 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 432 us MUL_MAT_ID(ffn_moe_down-41): 213 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 336 us MUL_MAT_ID(ffn_moe_down-42): 358 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 406 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 360 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 379 us MUL_MAT_ID(ffn_moe_down-45): 404 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 361 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 368 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 354 us MUL_MAT_ID(ffn_moe_down-48): 202 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 379 us MUL_MAT_ID(ffn_moe_down-49): 187 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 380 us MUL_MAT_ID(ffn_moe_down-50): 326 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 402 us MUL_MAT_ID(ffn_moe_down-51): 208 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 366 us MUL_MAT_ID(ffn_moe_down-52): 195 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 370 us MUL_MAT_ID(ffn_moe_down-53): 485 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 376 us MUL_MAT_ID(ffn_moe_down-54): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 370 us MUL_MAT_ID(ffn_moe_down-55): 186 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 400 us MUL_MAT_ID(ffn_moe_down-56): 235 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 349 us MUL_MAT_ID(ffn_moe_down-57): 201 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 391 us MUL_MAT_ID(ffn_moe_down-58): 187 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 349 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 407 us MUL_MAT_ID(ffn_moe_down-60): 368 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 464 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 375 us MUL_MAT_ID(ffn_moe_down-62): 381 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 334 us MUL_MAT_ID(ffn_moe_down-63): 392 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 373 us MUL_MAT_ID(ffn_moe_down-64): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 370 us MUL_MAT_ID(ffn_moe_down-65): 203 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 386 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 385 us MUL_MAT_ID(ffn_moe_down-67): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 382 us MUL_MAT_ID(ffn_moe_down-68): 183 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 347 us MUL_MAT_ID(ffn_moe_down-69): 183 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 365 us MUL_MAT_ID(ffn_moe_down-70): 393 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 372 us MUL_MAT_ID(ffn_moe_down-71): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 400 us MUL_MAT_ID(ffn_moe_down-72): 187 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 380 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 381 us MUL_MAT_ID(ffn_moe_down-74): 204 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 367 us MUL_MAT_ID(ffn_moe_down-75): 236 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 447 us MUL_MAT_ID(ffn_moe_down-76): 278 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 356 us MUL_MAT_ID(ffn_moe_down-77): 246 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 398 us MUL_MAT_ID(ffn_moe_down-78): 371 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 379 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 383 us MUL_MAT_ID(ffn_moe_down-80): 214 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 388 us MUL_MAT_ID(ffn_moe_down-81): 372 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 362 us MUL_MAT_ID(ffn_moe_down-82): 402 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 405 us MUL_MAT_ID(ffn_moe_down-83): 222 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 385 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 389 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 416 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 392 us MUL_MAT_ID(ffn_moe_down-87): 226 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 368 us MUL_MAT_ID(ffn_moe_down-88): 435 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 411 us MUL_MAT_ID(ffn_moe_down-89): 188 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 376 us MUL_MAT_ID(ffn_moe_down-90): 238 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 379 us MUL_MAT_ID(ffn_moe_down-91): 228 us ggml_barrier(...): 69 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 416 us MUL_MAT_ID(ffn_moe_down-25): 326 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 401 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 366 us MUL_MAT_ID(ffn_moe_down-27): 191 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 402 us MUL_MAT_ID(ffn_moe_down-28): 369 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 374 us MUL_MAT_ID(ffn_moe_down-29): 163 us ggml_barrier(...): 301 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 405 us MUL_MAT_ID(ffn_moe_down-30): 199 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 399 us MUL_MAT_ID(ffn_moe_down-31): 377 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 398 us MUL_MAT_ID(ffn_moe_down-32): 309 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 192 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 381 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 420 us MUL_MAT_ID(ffn_moe_down-35): 202 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 388 us MUL_MAT_ID(ffn_moe_down-36): 216 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 419 us MUL_MAT_ID(ffn_moe_down-37): 251 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 372 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 398 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 381 us MUL_MAT_ID(ffn_moe_down-40): 223 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 377 us MUL_MAT_ID(ffn_moe_down-41): 160 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 469 us MUL_MAT_ID(ffn_moe_down-42): 172 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 377 us MUL_MAT_ID(ffn_moe_down-43): 301 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 376 us MUL_MAT_ID(ffn_moe_down-44): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 490 us MUL_MAT_ID(ffn_moe_down-45): 171 us ggml_barrier(...): 198 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 408 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 372 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 358 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 481 us MUL_MAT_ID(ffn_moe_down-49): 240 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 388 us MUL_MAT_ID(ffn_moe_down-50): 245 us ggml_barrier(...): 220 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 391 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 359 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 389 us MUL_MAT_ID(ffn_moe_down-53): 195 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 391 us MUL_MAT_ID(ffn_moe_down-54): 199 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 377 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 352 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 353 us MUL_MAT_ID(ffn_moe_down-57): 185 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 386 us MUL_MAT_ID(ffn_moe_down-58): 202 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 350 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 372 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 393 us MUL_MAT_ID(ffn_moe_down-61): 170 us ggml_barrier(...): 265 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 398 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 413 us MUL_MAT_ID(ffn_moe_down-63): 204 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 398 us MUL_MAT_ID(ffn_moe_down-64): 204 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 397 us MUL_MAT_ID(ffn_moe_down-65): 198 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 399 us MUL_MAT_ID(ffn_moe_down-66): 217 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 377 us MUL_MAT_ID(ffn_moe_down-67): 344 us ggml_barrier(...): 286 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 371 us MUL_MAT_ID(ffn_moe_down-68): 177 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 358 us MUL_MAT_ID(ffn_moe_down-69): 184 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 380 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 362 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 349 us MUL_MAT_ID(ffn_moe_down-72): 236 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 390 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 348 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 372 us MUL_MAT_ID(ffn_moe_down-75): 186 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 355 us MUL_MAT_ID(ffn_moe_down-76): 215 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 406 us MUL_MAT_ID(ffn_moe_down-77): 226 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 398 us MUL_MAT_ID(ffn_moe_down-78): 178 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 358 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 353 us MUL_MAT_ID(ffn_moe_down-80): 493 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 393 us MUL_MAT_ID(ffn_moe_down-81): 191 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 489 us MUL_MAT_ID(ffn_moe_down-82): 180 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 406 us MUL_MAT_ID(ffn_moe_down-83): 187 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 344 us MUL_MAT_ID(ffn_moe_down-84): 239 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 391 us MUL_MAT_ID(ffn_moe_down-85): 549 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 392 us MUL_MAT_ID(ffn_moe_down-86): 212 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 413 us MUL_MAT_ID(ffn_moe_down-87): 189 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 402 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 400 us MUL_MAT_ID(ffn_moe_down-89): 207 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 372 us MUL_MAT_ID(ffn_moe_down-90): 664 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 381 us MUL_MAT_ID(ffn_moe_down-91): 227 us ggml_barrier(...): 35 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 379 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 398 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 399 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 354 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 394 us MUL_MAT_ID(ffn_moe_down-29): 190 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 360 us MUL_MAT_ID(ffn_moe_down-30): 215 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 377 us MUL_MAT_ID(ffn_moe_down-31): 207 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 391 us MUL_MAT_ID(ffn_moe_down-32): 181 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 412 us MUL_MAT_ID(ffn_moe_down-33): 385 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 383 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 388 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 370 us MUL_MAT_ID(ffn_moe_down-36): 381 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 377 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 387 us MUL_MAT_ID(ffn_moe_down-38): 184 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 415 us MUL_MAT_ID(ffn_moe_down-39): 242 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 352 us MUL_MAT_ID(ffn_moe_down-40): 245 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 389 us MUL_MAT_ID(ffn_moe_down-41): 212 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 407 us MUL_MAT_ID(ffn_moe_down-42): 190 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 375 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 384 us MUL_MAT_ID(ffn_moe_down-44): 422 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 375 us MUL_MAT_ID(ffn_moe_down-45): 333 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 369 us MUL_MAT_ID(ffn_moe_down-46): 195 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 419 us MUL_MAT_ID(ffn_moe_down-47): 222 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 354 us MUL_MAT_ID(ffn_moe_down-48): 224 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 400 us MUL_MAT_ID(ffn_moe_down-49): 192 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 386 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 400 us MUL_MAT_ID(ffn_moe_down-51): 186 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 407 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 355 us MUL_MAT_ID(ffn_moe_down-53): 211 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 383 us MUL_MAT_ID(ffn_moe_down-54): 189 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 373 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 361 us MUL_MAT_ID(ffn_moe_down-56): 184 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 407 us MUL_MAT_ID(ffn_moe_down-57): 237 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 399 us MUL_MAT_ID(ffn_moe_down-58): 214 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 349 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 116 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 378 us MUL_MAT_ID(ffn_moe_down-60): 265 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 355 us MUL_MAT_ID(ffn_moe_down-61): 412 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 401 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 385 us MUL_MAT_ID(ffn_moe_down-63): 390 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 387 us MUL_MAT_ID(ffn_moe_down-64): 180 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 394 us MUL_MAT_ID(ffn_moe_down-65): 256 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 508 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 396 us MUL_MAT_ID(ffn_moe_down-67): 204 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 409 us MUL_MAT_ID(ffn_moe_down-68): 385 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 356 us MUL_MAT_ID(ffn_moe_down-69): 201 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 381 us MUL_MAT_ID(ffn_moe_down-70): 538 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 418 us MUL_MAT_ID(ffn_moe_down-71): 212 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 364 us MUL_MAT_ID(ffn_moe_down-72): 211 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 226 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 408 us MUL_MAT_ID(ffn_moe_down-74): 195 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 376 us MUL_MAT_ID(ffn_moe_down-75): 182 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 375 us MUL_MAT_ID(ffn_moe_down-76): 199 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 379 us MUL_MAT_ID(ffn_moe_down-77): 199 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 413 us MUL_MAT_ID(ffn_moe_down-78): 383 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 389 us MUL_MAT_ID(ffn_moe_down-79): 180 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 377 us MUL_MAT_ID(ffn_moe_down-80): 200 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 435 us MUL_MAT_ID(ffn_moe_down-81): 341 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 341 us MUL_MAT_ID(ffn_moe_down-82): 190 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 403 us MUL_MAT_ID(ffn_moe_down-83): 329 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 366 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 394 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 390 us MUL_MAT_ID(ffn_moe_down-86): 228 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 428 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 386 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 369 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 401 us MUL_MAT_ID(ffn_moe_down-90): 244 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 368 us MUL_MAT_ID(ffn_moe_down-91): 231 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 411 us MUL_MAT_ID(ffn_moe_down-25): 181 us ggml_barrier(...): 149 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 362 us MUL_MAT_ID(ffn_moe_down-26): 185 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 380 us MUL_MAT_ID(ffn_moe_down-27): 188 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 369 us MUL_MAT_ID(ffn_moe_down-28): 219 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 367 us MUL_MAT_ID(ffn_moe_down-29): 192 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 409 us MUL_MAT_ID(ffn_moe_down-30): 192 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 410 us MUL_MAT_ID(ffn_moe_down-31): 197 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 376 us MUL_MAT_ID(ffn_moe_down-32): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 389 us MUL_MAT_ID(ffn_moe_down-33): 216 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 414 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 363 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 403 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 391 us MUL_MAT_ID(ffn_moe_down-37): 243 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 381 us MUL_MAT_ID(ffn_moe_down-38): 205 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 394 us MUL_MAT_ID(ffn_moe_down-39): 246 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 394 us MUL_MAT_ID(ffn_moe_down-40): 426 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 432 us MUL_MAT_ID(ffn_moe_down-41): 285 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 423 us MUL_MAT_ID(ffn_moe_down-42): 195 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 379 us MUL_MAT_ID(ffn_moe_down-43): 371 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 398 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 363 us MUL_MAT_ID(ffn_moe_down-45): 398 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 360 us MUL_MAT_ID(ffn_moe_down-46): 219 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 384 us MUL_MAT_ID(ffn_moe_down-47): 246 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 408 us MUL_MAT_ID(ffn_moe_down-48): 226 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 400 us MUL_MAT_ID(ffn_moe_down-49): 184 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 358 us MUL_MAT_ID(ffn_moe_down-50): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 391 us MUL_MAT_ID(ffn_moe_down-51): 201 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 368 us MUL_MAT_ID(ffn_moe_down-52): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 377 us MUL_MAT_ID(ffn_moe_down-53): 367 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 390 us MUL_MAT_ID(ffn_moe_down-54): 216 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 368 us MUL_MAT_ID(ffn_moe_down-55): 234 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 374 us MUL_MAT_ID(ffn_moe_down-56): 189 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 410 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 413 us MUL_MAT_ID(ffn_moe_down-58): 189 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 359 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 398 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 364 us MUL_MAT_ID(ffn_moe_down-61): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 365 us MUL_MAT_ID(ffn_moe_down-62): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 360 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 400 us MUL_MAT_ID(ffn_moe_down-64): 199 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 396 us MUL_MAT_ID(ffn_moe_down-65): 190 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 388 us MUL_MAT_ID(ffn_moe_down-66): 205 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 353 us MUL_MAT_ID(ffn_moe_down-67): 191 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 354 us MUL_MAT_ID(ffn_moe_down-68): 315 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 370 us MUL_MAT_ID(ffn_moe_down-69): 198 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 378 us MUL_MAT_ID(ffn_moe_down-70): 187 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 400 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 359 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 371 us MUL_MAT_ID(ffn_moe_down-73): 208 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 372 us MUL_MAT_ID(ffn_moe_down-74): 200 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 385 us MUL_MAT_ID(ffn_moe_down-75): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 387 us MUL_MAT_ID(ffn_moe_down-76): 371 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 388 us MUL_MAT_ID(ffn_moe_down-77): 197 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 402 us MUL_MAT_ID(ffn_moe_down-78): 330 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 361 us MUL_MAT_ID(ffn_moe_down-79): 187 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 349 us MUL_MAT_ID(ffn_moe_down-80): 194 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 363 us MUL_MAT_ID(ffn_moe_down-81): 219 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 364 us MUL_MAT_ID(ffn_moe_down-82): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 402 us MUL_MAT_ID(ffn_moe_down-83): 218 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 408 us MUL_MAT_ID(ffn_moe_down-84): 217 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 369 us MUL_MAT_ID(ffn_moe_down-85): 230 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 373 us MUL_MAT_ID(ffn_moe_down-86): 209 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 375 us MUL_MAT_ID(ffn_moe_down-87): 210 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 376 us MUL_MAT_ID(ffn_moe_down-88): 180 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 380 us MUL_MAT_ID(ffn_moe_down-89): 377 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 362 us MUL_MAT_ID(ffn_moe_down-90): 587 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 355 us MUL_MAT_ID(ffn_moe_down-91): 259 us ggml_barrier(...): 61 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 396 us MUL_MAT_ID(ffn_moe_down-25): 339 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 373 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 365 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 362 us MUL_MAT_ID(ffn_moe_down-28): 195 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 416 us MUL_MAT_ID(ffn_moe_down-29): 190 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 449 us MUL_MAT_ID(ffn_moe_down-30): 191 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 383 us MUL_MAT_ID(ffn_moe_down-31): 184 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 399 us MUL_MAT_ID(ffn_moe_down-32): 339 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 354 us MUL_MAT_ID(ffn_moe_down-33): 351 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 415 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 399 us MUL_MAT_ID(ffn_moe_down-35): 200 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 364 us MUL_MAT_ID(ffn_moe_down-36): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 425 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 398 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 363 us MUL_MAT_ID(ffn_moe_down-39): 247 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 347 us MUL_MAT_ID(ffn_moe_down-40): 253 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 388 us MUL_MAT_ID(ffn_moe_down-41): 227 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 390 us MUL_MAT_ID(ffn_moe_down-42): 247 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 385 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 386 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 378 us MUL_MAT_ID(ffn_moe_down-45): 383 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 406 us MUL_MAT_ID(ffn_moe_down-46): 205 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 370 us MUL_MAT_ID(ffn_moe_down-47): 186 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 342 us MUL_MAT_ID(ffn_moe_down-48): 208 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 404 us MUL_MAT_ID(ffn_moe_down-49): 519 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 394 us MUL_MAT_ID(ffn_moe_down-50): 204 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 420 us MUL_MAT_ID(ffn_moe_down-51): 203 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 387 us MUL_MAT_ID(ffn_moe_down-52): 369 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 385 us MUL_MAT_ID(ffn_moe_down-53): 406 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 391 us MUL_MAT_ID(ffn_moe_down-54): 182 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 371 us MUL_MAT_ID(ffn_moe_down-55): 186 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 393 us MUL_MAT_ID(ffn_moe_down-56): 222 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 394 us MUL_MAT_ID(ffn_moe_down-57): 208 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 425 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 238 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 412 us MUL_MAT_ID(ffn_moe_down-59): 209 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 358 us MUL_MAT_ID(ffn_moe_down-60): 196 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 401 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 386 us MUL_MAT_ID(ffn_moe_down-62): 400 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 376 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 386 us MUL_MAT_ID(ffn_moe_down-64): 189 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 389 us MUL_MAT_ID(ffn_moe_down-65): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 381 us MUL_MAT_ID(ffn_moe_down-66): 391 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 366 us MUL_MAT_ID(ffn_moe_down-67): 353 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 356 us MUL_MAT_ID(ffn_moe_down-68): 338 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 429 us MUL_MAT_ID(ffn_moe_down-69): 259 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 384 us MUL_MAT_ID(ffn_moe_down-70): 352 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 368 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 367 us MUL_MAT_ID(ffn_moe_down-72): 196 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 346 us MUL_MAT_ID(ffn_moe_down-73): 203 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 377 us MUL_MAT_ID(ffn_moe_down-74): 194 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 375 us MUL_MAT_ID(ffn_moe_down-75): 187 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 371 us MUL_MAT_ID(ffn_moe_down-76): 189 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 424 us MUL_MAT_ID(ffn_moe_down-77): 229 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 394 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 384 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 379 us MUL_MAT_ID(ffn_moe_down-80): 357 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 364 us MUL_MAT_ID(ffn_moe_down-81): 193 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 358 us MUL_MAT_ID(ffn_moe_down-82): 528 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 387 us MUL_MAT_ID(ffn_moe_down-83): 201 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 348 us MUL_MAT_ID(ffn_moe_down-84): 240 us ggml_barrier(...): 212 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 462 us MUL_MAT_ID(ffn_moe_down-85): 225 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 366 us MUL_MAT_ID(ffn_moe_down-86): 200 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 351 us MUL_MAT_ID(ffn_moe_down-87): 201 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 361 us MUL_MAT_ID(ffn_moe_down-88): 217 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 379 us MUL_MAT_ID(ffn_moe_down-89): 204 us ggml_barrier(...): 159 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 357 us MUL_MAT_ID(ffn_moe_down-90): 240 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 399 us MUL_MAT_ID(ffn_moe_down-91): 292 us ggml_barrier(...): 130 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 385 us MUL_MAT_ID(ffn_moe_down-25): 195 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 381 us MUL_MAT_ID(ffn_moe_down-26): 176 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 401 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 368 us MUL_MAT_ID(ffn_moe_down-28): 330 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 388 us MUL_MAT_ID(ffn_moe_down-29): 191 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 403 us MUL_MAT_ID(ffn_moe_down-30): 215 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 394 us MUL_MAT_ID(ffn_moe_down-31): 384 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 354 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 370 us MUL_MAT_ID(ffn_moe_down-33): 566 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 372 us MUL_MAT_ID(ffn_moe_down-34): 375 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 362 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 378 us MUL_MAT_ID(ffn_moe_down-36): 198 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 399 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 407 us MUL_MAT_ID(ffn_moe_down-38): 340 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 362 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 355 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 396 us MUL_MAT_ID(ffn_moe_down-41): 194 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 367 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 388 us MUL_MAT_ID(ffn_moe_down-43): 255 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 388 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 224 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 374 us MUL_MAT_ID(ffn_moe_down-45): 198 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 387 us MUL_MAT_ID(ffn_moe_down-46): 221 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 386 us MUL_MAT_ID(ffn_moe_down-47): 205 us ggml_barrier(...): 173 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 358 us MUL_MAT_ID(ffn_moe_down-48): 455 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 492 us MUL_MAT_ID(ffn_moe_down-49): 276 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 353 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 393 us MUL_MAT_ID(ffn_moe_down-51): 185 us ggml_barrier(...): 287 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 386 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 366 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 462 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 362 us MUL_MAT_ID(ffn_moe_down-55): 320 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 395 us MUL_MAT_ID(ffn_moe_down-56): 195 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 381 us MUL_MAT_ID(ffn_moe_down-57): 172 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 379 us MUL_MAT_ID(ffn_moe_down-58): 198 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 389 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 371 us MUL_MAT_ID(ffn_moe_down-60): 217 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 373 us MUL_MAT_ID(ffn_moe_down-61): 220 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 367 us MUL_MAT_ID(ffn_moe_down-62): 277 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 432 us MUL_MAT_ID(ffn_moe_down-63): 293 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 392 us MUL_MAT_ID(ffn_moe_down-64): 201 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 391 us MUL_MAT_ID(ffn_moe_down-65): 362 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 396 us MUL_MAT_ID(ffn_moe_down-66): 202 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 370 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 373 us MUL_MAT_ID(ffn_moe_down-68): 214 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 369 us MUL_MAT_ID(ffn_moe_down-69): 223 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 382 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 391 us MUL_MAT_ID(ffn_moe_down-71): 537 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 306 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 372 us MUL_MAT_ID(ffn_moe_down-73): 352 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 379 us MUL_MAT_ID(ffn_moe_down-74): 369 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 379 us MUL_MAT_ID(ffn_moe_down-75): 369 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 337 us MUL_MAT_ID(ffn_moe_down-76): 197 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 378 us MUL_MAT_ID(ffn_moe_down-77): 188 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 391 us MUL_MAT_ID(ffn_moe_down-78): 198 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 396 us MUL_MAT_ID(ffn_moe_down-79): 201 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 398 us MUL_MAT_ID(ffn_moe_down-80): 191 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 359 us MUL_MAT_ID(ffn_moe_down-81): 202 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 369 us MUL_MAT_ID(ffn_moe_down-82): 337 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 416 us MUL_MAT_ID(ffn_moe_down-83): 333 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 376 us MUL_MAT_ID(ffn_moe_down-84): 234 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 408 us MUL_MAT_ID(ffn_moe_down-85): 240 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 411 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 420 us MUL_MAT_ID(ffn_moe_down-87): 362 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 386 us MUL_MAT_ID(ffn_moe_down-88): 188 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 405 us MUL_MAT_ID(ffn_moe_down-89): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 386 us MUL_MAT_ID(ffn_moe_down-90): 256 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 353 us MUL_MAT_ID(ffn_moe_down-91): 258 us ggml_barrier(...): 44 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 375 us MUL_MAT_ID(ffn_moe_down-25): 192 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 428 us MUL_MAT_ID(ffn_moe_down-26): 204 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 363 us MUL_MAT_ID(ffn_moe_down-27): 239 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 390 us MUL_MAT_ID(ffn_moe_down-28): 190 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 391 us MUL_MAT_ID(ffn_moe_down-29): 197 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 377 us MUL_MAT_ID(ffn_moe_down-30): 210 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 404 us MUL_MAT_ID(ffn_moe_down-31): 214 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 391 us MUL_MAT_ID(ffn_moe_down-32): 210 us ggml_barrier(...): 281 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 379 us MUL_MAT_ID(ffn_moe_down-33): 207 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 429 us MUL_MAT_ID(ffn_moe_down-34): 228 us ggml_barrier(...): 157 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 408 us MUL_MAT_ID(ffn_moe_down-35): 217 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 401 us MUL_MAT_ID(ffn_moe_down-36): 216 us ggml_barrier(...): 354 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 436 us MUL_MAT_ID(ffn_moe_down-37): 568 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 388 us MUL_MAT_ID(ffn_moe_down-38): 188 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 385 us MUL_MAT_ID(ffn_moe_down-39): 231 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 366 us MUL_MAT_ID(ffn_moe_down-40): 222 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 397 us MUL_MAT_ID(ffn_moe_down-41): 201 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 371 us MUL_MAT_ID(ffn_moe_down-42): 205 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 376 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 354 us MUL_MAT_ID(ffn_moe_down-44): 384 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 398 us MUL_MAT_ID(ffn_moe_down-45): 264 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 405 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 267 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 402 us MUL_MAT_ID(ffn_moe_down-47): 197 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 366 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 369 us MUL_MAT_ID(ffn_moe_down-49): 202 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 394 us MUL_MAT_ID(ffn_moe_down-50): 204 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 342 us MUL_MAT_ID(ffn_moe_down-51): 205 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 411 us MUL_MAT_ID(ffn_moe_down-52): 203 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 382 us MUL_MAT_ID(ffn_moe_down-53): 185 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 374 us MUL_MAT_ID(ffn_moe_down-54): 212 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 209 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 374 us MUL_MAT_ID(ffn_moe_down-56): 196 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 412 us MUL_MAT_ID(ffn_moe_down-57): 189 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 390 us MUL_MAT_ID(ffn_moe_down-58): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 369 us MUL_MAT_ID(ffn_moe_down-59): 547 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 364 us MUL_MAT_ID(ffn_moe_down-60): 185 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 393 us MUL_MAT_ID(ffn_moe_down-61): 188 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 373 us MUL_MAT_ID(ffn_moe_down-62): 204 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 386 us MUL_MAT_ID(ffn_moe_down-63): 187 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 382 us MUL_MAT_ID(ffn_moe_down-64): 389 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 381 us MUL_MAT_ID(ffn_moe_down-65): 196 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 356 us MUL_MAT_ID(ffn_moe_down-66): 239 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 367 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 365 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 357 us MUL_MAT_ID(ffn_moe_down-69): 213 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 374 us MUL_MAT_ID(ffn_moe_down-70): 235 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 386 us MUL_MAT_ID(ffn_moe_down-71): 392 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 187 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 363 us MUL_MAT_ID(ffn_moe_down-73): 202 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 374 us MUL_MAT_ID(ffn_moe_down-74): 205 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 375 us MUL_MAT_ID(ffn_moe_down-75): 190 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 359 us MUL_MAT_ID(ffn_moe_down-76): 199 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 397 us MUL_MAT_ID(ffn_moe_down-77): 213 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 428 us MUL_MAT_ID(ffn_moe_down-78): 188 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 360 us MUL_MAT_ID(ffn_moe_down-79): 390 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 397 us MUL_MAT_ID(ffn_moe_down-80): 336 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 395 us MUL_MAT_ID(ffn_moe_down-81): 188 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 365 us MUL_MAT_ID(ffn_moe_down-82): 384 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 412 us MUL_MAT_ID(ffn_moe_down-83): 266 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 397 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 382 us MUL_MAT_ID(ffn_moe_down-85): 255 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 370 us MUL_MAT_ID(ffn_moe_down-86): 184 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 376 us MUL_MAT_ID(ffn_moe_down-87): 187 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 385 us MUL_MAT_ID(ffn_moe_down-88): 209 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 376 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 374 us MUL_MAT_ID(ffn_moe_down-90): 243 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 413 us MUL_MAT_ID(ffn_moe_down-91): 233 us ggml_barrier(...): 38 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 415 us MUL_MAT_ID(ffn_moe_down-25): 196 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 384 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 375 us MUL_MAT_ID(ffn_moe_down-27): 198 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 387 us MUL_MAT_ID(ffn_moe_down-28): 180 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 389 us MUL_MAT_ID(ffn_moe_down-29): 206 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 383 us MUL_MAT_ID(ffn_moe_down-30): 208 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 362 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 363 us MUL_MAT_ID(ffn_moe_down-32): 197 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 395 us MUL_MAT_ID(ffn_moe_down-33): 189 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 404 us MUL_MAT_ID(ffn_moe_down-34): 243 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 353 us MUL_MAT_ID(ffn_moe_down-35): 195 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 373 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 368 us MUL_MAT_ID(ffn_moe_down-37): 232 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 388 us MUL_MAT_ID(ffn_moe_down-38): 199 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 376 us MUL_MAT_ID(ffn_moe_down-39): 665 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 371 us MUL_MAT_ID(ffn_moe_down-40): 242 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 365 us MUL_MAT_ID(ffn_moe_down-41): 214 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 409 us MUL_MAT_ID(ffn_moe_down-42): 172 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 392 us MUL_MAT_ID(ffn_moe_down-43): 231 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 383 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 354 us MUL_MAT_ID(ffn_moe_down-45): 203 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 383 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 360 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 387 us MUL_MAT_ID(ffn_moe_down-48): 213 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 387 us MUL_MAT_ID(ffn_moe_down-49): 199 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 376 us MUL_MAT_ID(ffn_moe_down-50): 208 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 398 us MUL_MAT_ID(ffn_moe_down-51): 204 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 382 us MUL_MAT_ID(ffn_moe_down-52): 357 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 386 us MUL_MAT_ID(ffn_moe_down-53): 199 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 395 us MUL_MAT_ID(ffn_moe_down-54): 219 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 393 us MUL_MAT_ID(ffn_moe_down-55): 266 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 385 us MUL_MAT_ID(ffn_moe_down-56): 503 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 399 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 356 us MUL_MAT_ID(ffn_moe_down-58): 187 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 360 us MUL_MAT_ID(ffn_moe_down-59): 360 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 356 us MUL_MAT_ID(ffn_moe_down-60): 233 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 411 us MUL_MAT_ID(ffn_moe_down-61): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 383 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 363 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 395 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 389 us MUL_MAT_ID(ffn_moe_down-65): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 400 us MUL_MAT_ID(ffn_moe_down-66): 418 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 373 us MUL_MAT_ID(ffn_moe_down-67): 228 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 376 us MUL_MAT_ID(ffn_moe_down-68): 187 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 415 us MUL_MAT_ID(ffn_moe_down-69): 187 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 393 us MUL_MAT_ID(ffn_moe_down-70): 294 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 379 us MUL_MAT_ID(ffn_moe_down-71): 192 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 368 us MUL_MAT_ID(ffn_moe_down-72): 378 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 359 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 369 us MUL_MAT_ID(ffn_moe_down-74): 261 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 374 us MUL_MAT_ID(ffn_moe_down-75): 209 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 371 us MUL_MAT_ID(ffn_moe_down-76): 182 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 379 us MUL_MAT_ID(ffn_moe_down-77): 523 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 386 us MUL_MAT_ID(ffn_moe_down-78): 179 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 368 us MUL_MAT_ID(ffn_moe_down-79): 207 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 408 us MUL_MAT_ID(ffn_moe_down-80): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 393 us MUL_MAT_ID(ffn_moe_down-81): 185 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 389 us MUL_MAT_ID(ffn_moe_down-82): 329 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 438 us MUL_MAT_ID(ffn_moe_down-83): 189 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 390 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 389 us MUL_MAT_ID(ffn_moe_down-85): 227 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 379 us MUL_MAT_ID(ffn_moe_down-86): 322 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 396 us MUL_MAT_ID(ffn_moe_down-87): 198 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 396 us MUL_MAT_ID(ffn_moe_down-88): 347 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 377 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 348 us MUL_MAT_ID(ffn_moe_down-90): 227 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 412 us MUL_MAT_ID(ffn_moe_down-91): 537 us ggml_barrier(...): 269 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 409 us MUL_MAT_ID(ffn_moe_down-25): 181 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 414 us MUL_MAT_ID(ffn_moe_down-26): 219 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 389 us MUL_MAT_ID(ffn_moe_down-27): 207 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 370 us MUL_MAT_ID(ffn_moe_down-28): 511 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 425 us MUL_MAT_ID(ffn_moe_down-29): 398 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 411 us MUL_MAT_ID(ffn_moe_down-30): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 401 us MUL_MAT_ID(ffn_moe_down-31): 188 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 361 us MUL_MAT_ID(ffn_moe_down-32): 193 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 380 us MUL_MAT_ID(ffn_moe_down-33): 201 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 415 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 390 us MUL_MAT_ID(ffn_moe_down-35): 180 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 376 us MUL_MAT_ID(ffn_moe_down-36): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 403 us MUL_MAT_ID(ffn_moe_down-37): 614 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 377 us MUL_MAT_ID(ffn_moe_down-38): 202 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 384 us MUL_MAT_ID(ffn_moe_down-39): 627 us ggml_barrier(...): 229 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 382 us MUL_MAT_ID(ffn_moe_down-40): 243 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 406 us MUL_MAT_ID(ffn_moe_down-41): 210 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 374 us MUL_MAT_ID(ffn_moe_down-42): 491 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 386 us MUL_MAT_ID(ffn_moe_down-43): 195 us ggml_barrier(...): 237 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 381 us MUL_MAT_ID(ffn_moe_down-44): 223 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 417 us MUL_MAT_ID(ffn_moe_down-45): 401 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 385 us MUL_MAT_ID(ffn_moe_down-46): 511 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 374 us MUL_MAT_ID(ffn_moe_down-47): 199 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 527 us MUL_MAT_ID(ffn_moe_down-48): 212 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 398 us MUL_MAT_ID(ffn_moe_down-49): 197 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 369 us MUL_MAT_ID(ffn_moe_down-50): 364 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 354 us MUL_MAT_ID(ffn_moe_down-51): 237 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 417 us MUL_MAT_ID(ffn_moe_down-52): 199 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 387 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 393 us MUL_MAT_ID(ffn_moe_down-54): 181 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 390 us MUL_MAT_ID(ffn_moe_down-55): 486 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 400 us MUL_MAT_ID(ffn_moe_down-56): 217 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 430 us MUL_MAT_ID(ffn_moe_down-57): 200 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 384 us MUL_MAT_ID(ffn_moe_down-58): 178 us ggml_barrier(...): 291 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 388 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 353 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 366 us MUL_MAT_ID(ffn_moe_down-61): 426 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 422 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 374 us MUL_MAT_ID(ffn_moe_down-63): 190 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 386 us MUL_MAT_ID(ffn_moe_down-64): 186 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 390 us MUL_MAT_ID(ffn_moe_down-65): 207 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 352 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 396 us MUL_MAT_ID(ffn_moe_down-67): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 503 us MUL_MAT_ID(ffn_moe_down-68): 214 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 373 us MUL_MAT_ID(ffn_moe_down-69): 197 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 380 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 358 us MUL_MAT_ID(ffn_moe_down-71): 187 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 361 us MUL_MAT_ID(ffn_moe_down-72): 384 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 367 us MUL_MAT_ID(ffn_moe_down-73): 284 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 369 us MUL_MAT_ID(ffn_moe_down-74): 215 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 365 us MUL_MAT_ID(ffn_moe_down-75): 244 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 521 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 182 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 384 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 403 us MUL_MAT_ID(ffn_moe_down-78): 372 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 414 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 350 us MUL_MAT_ID(ffn_moe_down-80): 226 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 365 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 389 us MUL_MAT_ID(ffn_moe_down-82): 179 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 407 us MUL_MAT_ID(ffn_moe_down-83): 202 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 367 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 379 us MUL_MAT_ID(ffn_moe_down-85): 237 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 411 us MUL_MAT_ID(ffn_moe_down-86): 207 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 360 us MUL_MAT_ID(ffn_moe_down-87): 334 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 405 us MUL_MAT_ID(ffn_moe_down-88): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 409 us MUL_MAT_ID(ffn_moe_down-89): 384 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 384 us MUL_MAT_ID(ffn_moe_down-90): 235 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 385 us MUL_MAT_ID(ffn_moe_down-91): 221 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 266 us ggml_barrier(...): 350 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 372 us MUL_MAT_ID(ffn_moe_down-26): 191 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 406 us MUL_MAT_ID(ffn_moe_down-27): 202 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 392 us MUL_MAT_ID(ffn_moe_down-28): 200 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 409 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 405 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 427 us MUL_MAT_ID(ffn_moe_down-31): 447 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 402 us MUL_MAT_ID(ffn_moe_down-32): 240 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 387 us MUL_MAT_ID(ffn_moe_down-33): 213 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 409 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 359 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 474 us MUL_MAT_ID(ffn_moe_down-36): 187 us ggml_barrier(...): 489 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 401 us MUL_MAT_ID(ffn_moe_down-37): 244 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 454 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 369 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 356 us MUL_MAT_ID(ffn_moe_down-40): 743 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 376 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 390 us MUL_MAT_ID(ffn_moe_down-42): 203 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 371 us MUL_MAT_ID(ffn_moe_down-43): 284 us ggml_barrier(...): 176 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 355 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 375 us MUL_MAT_ID(ffn_moe_down-45): 210 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 348 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 362 us MUL_MAT_ID(ffn_moe_down-47): 266 us ggml_barrier(...): 258 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 358 us MUL_MAT_ID(ffn_moe_down-48): 318 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 391 us MUL_MAT_ID(ffn_moe_down-49): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 390 us MUL_MAT_ID(ffn_moe_down-50): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 368 us MUL_MAT_ID(ffn_moe_down-51): 207 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 428 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 394 us MUL_MAT_ID(ffn_moe_down-53): 206 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 368 us MUL_MAT_ID(ffn_moe_down-54): 242 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 433 us MUL_MAT_ID(ffn_moe_down-56): 179 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 387 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 366 us MUL_MAT_ID(ffn_moe_down-58): 187 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 363 us MUL_MAT_ID(ffn_moe_down-59): 211 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 461 us MUL_MAT_ID(ffn_moe_down-60): 612 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 384 us MUL_MAT_ID(ffn_moe_down-61): 182 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 396 us MUL_MAT_ID(ffn_moe_down-62): 273 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 497 us MUL_MAT_ID(ffn_moe_down-63): 367 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 375 us MUL_MAT_ID(ffn_moe_down-64): 202 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 356 us MUL_MAT_ID(ffn_moe_down-65): 201 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 390 us MUL_MAT_ID(ffn_moe_down-66): 300 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 387 us MUL_MAT_ID(ffn_moe_down-67): 457 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 376 us MUL_MAT_ID(ffn_moe_down-68): 219 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 366 us MUL_MAT_ID(ffn_moe_down-69): 203 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 378 us MUL_MAT_ID(ffn_moe_down-70): 210 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 376 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 356 us MUL_MAT_ID(ffn_moe_down-72): 192 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 402 us MUL_MAT_ID(ffn_moe_down-73): 185 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 365 us MUL_MAT_ID(ffn_moe_down-74): 556 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 356 us MUL_MAT_ID(ffn_moe_down-75): 188 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 394 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 407 us MUL_MAT_ID(ffn_moe_down-77): 217 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 417 us MUL_MAT_ID(ffn_moe_down-78): 204 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 392 us MUL_MAT_ID(ffn_moe_down-79): 208 us ggml_barrier(...): 246 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 403 us MUL_MAT_ID(ffn_moe_down-80): 196 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 377 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 411 us MUL_MAT_ID(ffn_moe_down-82): 205 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 375 us MUL_MAT_ID(ffn_moe_down-83): 228 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 364 us MUL_MAT_ID(ffn_moe_down-84): 226 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 356 us MUL_MAT_ID(ffn_moe_down-85): 672 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 383 us MUL_MAT_ID(ffn_moe_down-86): 196 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 432 us MUL_MAT_ID(ffn_moe_down-87): 574 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 366 us MUL_MAT_ID(ffn_moe_down-88): 188 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 348 us MUL_MAT_ID(ffn_moe_down-89): 221 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 344 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 364 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 61 us GET_ROWS(inp_embd): 26 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 407 us MUL_MAT_ID(ffn_moe_down-25): 183 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 526 us MUL_MAT_ID(ffn_moe_down-26): 224 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 402 us MUL_MAT_ID(ffn_moe_down-27): 343 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 186 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 389 us MUL_MAT_ID(ffn_moe_down-29): 370 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 377 us MUL_MAT_ID(ffn_moe_down-30): 346 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 364 us MUL_MAT_ID(ffn_moe_down-31): 386 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 364 us MUL_MAT_ID(ffn_moe_down-32): 185 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 397 us MUL_MAT_ID(ffn_moe_down-33): 355 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 390 us MUL_MAT_ID(ffn_moe_down-34): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 419 us MUL_MAT_ID(ffn_moe_down-35): 197 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 371 us MUL_MAT_ID(ffn_moe_down-36): 217 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 385 us MUL_MAT_ID(ffn_moe_down-37): 225 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 364 us MUL_MAT_ID(ffn_moe_down-38): 192 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 361 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 393 us MUL_MAT_ID(ffn_moe_down-40): 395 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 353 us MUL_MAT_ID(ffn_moe_down-41): 180 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 387 us MUL_MAT_ID(ffn_moe_down-42): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 364 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 364 us MUL_MAT_ID(ffn_moe_down-44): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 462 us MUL_MAT_ID(ffn_moe_down-45): 221 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 380 us MUL_MAT_ID(ffn_moe_down-46): 206 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 454 us MUL_MAT_ID(ffn_moe_down-47): 248 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 395 us MUL_MAT_ID(ffn_moe_down-48): 535 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 360 us MUL_MAT_ID(ffn_moe_down-49): 193 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 353 us MUL_MAT_ID(ffn_moe_down-50): 192 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 381 us MUL_MAT_ID(ffn_moe_down-51): 194 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 505 us MUL_MAT_ID(ffn_moe_down-52): 213 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 411 us MUL_MAT_ID(ffn_moe_down-53): 201 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 365 us MUL_MAT_ID(ffn_moe_down-54): 272 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 416 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 357 us MUL_MAT_ID(ffn_moe_down-56): 218 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 376 us MUL_MAT_ID(ffn_moe_down-57): 241 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 406 us MUL_MAT_ID(ffn_moe_down-58): 204 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 405 us MUL_MAT_ID(ffn_moe_down-59): 203 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 382 us MUL_MAT_ID(ffn_moe_down-60): 365 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 387 us MUL_MAT_ID(ffn_moe_down-61): 333 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 397 us MUL_MAT_ID(ffn_moe_down-62): 223 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 391 us MUL_MAT_ID(ffn_moe_down-63): 200 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 356 us MUL_MAT_ID(ffn_moe_down-64): 229 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 348 us MUL_MAT_ID(ffn_moe_down-65): 186 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 355 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 359 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 346 us MUL_MAT_ID(ffn_moe_down-68): 303 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 369 us MUL_MAT_ID(ffn_moe_down-69): 245 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 397 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 383 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 393 us MUL_MAT_ID(ffn_moe_down-73): 221 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 374 us MUL_MAT_ID(ffn_moe_down-74): 358 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 357 us MUL_MAT_ID(ffn_moe_down-75): 194 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 375 us MUL_MAT_ID(ffn_moe_down-76): 216 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 357 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 390 us MUL_MAT_ID(ffn_moe_down-78): 411 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 392 us MUL_MAT_ID(ffn_moe_down-79): 171 us ggml_barrier(...): 321 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 384 us MUL_MAT_ID(ffn_moe_down-80): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 382 us MUL_MAT_ID(ffn_moe_down-81): 190 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 369 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 424 us MUL_MAT_ID(ffn_moe_down-83): 426 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 374 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 372 us MUL_MAT_ID(ffn_moe_down-85): 233 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 389 us MUL_MAT_ID(ffn_moe_down-86): 206 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 381 us MUL_MAT_ID(ffn_moe_down-87): 222 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 387 us MUL_MAT_ID(ffn_moe_down-88): 411 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 364 us MUL_MAT_ID(ffn_moe_down-89): 228 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 395 us MUL_MAT_ID(ffn_moe_down-90): 236 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 362 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 49 us GET_ROWS(inp_embd): 23 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 203 us ggml_barrier(...): 211 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 376 us MUL_MAT_ID(ffn_moe_down-26): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 388 us MUL_MAT_ID(ffn_moe_down-27): 183 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 398 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 393 us MUL_MAT_ID(ffn_moe_down-29): 192 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 394 us MUL_MAT_ID(ffn_moe_down-30): 212 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 407 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 380 us MUL_MAT_ID(ffn_moe_down-32): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 370 us MUL_MAT_ID(ffn_moe_down-33): 215 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 396 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 357 us MUL_MAT_ID(ffn_moe_down-35): 201 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 435 us MUL_MAT_ID(ffn_moe_down-36): 263 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 382 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 375 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 388 us MUL_MAT_ID(ffn_moe_down-39): 249 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 391 us MUL_MAT_ID(ffn_moe_down-40): 395 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 365 us MUL_MAT_ID(ffn_moe_down-41): 203 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 372 us MUL_MAT_ID(ffn_moe_down-42): 206 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 379 us MUL_MAT_ID(ffn_moe_down-43): 192 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 401 us MUL_MAT_ID(ffn_moe_down-44): 196 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 425 us MUL_MAT_ID(ffn_moe_down-45): 203 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 396 us MUL_MAT_ID(ffn_moe_down-46): 196 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 402 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 360 us MUL_MAT_ID(ffn_moe_down-48): 183 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 204 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 404 us MUL_MAT_ID(ffn_moe_down-50): 189 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 370 us MUL_MAT_ID(ffn_moe_down-51): 210 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 423 us MUL_MAT_ID(ffn_moe_down-52): 201 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 397 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 385 us MUL_MAT_ID(ffn_moe_down-54): 270 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 203 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 366 us MUL_MAT_ID(ffn_moe_down-56): 365 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 404 us MUL_MAT_ID(ffn_moe_down-57): 416 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 410 us MUL_MAT_ID(ffn_moe_down-58): 186 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 374 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 386 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 357 us MUL_MAT_ID(ffn_moe_down-61): 199 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 379 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 364 us MUL_MAT_ID(ffn_moe_down-63): 356 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 385 us MUL_MAT_ID(ffn_moe_down-64): 209 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 365 us MUL_MAT_ID(ffn_moe_down-65): 227 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 404 us MUL_MAT_ID(ffn_moe_down-66): 199 us ggml_barrier(...): 352 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 367 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 415 us MUL_MAT_ID(ffn_moe_down-68): 191 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 382 us MUL_MAT_ID(ffn_moe_down-69): 202 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 375 us MUL_MAT_ID(ffn_moe_down-70): 212 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 379 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 395 us MUL_MAT_ID(ffn_moe_down-72): 206 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 394 us MUL_MAT_ID(ffn_moe_down-73): 186 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 379 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 371 us MUL_MAT_ID(ffn_moe_down-75): 185 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 365 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 412 us MUL_MAT_ID(ffn_moe_down-77): 214 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 406 us MUL_MAT_ID(ffn_moe_down-78): 185 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 418 us MUL_MAT_ID(ffn_moe_down-79): 197 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 463 us MUL_MAT_ID(ffn_moe_down-80): 551 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 399 us MUL_MAT_ID(ffn_moe_down-81): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 404 us MUL_MAT_ID(ffn_moe_down-82): 189 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 505 us MUL_MAT_ID(ffn_moe_down-83): 207 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 398 us MUL_MAT_ID(ffn_moe_down-84): 250 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 400 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 399 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 396 us MUL_MAT_ID(ffn_moe_down-87): 230 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 393 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 356 us MUL_MAT_ID(ffn_moe_down-89): 274 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 380 us MUL_MAT_ID(ffn_moe_down-90): 292 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 419 us MUL_MAT_ID(ffn_moe_down-91): 242 us ggml_barrier(...): 32 us GET_ROWS(inp_embd): 12 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 404 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 199 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 400 us MUL_MAT_ID(ffn_moe_down-26): 183 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 400 us MUL_MAT_ID(ffn_moe_down-27): 392 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 358 us MUL_MAT_ID(ffn_moe_down-28): 376 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 367 us MUL_MAT_ID(ffn_moe_down-29): 366 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 374 us MUL_MAT_ID(ffn_moe_down-30): 215 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 369 us MUL_MAT_ID(ffn_moe_down-31): 361 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 381 us MUL_MAT_ID(ffn_moe_down-32): 202 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 406 us MUL_MAT_ID(ffn_moe_down-33): 218 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 377 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 356 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 359 us MUL_MAT_ID(ffn_moe_down-36): 217 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 365 us MUL_MAT_ID(ffn_moe_down-37): 237 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 393 us MUL_MAT_ID(ffn_moe_down-38): 216 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 387 us MUL_MAT_ID(ffn_moe_down-39): 243 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 394 us MUL_MAT_ID(ffn_moe_down-40): 287 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 359 us MUL_MAT_ID(ffn_moe_down-41): 195 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 423 us MUL_MAT_ID(ffn_moe_down-42): 264 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 429 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 356 us MUL_MAT_ID(ffn_moe_down-44): 198 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 384 us MUL_MAT_ID(ffn_moe_down-45): 207 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 400 us MUL_MAT_ID(ffn_moe_down-46): 218 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 391 us MUL_MAT_ID(ffn_moe_down-47): 187 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 349 us MUL_MAT_ID(ffn_moe_down-48): 199 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 401 us MUL_MAT_ID(ffn_moe_down-49): 210 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 384 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 348 us MUL_MAT_ID(ffn_moe_down-51): 212 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 382 us MUL_MAT_ID(ffn_moe_down-52): 201 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 390 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 397 us MUL_MAT_ID(ffn_moe_down-54): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 377 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 364 us MUL_MAT_ID(ffn_moe_down-56): 241 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 381 us MUL_MAT_ID(ffn_moe_down-57): 193 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 409 us MUL_MAT_ID(ffn_moe_down-58): 209 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 367 us MUL_MAT_ID(ffn_moe_down-59): 208 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 375 us MUL_MAT_ID(ffn_moe_down-60): 331 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 401 us MUL_MAT_ID(ffn_moe_down-61): 222 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 398 us MUL_MAT_ID(ffn_moe_down-62): 328 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 379 us MUL_MAT_ID(ffn_moe_down-63): 322 us ggml_barrier(...): 95 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 358 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 403 us MUL_MAT_ID(ffn_moe_down-65): 384 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 383 us MUL_MAT_ID(ffn_moe_down-66): 187 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 427 us MUL_MAT_ID(ffn_moe_down-67): 241 us ggml_barrier(...): 232 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 394 us MUL_MAT_ID(ffn_moe_down-68): 195 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 399 us MUL_MAT_ID(ffn_moe_down-69): 200 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 365 us MUL_MAT_ID(ffn_moe_down-70): 255 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 358 us MUL_MAT_ID(ffn_moe_down-71): 377 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 407 us MUL_MAT_ID(ffn_moe_down-72): 213 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 401 us MUL_MAT_ID(ffn_moe_down-73): 320 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 386 us MUL_MAT_ID(ffn_moe_down-74): 185 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 360 us MUL_MAT_ID(ffn_moe_down-75): 199 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 414 us MUL_MAT_ID(ffn_moe_down-76): 253 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 348 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 384 us MUL_MAT_ID(ffn_moe_down-78): 376 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 394 us MUL_MAT_ID(ffn_moe_down-79): 344 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 407 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 374 us MUL_MAT_ID(ffn_moe_down-81): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 399 us MUL_MAT_ID(ffn_moe_down-82): 322 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 396 us MUL_MAT_ID(ffn_moe_down-83): 216 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 383 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 409 us MUL_MAT_ID(ffn_moe_down-85): 239 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 344 us MUL_MAT_ID(ffn_moe_down-86): 298 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 375 us MUL_MAT_ID(ffn_moe_down-87): 236 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 363 us MUL_MAT_ID(ffn_moe_down-88): 201 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 400 us MUL_MAT_ID(ffn_moe_down-89): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 361 us MUL_MAT_ID(ffn_moe_down-90): 243 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 395 us MUL_MAT_ID(ffn_moe_down-91): 245 us ggml_barrier(...): 21 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 406 us MUL_MAT_ID(ffn_moe_down-25): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 363 us MUL_MAT_ID(ffn_moe_down-26): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 376 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 389 us MUL_MAT_ID(ffn_moe_down-28): 196 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 361 us MUL_MAT_ID(ffn_moe_down-29): 180 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 400 us MUL_MAT_ID(ffn_moe_down-30): 206 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 378 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 389 us MUL_MAT_ID(ffn_moe_down-32): 212 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 360 us MUL_MAT_ID(ffn_moe_down-33): 224 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 412 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 399 us MUL_MAT_ID(ffn_moe_down-35): 194 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 378 us MUL_MAT_ID(ffn_moe_down-36): 193 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 383 us MUL_MAT_ID(ffn_moe_down-37): 224 us ggml_barrier(...): 195 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 386 us MUL_MAT_ID(ffn_moe_down-38): 184 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 365 us MUL_MAT_ID(ffn_moe_down-39): 694 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 376 us MUL_MAT_ID(ffn_moe_down-40): 700 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 390 us MUL_MAT_ID(ffn_moe_down-41): 193 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 514 us MUL_MAT_ID(ffn_moe_down-42): 187 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 393 us MUL_MAT_ID(ffn_moe_down-43): 270 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 361 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 518 us MUL_MAT_ID(ffn_moe_down-45): 208 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 381 us MUL_MAT_ID(ffn_moe_down-46): 201 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 409 us MUL_MAT_ID(ffn_moe_down-47): 201 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 393 us MUL_MAT_ID(ffn_moe_down-48): 365 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 361 us MUL_MAT_ID(ffn_moe_down-49): 201 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 377 us MUL_MAT_ID(ffn_moe_down-50): 220 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 369 us MUL_MAT_ID(ffn_moe_down-51): 200 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 400 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 354 us MUL_MAT_ID(ffn_moe_down-53): 190 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 412 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 454 us MUL_MAT_ID(ffn_moe_down-55): 190 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 397 us MUL_MAT_ID(ffn_moe_down-56): 532 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 381 us MUL_MAT_ID(ffn_moe_down-57): 187 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 398 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 408 us MUL_MAT_ID(ffn_moe_down-59): 210 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 381 us MUL_MAT_ID(ffn_moe_down-60): 358 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 378 us MUL_MAT_ID(ffn_moe_down-61): 222 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 373 us MUL_MAT_ID(ffn_moe_down-63): 552 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 381 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 369 us MUL_MAT_ID(ffn_moe_down-65): 182 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 352 us MUL_MAT_ID(ffn_moe_down-66): 304 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 380 us MUL_MAT_ID(ffn_moe_down-67): 415 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 358 us MUL_MAT_ID(ffn_moe_down-68): 224 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 394 us MUL_MAT_ID(ffn_moe_down-69): 314 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 369 us MUL_MAT_ID(ffn_moe_down-70): 207 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 435 us MUL_MAT_ID(ffn_moe_down-71): 346 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 423 us MUL_MAT_ID(ffn_moe_down-72): 233 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 362 us MUL_MAT_ID(ffn_moe_down-73): 315 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 358 us MUL_MAT_ID(ffn_moe_down-74): 199 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 381 us MUL_MAT_ID(ffn_moe_down-75): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 380 us MUL_MAT_ID(ffn_moe_down-76): 360 us ggml_barrier(...): 344 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 405 us MUL_MAT_ID(ffn_moe_down-77): 256 us ggml_barrier(...): 422 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 408 us MUL_MAT_ID(ffn_moe_down-78): 202 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 395 us MUL_MAT_ID(ffn_moe_down-79): 186 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 447 us MUL_MAT_ID(ffn_moe_down-80): 180 us ggml_barrier(...): 275 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 367 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 390 us MUL_MAT_ID(ffn_moe_down-82): 204 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 395 us MUL_MAT_ID(ffn_moe_down-83): 221 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 364 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 351 us MUL_MAT_ID(ffn_moe_down-85): 249 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 371 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 376 us MUL_MAT_ID(ffn_moe_down-87): 209 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 381 us MUL_MAT_ID(ffn_moe_down-88): 206 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 359 us MUL_MAT_ID(ffn_moe_down-89): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 470 us MUL_MAT_ID(ffn_moe_down-90): 217 us ggml_barrier(...): 218 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 369 us MUL_MAT_ID(ffn_moe_down-91): 658 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 414 us MUL_MAT_ID(ffn_moe_down-25): 199 us ggml_barrier(...): 131 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 386 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 369 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 386 us MUL_MAT_ID(ffn_moe_down-28): 214 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 383 us MUL_MAT_ID(ffn_moe_down-29): 192 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 402 us MUL_MAT_ID(ffn_moe_down-30): 399 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 361 us MUL_MAT_ID(ffn_moe_down-31): 190 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 384 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 353 us MUL_MAT_ID(ffn_moe_down-33): 197 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 377 us MUL_MAT_ID(ffn_moe_down-34): 199 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 407 us MUL_MAT_ID(ffn_moe_down-35): 251 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 368 us MUL_MAT_ID(ffn_moe_down-36): 197 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 351 us MUL_MAT_ID(ffn_moe_down-37): 226 us ggml_barrier(...): 193 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 390 us MUL_MAT_ID(ffn_moe_down-38): 197 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 366 us MUL_MAT_ID(ffn_moe_down-39): 252 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 366 us MUL_MAT_ID(ffn_moe_down-40): 240 us ggml_barrier(...): 147 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 374 us MUL_MAT_ID(ffn_moe_down-41): 359 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 364 us MUL_MAT_ID(ffn_moe_down-42): 181 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 412 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 145 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 366 us MUL_MAT_ID(ffn_moe_down-44): 194 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 396 us MUL_MAT_ID(ffn_moe_down-45): 218 us ggml_barrier(...): 180 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 338 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 408 us MUL_MAT_ID(ffn_moe_down-47): 299 us ggml_barrier(...): 202 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 370 us MUL_MAT_ID(ffn_moe_down-48): 404 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 376 us MUL_MAT_ID(ffn_moe_down-49): 371 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 332 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 356 us MUL_MAT_ID(ffn_moe_down-51): 220 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 428 us MUL_MAT_ID(ffn_moe_down-52): 205 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 385 us MUL_MAT_ID(ffn_moe_down-53): 200 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 363 us MUL_MAT_ID(ffn_moe_down-54): 225 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 363 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 393 us MUL_MAT_ID(ffn_moe_down-56): 225 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 403 us MUL_MAT_ID(ffn_moe_down-57): 378 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 373 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 383 us MUL_MAT_ID(ffn_moe_down-59): 191 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 377 us MUL_MAT_ID(ffn_moe_down-60): 189 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 378 us MUL_MAT_ID(ffn_moe_down-61): 176 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 356 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 375 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 399 us MUL_MAT_ID(ffn_moe_down-64): 388 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 401 us MUL_MAT_ID(ffn_moe_down-65): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 394 us MUL_MAT_ID(ffn_moe_down-66): 185 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 382 us MUL_MAT_ID(ffn_moe_down-67): 206 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 378 us MUL_MAT_ID(ffn_moe_down-68): 210 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 383 us MUL_MAT_ID(ffn_moe_down-69): 537 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 375 us MUL_MAT_ID(ffn_moe_down-70): 190 us ggml_barrier(...): 81 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 379 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 343 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 366 us MUL_MAT_ID(ffn_moe_down-73): 186 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 391 us MUL_MAT_ID(ffn_moe_down-74): 189 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 409 us MUL_MAT_ID(ffn_moe_down-75): 204 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 403 us MUL_MAT_ID(ffn_moe_down-76): 387 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 389 us MUL_MAT_ID(ffn_moe_down-77): 194 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 384 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 347 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 185 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 387 us MUL_MAT_ID(ffn_moe_down-80): 225 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 442 us MUL_MAT_ID(ffn_moe_down-81): 206 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 356 us MUL_MAT_ID(ffn_moe_down-82): 259 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 383 us MUL_MAT_ID(ffn_moe_down-83): 272 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 388 us MUL_MAT_ID(ffn_moe_down-84): 237 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 402 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 399 us MUL_MAT_ID(ffn_moe_down-86): 190 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 382 us MUL_MAT_ID(ffn_moe_down-87): 212 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 404 us MUL_MAT_ID(ffn_moe_down-88): 301 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 354 us MUL_MAT_ID(ffn_moe_down-89): 410 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 364 us MUL_MAT_ID(ffn_moe_down-90): 230 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 414 us MUL_MAT_ID(ffn_moe_down-91): 238 us ggml_barrier(...): 7 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 417 us MUL_MAT_ID(ffn_moe_down-25): 251 us ggml_barrier(...): 223 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 403 us MUL_MAT_ID(ffn_moe_down-26): 185 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 395 us MUL_MAT_ID(ffn_moe_down-27): 504 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 401 us MUL_MAT_ID(ffn_moe_down-28): 341 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 371 us MUL_MAT_ID(ffn_moe_down-29): 186 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 395 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 403 us MUL_MAT_ID(ffn_moe_down-32): 207 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 430 us MUL_MAT_ID(ffn_moe_down-33): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 408 us MUL_MAT_ID(ffn_moe_down-34): 529 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 403 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 368 us MUL_MAT_ID(ffn_moe_down-36): 399 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 403 us MUL_MAT_ID(ffn_moe_down-37): 244 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 406 us MUL_MAT_ID(ffn_moe_down-38): 499 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 390 us MUL_MAT_ID(ffn_moe_down-39): 227 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 372 us MUL_MAT_ID(ffn_moe_down-40): 232 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 359 us MUL_MAT_ID(ffn_moe_down-41): 196 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 385 us MUL_MAT_ID(ffn_moe_down-42): 223 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 408 us MUL_MAT_ID(ffn_moe_down-43): 209 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 411 us MUL_MAT_ID(ffn_moe_down-44): 247 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 401 us MUL_MAT_ID(ffn_moe_down-45): 208 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 443 us MUL_MAT_ID(ffn_moe_down-46): 208 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 398 us MUL_MAT_ID(ffn_moe_down-47): 200 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 368 us MUL_MAT_ID(ffn_moe_down-48): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 342 us MUL_MAT_ID(ffn_moe_down-49): 202 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 366 us MUL_MAT_ID(ffn_moe_down-50): 187 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 357 us MUL_MAT_ID(ffn_moe_down-51): 393 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 379 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 391 us MUL_MAT_ID(ffn_moe_down-53): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 362 us MUL_MAT_ID(ffn_moe_down-54): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 370 us MUL_MAT_ID(ffn_moe_down-55): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 363 us MUL_MAT_ID(ffn_moe_down-56): 211 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 362 us MUL_MAT_ID(ffn_moe_down-57): 215 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 387 us MUL_MAT_ID(ffn_moe_down-58): 192 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 383 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 516 us MUL_MAT_ID(ffn_moe_down-60): 214 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 415 us MUL_MAT_ID(ffn_moe_down-61): 213 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 376 us MUL_MAT_ID(ffn_moe_down-62): 207 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 382 us MUL_MAT_ID(ffn_moe_down-63): 298 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 399 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 372 us MUL_MAT_ID(ffn_moe_down-65): 186 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 378 us MUL_MAT_ID(ffn_moe_down-66): 208 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 399 us MUL_MAT_ID(ffn_moe_down-67): 208 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 387 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 408 us MUL_MAT_ID(ffn_moe_down-69): 214 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 401 us MUL_MAT_ID(ffn_moe_down-70): 209 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 396 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 382 us MUL_MAT_ID(ffn_moe_down-72): 456 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 396 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 372 us MUL_MAT_ID(ffn_moe_down-74): 217 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 361 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 370 us MUL_MAT_ID(ffn_moe_down-76): 182 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 395 us MUL_MAT_ID(ffn_moe_down-77): 207 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 410 us MUL_MAT_ID(ffn_moe_down-78): 189 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 370 us MUL_MAT_ID(ffn_moe_down-79): 191 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 386 us MUL_MAT_ID(ffn_moe_down-80): 187 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 373 us MUL_MAT_ID(ffn_moe_down-81): 177 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 351 us MUL_MAT_ID(ffn_moe_down-82): 184 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 337 us MUL_MAT_ID(ffn_moe_down-83): 187 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 385 us MUL_MAT_ID(ffn_moe_down-84): 225 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 367 us MUL_MAT_ID(ffn_moe_down-85): 247 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 386 us MUL_MAT_ID(ffn_moe_down-86): 204 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 365 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 397 us MUL_MAT_ID(ffn_moe_down-88): 183 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 401 us MUL_MAT_ID(ffn_moe_down-89): 195 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 382 us MUL_MAT_ID(ffn_moe_down-90): 231 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 412 us MUL_MAT_ID(ffn_moe_down-91): 237 us ggml_barrier(...): 20 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 393 us MUL_MAT_ID(ffn_moe_down-25): 216 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 371 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 378 us MUL_MAT_ID(ffn_moe_down-27): 214 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 359 us MUL_MAT_ID(ffn_moe_down-28): 182 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 391 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 377 us MUL_MAT_ID(ffn_moe_down-30): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 363 us MUL_MAT_ID(ffn_moe_down-31): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 374 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 403 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 359 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 419 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 373 us MUL_MAT_ID(ffn_moe_down-36): 197 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 411 us MUL_MAT_ID(ffn_moe_down-37): 240 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 372 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 361 us MUL_MAT_ID(ffn_moe_down-39): 246 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 370 us MUL_MAT_ID(ffn_moe_down-40): 235 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 403 us MUL_MAT_ID(ffn_moe_down-41): 190 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 356 us MUL_MAT_ID(ffn_moe_down-42): 266 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 412 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 383 us MUL_MAT_ID(ffn_moe_down-44): 214 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 363 us MUL_MAT_ID(ffn_moe_down-45): 426 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 370 us MUL_MAT_ID(ffn_moe_down-46): 205 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 393 us MUL_MAT_ID(ffn_moe_down-47): 188 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 373 us MUL_MAT_ID(ffn_moe_down-48): 185 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 367 us MUL_MAT_ID(ffn_moe_down-49): 523 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 404 us MUL_MAT_ID(ffn_moe_down-50): 404 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 374 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 358 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 395 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 410 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 366 us MUL_MAT_ID(ffn_moe_down-55): 331 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 376 us MUL_MAT_ID(ffn_moe_down-56): 211 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 357 us MUL_MAT_ID(ffn_moe_down-57): 194 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 369 us MUL_MAT_ID(ffn_moe_down-58): 235 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 384 us MUL_MAT_ID(ffn_moe_down-59): 206 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 381 us MUL_MAT_ID(ffn_moe_down-60): 207 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 366 us MUL_MAT_ID(ffn_moe_down-61): 182 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 392 us MUL_MAT_ID(ffn_moe_down-62): 189 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 378 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 391 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 383 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 366 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 434 us MUL_MAT_ID(ffn_moe_down-67): 211 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 361 us MUL_MAT_ID(ffn_moe_down-68): 179 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 382 us MUL_MAT_ID(ffn_moe_down-69): 195 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 403 us MUL_MAT_ID(ffn_moe_down-70): 215 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 410 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 374 us MUL_MAT_ID(ffn_moe_down-72): 187 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 383 us MUL_MAT_ID(ffn_moe_down-73): 185 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 370 us MUL_MAT_ID(ffn_moe_down-74): 217 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 355 us MUL_MAT_ID(ffn_moe_down-75): 546 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 425 us MUL_MAT_ID(ffn_moe_down-76): 432 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 385 us MUL_MAT_ID(ffn_moe_down-77): 199 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 412 us MUL_MAT_ID(ffn_moe_down-78): 211 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 377 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 396 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 411 us MUL_MAT_ID(ffn_moe_down-81): 539 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 402 us MUL_MAT_ID(ffn_moe_down-82): 207 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 404 us MUL_MAT_ID(ffn_moe_down-83): 398 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 398 us MUL_MAT_ID(ffn_moe_down-84): 366 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 429 us MUL_MAT_ID(ffn_moe_down-85): 242 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 365 us MUL_MAT_ID(ffn_moe_down-86): 231 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 339 us MUL_MAT_ID(ffn_moe_down-87): 183 us ggml_barrier(...): 83 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 349 us MUL_MAT_ID(ffn_moe_down-88): 203 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 390 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 363 us MUL_MAT_ID(ffn_moe_down-90): 299 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 386 us MUL_MAT_ID(ffn_moe_down-91): 276 us ggml_barrier(...): 28 us GET_ROWS(inp_embd): 22 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 386 us MUL_MAT_ID(ffn_moe_down-25): 196 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 370 us MUL_MAT_ID(ffn_moe_down-26): 375 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 390 us MUL_MAT_ID(ffn_moe_down-27): 209 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 411 us MUL_MAT_ID(ffn_moe_down-28): 191 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 403 us MUL_MAT_ID(ffn_moe_down-29): 214 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 543 us MUL_MAT_ID(ffn_moe_down-30): 183 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 353 us MUL_MAT_ID(ffn_moe_down-31): 216 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 379 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 356 us MUL_MAT_ID(ffn_moe_down-33): 528 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 372 us MUL_MAT_ID(ffn_moe_down-34): 198 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 389 us MUL_MAT_ID(ffn_moe_down-35): 191 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 383 us MUL_MAT_ID(ffn_moe_down-36): 197 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 390 us MUL_MAT_ID(ffn_moe_down-37): 245 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 377 us MUL_MAT_ID(ffn_moe_down-38): 205 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 365 us MUL_MAT_ID(ffn_moe_down-39): 248 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 363 us MUL_MAT_ID(ffn_moe_down-40): 228 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 482 us MUL_MAT_ID(ffn_moe_down-41): 292 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 386 us MUL_MAT_ID(ffn_moe_down-42): 301 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 370 us MUL_MAT_ID(ffn_moe_down-43): 535 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 348 us MUL_MAT_ID(ffn_moe_down-44): 228 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 374 us MUL_MAT_ID(ffn_moe_down-45): 212 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 366 us MUL_MAT_ID(ffn_moe_down-46): 210 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 376 us MUL_MAT_ID(ffn_moe_down-47): 190 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 382 us MUL_MAT_ID(ffn_moe_down-48): 212 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 398 us MUL_MAT_ID(ffn_moe_down-49): 213 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 399 us MUL_MAT_ID(ffn_moe_down-50): 182 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 401 us MUL_MAT_ID(ffn_moe_down-51): 188 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 414 us MUL_MAT_ID(ffn_moe_down-52): 193 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 426 us MUL_MAT_ID(ffn_moe_down-53): 389 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 480 us MUL_MAT_ID(ffn_moe_down-54): 392 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 415 us MUL_MAT_ID(ffn_moe_down-55): 202 us ggml_barrier(...): 103 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 412 us MUL_MAT_ID(ffn_moe_down-56): 253 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 374 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 391 us MUL_MAT_ID(ffn_moe_down-58): 260 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 364 us MUL_MAT_ID(ffn_moe_down-59): 190 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 403 us MUL_MAT_ID(ffn_moe_down-60): 535 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 395 us MUL_MAT_ID(ffn_moe_down-61): 250 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 385 us MUL_MAT_ID(ffn_moe_down-62): 229 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 375 us MUL_MAT_ID(ffn_moe_down-63): 205 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 372 us MUL_MAT_ID(ffn_moe_down-64): 406 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 379 us MUL_MAT_ID(ffn_moe_down-65): 197 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 379 us MUL_MAT_ID(ffn_moe_down-66): 211 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 379 us MUL_MAT_ID(ffn_moe_down-67): 194 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 405 us MUL_MAT_ID(ffn_moe_down-68): 211 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 368 us MUL_MAT_ID(ffn_moe_down-69): 456 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 393 us MUL_MAT_ID(ffn_moe_down-70): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 396 us MUL_MAT_ID(ffn_moe_down-71): 189 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 393 us MUL_MAT_ID(ffn_moe_down-72): 271 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 386 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 363 us MUL_MAT_ID(ffn_moe_down-74): 442 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 411 us MUL_MAT_ID(ffn_moe_down-75): 354 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 553 us MUL_MAT_ID(ffn_moe_down-76): 219 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 357 us MUL_MAT_ID(ffn_moe_down-77): 540 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 389 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 368 us MUL_MAT_ID(ffn_moe_down-79): 189 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 406 us MUL_MAT_ID(ffn_moe_down-80): 217 us ggml_barrier(...): 7 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 379 us MUL_MAT_ID(ffn_moe_down-81): 201 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 375 us MUL_MAT_ID(ffn_moe_down-82): 182 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 375 us MUL_MAT_ID(ffn_moe_down-83): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 496 us MUL_MAT_ID(ffn_moe_down-84): 252 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 416 us MUL_MAT_ID(ffn_moe_down-85): 248 us ggml_barrier(...): 6 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 398 us MUL_MAT_ID(ffn_moe_down-86): 186 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 394 us MUL_MAT_ID(ffn_moe_down-87): 205 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 394 us MUL_MAT_ID(ffn_moe_down-88): 303 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 359 us MUL_MAT_ID(ffn_moe_down-89): 215 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 401 us MUL_MAT_ID(ffn_moe_down-90): 249 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 408 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 51 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 359 us MUL_MAT_ID(ffn_moe_down-25): 206 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 359 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 374 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 361 us MUL_MAT_ID(ffn_moe_down-28): 197 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 394 us MUL_MAT_ID(ffn_moe_down-29): 236 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 377 us MUL_MAT_ID(ffn_moe_down-30): 185 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 389 us MUL_MAT_ID(ffn_moe_down-31): 198 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 501 us MUL_MAT_ID(ffn_moe_down-32): 251 us ggml_barrier(...): 250 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 370 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 162 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 398 us MUL_MAT_ID(ffn_moe_down-34): 196 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 376 us MUL_MAT_ID(ffn_moe_down-35): 214 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 465 us MUL_MAT_ID(ffn_moe_down-36): 176 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 385 us MUL_MAT_ID(ffn_moe_down-37): 244 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 530 us MUL_MAT_ID(ffn_moe_down-38): 207 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 410 us MUL_MAT_ID(ffn_moe_down-39): 264 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 361 us MUL_MAT_ID(ffn_moe_down-40): 738 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 362 us MUL_MAT_ID(ffn_moe_down-41): 215 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 375 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 420 us MUL_MAT_ID(ffn_moe_down-43): 189 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 335 us MUL_MAT_ID(ffn_moe_down-44): 198 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 408 us MUL_MAT_ID(ffn_moe_down-45): 205 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 372 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 393 us MUL_MAT_ID(ffn_moe_down-47): 187 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 363 us MUL_MAT_ID(ffn_moe_down-48): 232 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 362 us MUL_MAT_ID(ffn_moe_down-49): 222 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 366 us MUL_MAT_ID(ffn_moe_down-50): 347 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 397 us MUL_MAT_ID(ffn_moe_down-51): 521 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 367 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 372 us MUL_MAT_ID(ffn_moe_down-53): 208 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 414 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 403 us MUL_MAT_ID(ffn_moe_down-55): 372 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 370 us MUL_MAT_ID(ffn_moe_down-56): 201 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 385 us MUL_MAT_ID(ffn_moe_down-57): 190 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 395 us MUL_MAT_ID(ffn_moe_down-58): 188 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 369 us MUL_MAT_ID(ffn_moe_down-59): 206 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 398 us MUL_MAT_ID(ffn_moe_down-60): 207 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 392 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 399 us MUL_MAT_ID(ffn_moe_down-62): 198 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 384 us MUL_MAT_ID(ffn_moe_down-63): 394 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 344 us MUL_MAT_ID(ffn_moe_down-64): 189 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 377 us MUL_MAT_ID(ffn_moe_down-65): 217 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 400 us MUL_MAT_ID(ffn_moe_down-66): 183 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 382 us MUL_MAT_ID(ffn_moe_down-67): 189 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 360 us MUL_MAT_ID(ffn_moe_down-68): 203 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 369 us MUL_MAT_ID(ffn_moe_down-69): 333 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 389 us MUL_MAT_ID(ffn_moe_down-70): 194 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 392 us MUL_MAT_ID(ffn_moe_down-71): 186 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 374 us MUL_MAT_ID(ffn_moe_down-72): 201 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 348 us MUL_MAT_ID(ffn_moe_down-73): 191 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 391 us MUL_MAT_ID(ffn_moe_down-74): 561 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 351 us MUL_MAT_ID(ffn_moe_down-75): 192 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 406 us MUL_MAT_ID(ffn_moe_down-76): 190 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 403 us MUL_MAT_ID(ffn_moe_down-77): 189 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 384 us MUL_MAT_ID(ffn_moe_down-78): 382 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 382 us MUL_MAT_ID(ffn_moe_down-79): 202 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 348 us MUL_MAT_ID(ffn_moe_down-80): 201 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 383 us MUL_MAT_ID(ffn_moe_down-81): 384 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 377 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 368 us MUL_MAT_ID(ffn_moe_down-83): 207 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 406 us MUL_MAT_ID(ffn_moe_down-84): 226 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 359 us MUL_MAT_ID(ffn_moe_down-85): 225 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 363 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 379 us MUL_MAT_ID(ffn_moe_down-87): 191 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 390 us MUL_MAT_ID(ffn_moe_down-88): 189 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 402 us MUL_MAT_ID(ffn_moe_down-89): 215 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 369 us MUL_MAT_ID(ffn_moe_down-90): 222 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 384 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 34 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 404 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 388 us MUL_MAT_ID(ffn_moe_down-26): 557 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 358 us MUL_MAT_ID(ffn_moe_down-27): 212 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 398 us MUL_MAT_ID(ffn_moe_down-28): 219 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 387 us MUL_MAT_ID(ffn_moe_down-29): 541 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 376 us MUL_MAT_ID(ffn_moe_down-30): 379 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 401 us MUL_MAT_ID(ffn_moe_down-31): 195 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 397 us MUL_MAT_ID(ffn_moe_down-32): 207 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 415 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 379 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 368 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 363 us MUL_MAT_ID(ffn_moe_down-36): 189 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 400 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 375 us MUL_MAT_ID(ffn_moe_down-38): 206 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 395 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 412 us MUL_MAT_ID(ffn_moe_down-40): 418 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 420 us MUL_MAT_ID(ffn_moe_down-41): 264 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 378 us MUL_MAT_ID(ffn_moe_down-42): 196 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 386 us MUL_MAT_ID(ffn_moe_down-43): 203 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 388 us MUL_MAT_ID(ffn_moe_down-44): 189 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 402 us MUL_MAT_ID(ffn_moe_down-45): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 365 us MUL_MAT_ID(ffn_moe_down-46): 206 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 381 us MUL_MAT_ID(ffn_moe_down-47): 394 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 377 us MUL_MAT_ID(ffn_moe_down-48): 187 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 390 us MUL_MAT_ID(ffn_moe_down-49): 192 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 360 us MUL_MAT_ID(ffn_moe_down-50): 244 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 367 us MUL_MAT_ID(ffn_moe_down-51): 203 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 395 us MUL_MAT_ID(ffn_moe_down-52): 215 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 359 us MUL_MAT_ID(ffn_moe_down-53): 394 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 367 us MUL_MAT_ID(ffn_moe_down-54): 187 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 390 us MUL_MAT_ID(ffn_moe_down-55): 182 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 367 us MUL_MAT_ID(ffn_moe_down-56): 187 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 409 us MUL_MAT_ID(ffn_moe_down-57): 187 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 397 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 358 us MUL_MAT_ID(ffn_moe_down-59): 187 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 384 us MUL_MAT_ID(ffn_moe_down-60): 185 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 405 us MUL_MAT_ID(ffn_moe_down-61): 202 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 391 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 360 us MUL_MAT_ID(ffn_moe_down-63): 191 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 356 us MUL_MAT_ID(ffn_moe_down-64): 245 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 405 us MUL_MAT_ID(ffn_moe_down-65): 509 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 379 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 377 us MUL_MAT_ID(ffn_moe_down-67): 196 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 380 us MUL_MAT_ID(ffn_moe_down-68): 205 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 380 us MUL_MAT_ID(ffn_moe_down-69): 402 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 400 us MUL_MAT_ID(ffn_moe_down-70): 184 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 346 us MUL_MAT_ID(ffn_moe_down-71): 189 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 399 us MUL_MAT_ID(ffn_moe_down-72): 201 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 368 us MUL_MAT_ID(ffn_moe_down-73): 523 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 352 us MUL_MAT_ID(ffn_moe_down-74): 300 us ggml_barrier(...): 109 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 389 us MUL_MAT_ID(ffn_moe_down-75): 204 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 408 us MUL_MAT_ID(ffn_moe_down-76): 200 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 374 us MUL_MAT_ID(ffn_moe_down-77): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 351 us MUL_MAT_ID(ffn_moe_down-78): 213 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 365 us MUL_MAT_ID(ffn_moe_down-79): 210 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 347 us MUL_MAT_ID(ffn_moe_down-80): 201 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 463 us MUL_MAT_ID(ffn_moe_down-81): 219 us ggml_barrier(...): 187 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 371 us MUL_MAT_ID(ffn_moe_down-82): 315 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 392 us MUL_MAT_ID(ffn_moe_down-83): 206 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 362 us MUL_MAT_ID(ffn_moe_down-84): 491 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 390 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 347 us MUL_MAT_ID(ffn_moe_down-86): 196 us ggml_barrier(...): 216 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 375 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 381 us MUL_MAT_ID(ffn_moe_down-88): 197 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 385 us MUL_MAT_ID(ffn_moe_down-89): 201 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 385 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 416 us MUL_MAT_ID(ffn_moe_down-91): 245 us ggml_barrier(...): 7 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 397 us MUL_MAT_ID(ffn_moe_down-25): 467 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 382 us MUL_MAT_ID(ffn_moe_down-26): 376 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 373 us MUL_MAT_ID(ffn_moe_down-27): 182 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 382 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 381 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 360 us MUL_MAT_ID(ffn_moe_down-30): 195 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 399 us MUL_MAT_ID(ffn_moe_down-31): 217 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 384 us MUL_MAT_ID(ffn_moe_down-32): 191 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 372 us MUL_MAT_ID(ffn_moe_down-33): 211 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 374 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 376 us MUL_MAT_ID(ffn_moe_down-35): 333 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 396 us MUL_MAT_ID(ffn_moe_down-36): 220 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 369 us MUL_MAT_ID(ffn_moe_down-37): 245 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 402 us MUL_MAT_ID(ffn_moe_down-38): 189 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 350 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 390 us MUL_MAT_ID(ffn_moe_down-40): 464 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 369 us MUL_MAT_ID(ffn_moe_down-41): 221 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 394 us MUL_MAT_ID(ffn_moe_down-42): 206 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 397 us MUL_MAT_ID(ffn_moe_down-43): 207 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 358 us MUL_MAT_ID(ffn_moe_down-44): 224 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 373 us MUL_MAT_ID(ffn_moe_down-45): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 359 us MUL_MAT_ID(ffn_moe_down-46): 199 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 373 us MUL_MAT_ID(ffn_moe_down-47): 196 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 368 us MUL_MAT_ID(ffn_moe_down-48): 419 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 365 us MUL_MAT_ID(ffn_moe_down-49): 210 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 350 us MUL_MAT_ID(ffn_moe_down-50): 187 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 408 us MUL_MAT_ID(ffn_moe_down-51): 207 us ggml_barrier(...): 4 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 390 us MUL_MAT_ID(ffn_moe_down-52): 207 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 416 us MUL_MAT_ID(ffn_moe_down-53): 407 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 344 us MUL_MAT_ID(ffn_moe_down-54): 183 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 357 us MUL_MAT_ID(ffn_moe_down-55): 194 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 366 us MUL_MAT_ID(ffn_moe_down-56): 299 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 376 us MUL_MAT_ID(ffn_moe_down-57): 199 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 347 us MUL_MAT_ID(ffn_moe_down-58): 203 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 364 us MUL_MAT_ID(ffn_moe_down-59): 201 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 421 us MUL_MAT_ID(ffn_moe_down-60): 297 us ggml_barrier(...): 155 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 401 us MUL_MAT_ID(ffn_moe_down-61): 200 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 375 us MUL_MAT_ID(ffn_moe_down-62): 378 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 390 us MUL_MAT_ID(ffn_moe_down-63): 387 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 395 us MUL_MAT_ID(ffn_moe_down-64): 399 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 430 us MUL_MAT_ID(ffn_moe_down-65): 232 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 403 us MUL_MAT_ID(ffn_moe_down-66): 203 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 407 us MUL_MAT_ID(ffn_moe_down-67): 280 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 348 us MUL_MAT_ID(ffn_moe_down-68): 412 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 394 us MUL_MAT_ID(ffn_moe_down-69): 186 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 378 us MUL_MAT_ID(ffn_moe_down-70): 258 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 368 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 362 us MUL_MAT_ID(ffn_moe_down-72): 189 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 395 us MUL_MAT_ID(ffn_moe_down-73): 200 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 374 us MUL_MAT_ID(ffn_moe_down-74): 195 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 355 us MUL_MAT_ID(ffn_moe_down-75): 190 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 346 us MUL_MAT_ID(ffn_moe_down-76): 234 us ggml_barrier(...): 244 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 397 us MUL_MAT_ID(ffn_moe_down-77): 187 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 370 us MUL_MAT_ID(ffn_moe_down-78): 188 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 405 us MUL_MAT_ID(ffn_moe_down-79): 198 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 426 us MUL_MAT_ID(ffn_moe_down-80): 526 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 351 us MUL_MAT_ID(ffn_moe_down-81): 400 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 409 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 389 us MUL_MAT_ID(ffn_moe_down-83): 193 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 370 us MUL_MAT_ID(ffn_moe_down-84): 231 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 370 us MUL_MAT_ID(ffn_moe_down-85): 631 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 400 us MUL_MAT_ID(ffn_moe_down-86): 205 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 350 us MUL_MAT_ID(ffn_moe_down-87): 183 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 352 us MUL_MAT_ID(ffn_moe_down-88): 386 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 369 us MUL_MAT_ID(ffn_moe_down-89): 230 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 355 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 394 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 22 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 375 us MUL_MAT_ID(ffn_moe_down-25): 238 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 400 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 396 us MUL_MAT_ID(ffn_moe_down-27): 194 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 404 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 362 us MUL_MAT_ID(ffn_moe_down-29): 206 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 379 us MUL_MAT_ID(ffn_moe_down-30): 202 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 385 us MUL_MAT_ID(ffn_moe_down-31): 234 us ggml_barrier(...): 110 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 464 us MUL_MAT_ID(ffn_moe_down-32): 213 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 386 us MUL_MAT_ID(ffn_moe_down-33): 212 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 368 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 426 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 385 us MUL_MAT_ID(ffn_moe_down-36): 377 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 390 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 362 us MUL_MAT_ID(ffn_moe_down-38): 180 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 470 us MUL_MAT_ID(ffn_moe_down-39): 302 us ggml_barrier(...): 132 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 363 us MUL_MAT_ID(ffn_moe_down-40): 247 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 366 us MUL_MAT_ID(ffn_moe_down-41): 291 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 355 us MUL_MAT_ID(ffn_moe_down-42): 355 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 375 us MUL_MAT_ID(ffn_moe_down-43): 233 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 390 us MUL_MAT_ID(ffn_moe_down-44): 244 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 386 us MUL_MAT_ID(ffn_moe_down-45): 199 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 391 us MUL_MAT_ID(ffn_moe_down-46): 192 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 386 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 384 us MUL_MAT_ID(ffn_moe_down-48): 186 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 380 us MUL_MAT_ID(ffn_moe_down-49): 192 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 391 us MUL_MAT_ID(ffn_moe_down-50): 206 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 383 us MUL_MAT_ID(ffn_moe_down-51): 196 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 401 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 364 us MUL_MAT_ID(ffn_moe_down-53): 184 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 356 us MUL_MAT_ID(ffn_moe_down-54): 191 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 364 us MUL_MAT_ID(ffn_moe_down-55): 197 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 377 us MUL_MAT_ID(ffn_moe_down-56): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 341 us MUL_MAT_ID(ffn_moe_down-57): 205 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 388 us MUL_MAT_ID(ffn_moe_down-58): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 362 us MUL_MAT_ID(ffn_moe_down-59): 461 us ggml_barrier(...): 207 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 368 us MUL_MAT_ID(ffn_moe_down-60): 189 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 367 us MUL_MAT_ID(ffn_moe_down-61): 189 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 402 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 404 us MUL_MAT_ID(ffn_moe_down-63): 203 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 391 us MUL_MAT_ID(ffn_moe_down-64): 507 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 372 us MUL_MAT_ID(ffn_moe_down-65): 213 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 379 us MUL_MAT_ID(ffn_moe_down-66): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 366 us MUL_MAT_ID(ffn_moe_down-67): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 348 us MUL_MAT_ID(ffn_moe_down-68): 187 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 371 us MUL_MAT_ID(ffn_moe_down-69): 383 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 362 us MUL_MAT_ID(ffn_moe_down-70): 258 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 376 us MUL_MAT_ID(ffn_moe_down-71): 381 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 367 us MUL_MAT_ID(ffn_moe_down-72): 188 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 481 us MUL_MAT_ID(ffn_moe_down-73): 295 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 366 us MUL_MAT_ID(ffn_moe_down-74): 194 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 365 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 389 us MUL_MAT_ID(ffn_moe_down-76): 173 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 362 us MUL_MAT_ID(ffn_moe_down-77): 195 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 393 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 402 us MUL_MAT_ID(ffn_moe_down-79): 240 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 381 us MUL_MAT_ID(ffn_moe_down-80): 228 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 380 us MUL_MAT_ID(ffn_moe_down-81): 240 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 372 us MUL_MAT_ID(ffn_moe_down-82): 189 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 380 us MUL_MAT_ID(ffn_moe_down-83): 303 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 360 us MUL_MAT_ID(ffn_moe_down-84): 225 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 371 us MUL_MAT_ID(ffn_moe_down-85): 568 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 370 us MUL_MAT_ID(ffn_moe_down-86): 270 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 377 us MUL_MAT_ID(ffn_moe_down-87): 199 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 382 us MUL_MAT_ID(ffn_moe_down-88): 201 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 366 us MUL_MAT_ID(ffn_moe_down-89): 216 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 405 us MUL_MAT_ID(ffn_moe_down-90): 244 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 380 us MUL_MAT_ID(ffn_moe_down-91): 647 us ggml_barrier(...): 70 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 384 us MUL_MAT_ID(ffn_moe_down-25): 203 us ggml_barrier(...): 490 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 399 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 385 us MUL_MAT_ID(ffn_moe_down-27): 306 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 388 us MUL_MAT_ID(ffn_moe_down-28): 191 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 418 us MUL_MAT_ID(ffn_moe_down-29): 185 us ggml_barrier(...): 235 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 412 us MUL_MAT_ID(ffn_moe_down-30): 239 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 355 us MUL_MAT_ID(ffn_moe_down-31): 203 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 388 us MUL_MAT_ID(ffn_moe_down-32): 209 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 407 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 366 us MUL_MAT_ID(ffn_moe_down-34): 192 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 378 us MUL_MAT_ID(ffn_moe_down-35): 362 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 379 us MUL_MAT_ID(ffn_moe_down-36): 208 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 388 us MUL_MAT_ID(ffn_moe_down-37): 217 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 368 us MUL_MAT_ID(ffn_moe_down-38): 295 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 397 us MUL_MAT_ID(ffn_moe_down-39): 347 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 400 us MUL_MAT_ID(ffn_moe_down-40): 231 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 388 us MUL_MAT_ID(ffn_moe_down-41): 219 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 465 us MUL_MAT_ID(ffn_moe_down-42): 283 us ggml_barrier(...): 113 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 356 us MUL_MAT_ID(ffn_moe_down-43): 206 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 395 us MUL_MAT_ID(ffn_moe_down-44): 271 us ggml_barrier(...): 114 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 382 us MUL_MAT_ID(ffn_moe_down-45): 201 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 353 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 89 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 398 us MUL_MAT_ID(ffn_moe_down-47): 198 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 417 us MUL_MAT_ID(ffn_moe_down-48): 250 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 393 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 411 us MUL_MAT_ID(ffn_moe_down-50): 362 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 422 us MUL_MAT_ID(ffn_moe_down-51): 205 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 414 us MUL_MAT_ID(ffn_moe_down-52): 199 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 375 us MUL_MAT_ID(ffn_moe_down-53): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 410 us MUL_MAT_ID(ffn_moe_down-54): 193 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 376 us MUL_MAT_ID(ffn_moe_down-55): 379 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 406 us MUL_MAT_ID(ffn_moe_down-56): 226 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 365 us MUL_MAT_ID(ffn_moe_down-57): 206 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 399 us MUL_MAT_ID(ffn_moe_down-58): 354 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 418 us MUL_MAT_ID(ffn_moe_down-59): 198 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 421 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 409 us MUL_MAT_ID(ffn_moe_down-61): 214 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 394 us MUL_MAT_ID(ffn_moe_down-62): 210 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 379 us MUL_MAT_ID(ffn_moe_down-63): 197 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 386 us MUL_MAT_ID(ffn_moe_down-64): 217 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 399 us MUL_MAT_ID(ffn_moe_down-65): 542 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 378 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 387 us MUL_MAT_ID(ffn_moe_down-67): 199 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 376 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 402 us MUL_MAT_ID(ffn_moe_down-69): 184 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 355 us MUL_MAT_ID(ffn_moe_down-70): 218 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 366 us MUL_MAT_ID(ffn_moe_down-71): 354 us ggml_barrier(...): 261 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 416 us MUL_MAT_ID(ffn_moe_down-72): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 369 us MUL_MAT_ID(ffn_moe_down-73): 199 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 389 us MUL_MAT_ID(ffn_moe_down-74): 193 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 416 us MUL_MAT_ID(ffn_moe_down-75): 202 us ggml_barrier(...): 482 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 384 us MUL_MAT_ID(ffn_moe_down-76): 192 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 404 us MUL_MAT_ID(ffn_moe_down-77): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 407 us MUL_MAT_ID(ffn_moe_down-78): 205 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 352 us MUL_MAT_ID(ffn_moe_down-79): 375 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 354 us MUL_MAT_ID(ffn_moe_down-80): 201 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 358 us MUL_MAT_ID(ffn_moe_down-81): 194 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 394 us MUL_MAT_ID(ffn_moe_down-82): 211 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 383 us MUL_MAT_ID(ffn_moe_down-83): 202 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 365 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 230 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 381 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 370 us MUL_MAT_ID(ffn_moe_down-86): 195 us ggml_barrier(...): 183 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 354 us MUL_MAT_ID(ffn_moe_down-87): 204 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 212 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 374 us MUL_MAT_ID(ffn_moe_down-89): 321 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 412 us MUL_MAT_ID(ffn_moe_down-90): 248 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 345 us MUL_MAT_ID(ffn_moe_down-91): 715 us ggml_barrier(...): 59 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 435 us MUL_MAT_ID(ffn_moe_down-25): 191 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 384 us MUL_MAT_ID(ffn_moe_down-26): 203 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 392 us MUL_MAT_ID(ffn_moe_down-27): 207 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 382 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 366 us MUL_MAT_ID(ffn_moe_down-29): 381 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 380 us MUL_MAT_ID(ffn_moe_down-30): 222 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 415 us MUL_MAT_ID(ffn_moe_down-31): 184 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 363 us MUL_MAT_ID(ffn_moe_down-32): 195 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 365 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 429 us MUL_MAT_ID(ffn_moe_down-34): 219 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 354 us MUL_MAT_ID(ffn_moe_down-35): 203 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 407 us MUL_MAT_ID(ffn_moe_down-36): 294 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 391 us MUL_MAT_ID(ffn_moe_down-37): 247 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 366 us MUL_MAT_ID(ffn_moe_down-38): 389 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 384 us MUL_MAT_ID(ffn_moe_down-39): 259 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 411 us MUL_MAT_ID(ffn_moe_down-40): 225 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 379 us MUL_MAT_ID(ffn_moe_down-41): 394 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 400 us MUL_MAT_ID(ffn_moe_down-42): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 395 us MUL_MAT_ID(ffn_moe_down-43): 196 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 385 us MUL_MAT_ID(ffn_moe_down-44): 212 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 352 us MUL_MAT_ID(ffn_moe_down-45): 208 us ggml_barrier(...): 184 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 395 us MUL_MAT_ID(ffn_moe_down-46): 204 us ggml_barrier(...): 91 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 392 us MUL_MAT_ID(ffn_moe_down-47): 338 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 420 us MUL_MAT_ID(ffn_moe_down-48): 223 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 360 us MUL_MAT_ID(ffn_moe_down-49): 458 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 379 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 366 us MUL_MAT_ID(ffn_moe_down-51): 194 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 385 us MUL_MAT_ID(ffn_moe_down-52): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 430 us MUL_MAT_ID(ffn_moe_down-53): 188 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 403 us MUL_MAT_ID(ffn_moe_down-54): 213 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 376 us MUL_MAT_ID(ffn_moe_down-55): 370 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 409 us MUL_MAT_ID(ffn_moe_down-56): 187 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 355 us MUL_MAT_ID(ffn_moe_down-57): 221 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 405 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 375 us MUL_MAT_ID(ffn_moe_down-59): 190 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 375 us MUL_MAT_ID(ffn_moe_down-60): 192 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 412 us MUL_MAT_ID(ffn_moe_down-61): 198 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 425 us MUL_MAT_ID(ffn_moe_down-62): 268 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 406 us MUL_MAT_ID(ffn_moe_down-63): 533 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 356 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 399 us MUL_MAT_ID(ffn_moe_down-65): 189 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 408 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 400 us MUL_MAT_ID(ffn_moe_down-67): 195 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 379 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 387 us MUL_MAT_ID(ffn_moe_down-69): 180 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 345 us MUL_MAT_ID(ffn_moe_down-70): 184 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 361 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 388 us MUL_MAT_ID(ffn_moe_down-72): 207 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 429 us MUL_MAT_ID(ffn_moe_down-73): 205 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 412 us MUL_MAT_ID(ffn_moe_down-74): 517 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 336 us MUL_MAT_ID(ffn_moe_down-75): 206 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 357 us MUL_MAT_ID(ffn_moe_down-76): 376 us ggml_barrier(...): 161 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 369 us MUL_MAT_ID(ffn_moe_down-77): 231 us ggml_barrier(...): 139 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 417 us MUL_MAT_ID(ffn_moe_down-78): 296 us ggml_barrier(...): 104 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 356 us MUL_MAT_ID(ffn_moe_down-79): 373 us ggml_barrier(...): 243 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 346 us MUL_MAT_ID(ffn_moe_down-80): 229 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 375 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 386 us MUL_MAT_ID(ffn_moe_down-82): 184 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 386 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 378 us MUL_MAT_ID(ffn_moe_down-84): 229 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 388 us MUL_MAT_ID(ffn_moe_down-85): 231 us ggml_barrier(...): 164 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 356 us MUL_MAT_ID(ffn_moe_down-86): 192 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 401 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 414 us MUL_MAT_ID(ffn_moe_down-88): 225 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 386 us MUL_MAT_ID(ffn_moe_down-89): 336 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 402 us MUL_MAT_ID(ffn_moe_down-90): 232 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 407 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 7 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 381 us MUL_MAT_ID(ffn_moe_down-25): 194 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 389 us MUL_MAT_ID(ffn_moe_down-26): 209 us ggml_barrier(...): 257 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 371 us MUL_MAT_ID(ffn_moe_down-27): 208 us ggml_barrier(...): 255 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 369 us MUL_MAT_ID(ffn_moe_down-28): 221 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 392 us MUL_MAT_ID(ffn_moe_down-29): 191 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 376 us MUL_MAT_ID(ffn_moe_down-30): 202 us ggml_barrier(...): 98 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 340 us MUL_MAT_ID(ffn_moe_down-31): 189 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 385 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 363 us MUL_MAT_ID(ffn_moe_down-33): 204 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 384 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 359 us MUL_MAT_ID(ffn_moe_down-35): 195 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 367 us MUL_MAT_ID(ffn_moe_down-36): 191 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 362 us MUL_MAT_ID(ffn_moe_down-37): 246 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 375 us MUL_MAT_ID(ffn_moe_down-38): 198 us ggml_barrier(...): 124 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 376 us MUL_MAT_ID(ffn_moe_down-39): 529 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 367 us MUL_MAT_ID(ffn_moe_down-40): 578 us ggml_barrier(...): 370 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 391 us MUL_MAT_ID(ffn_moe_down-41): 518 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 405 us MUL_MAT_ID(ffn_moe_down-42): 281 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 405 us MUL_MAT_ID(ffn_moe_down-43): 212 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 345 us MUL_MAT_ID(ffn_moe_down-44): 217 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 350 us MUL_MAT_ID(ffn_moe_down-45): 324 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 369 us MUL_MAT_ID(ffn_moe_down-46): 212 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 400 us MUL_MAT_ID(ffn_moe_down-47): 187 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 352 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 373 us MUL_MAT_ID(ffn_moe_down-49): 369 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 380 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 389 us MUL_MAT_ID(ffn_moe_down-51): 194 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 371 us MUL_MAT_ID(ffn_moe_down-52): 204 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 413 us MUL_MAT_ID(ffn_moe_down-53): 219 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 379 us MUL_MAT_ID(ffn_moe_down-54): 183 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 400 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 359 us MUL_MAT_ID(ffn_moe_down-56): 200 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 417 us MUL_MAT_ID(ffn_moe_down-57): 266 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 407 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 411 us MUL_MAT_ID(ffn_moe_down-59): 210 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 384 us MUL_MAT_ID(ffn_moe_down-60): 217 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 382 us MUL_MAT_ID(ffn_moe_down-61): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 363 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 403 us MUL_MAT_ID(ffn_moe_down-63): 199 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 357 us MUL_MAT_ID(ffn_moe_down-64): 204 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 372 us MUL_MAT_ID(ffn_moe_down-65): 329 us ggml_barrier(...): 179 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 431 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 399 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 373 us MUL_MAT_ID(ffn_moe_down-68): 246 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 356 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 409 us MUL_MAT_ID(ffn_moe_down-70): 203 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 366 us MUL_MAT_ID(ffn_moe_down-71): 371 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 403 us MUL_MAT_ID(ffn_moe_down-72): 201 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 380 us MUL_MAT_ID(ffn_moe_down-73): 217 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 343 us MUL_MAT_ID(ffn_moe_down-74): 249 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 422 us MUL_MAT_ID(ffn_moe_down-75): 197 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 402 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 361 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 393 us MUL_MAT_ID(ffn_moe_down-78): 182 us ggml_barrier(...): 309 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 387 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 373 us MUL_MAT_ID(ffn_moe_down-80): 195 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 366 us MUL_MAT_ID(ffn_moe_down-81): 192 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 405 us MUL_MAT_ID(ffn_moe_down-82): 196 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 392 us MUL_MAT_ID(ffn_moe_down-83): 199 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 381 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 377 us MUL_MAT_ID(ffn_moe_down-85): 234 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 377 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 369 us MUL_MAT_ID(ffn_moe_down-87): 201 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 404 us MUL_MAT_ID(ffn_moe_down-88): 365 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 373 us MUL_MAT_ID(ffn_moe_down-89): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 400 us MUL_MAT_ID(ffn_moe_down-90): 226 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 392 us MUL_MAT_ID(ffn_moe_down-91): 262 us ggml_barrier(...): 13 us GET_ROWS(inp_embd): 20 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 434 us MUL_MAT_ID(ffn_moe_down-25): 376 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 409 us MUL_MAT_ID(ffn_moe_down-26): 187 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 397 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 348 us MUL_MAT_ID(ffn_moe_down-28): 494 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 195 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 485 us MUL_MAT_ID(ffn_moe_down-30): 196 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 392 us MUL_MAT_ID(ffn_moe_down-31): 200 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 395 us MUL_MAT_ID(ffn_moe_down-32): 490 us ggml_barrier(...): 146 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 373 us MUL_MAT_ID(ffn_moe_down-33): 196 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 386 us MUL_MAT_ID(ffn_moe_down-34): 191 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 387 us MUL_MAT_ID(ffn_moe_down-35): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 440 us MUL_MAT_ID(ffn_moe_down-36): 295 us ggml_barrier(...): 135 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 388 us MUL_MAT_ID(ffn_moe_down-37): 243 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 384 us MUL_MAT_ID(ffn_moe_down-38): 203 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 396 us MUL_MAT_ID(ffn_moe_down-39): 546 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 406 us MUL_MAT_ID(ffn_moe_down-40): 375 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 384 us MUL_MAT_ID(ffn_moe_down-41): 205 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 380 us MUL_MAT_ID(ffn_moe_down-42): 488 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 381 us MUL_MAT_ID(ffn_moe_down-43): 196 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 365 us MUL_MAT_ID(ffn_moe_down-44): 200 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 377 us MUL_MAT_ID(ffn_moe_down-45): 537 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 356 us MUL_MAT_ID(ffn_moe_down-46): 193 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 364 us MUL_MAT_ID(ffn_moe_down-47): 200 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 373 us MUL_MAT_ID(ffn_moe_down-48): 204 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 358 us MUL_MAT_ID(ffn_moe_down-49): 197 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 377 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 357 us MUL_MAT_ID(ffn_moe_down-51): 202 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 360 us MUL_MAT_ID(ffn_moe_down-52): 189 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 389 us MUL_MAT_ID(ffn_moe_down-53): 181 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 394 us MUL_MAT_ID(ffn_moe_down-54): 187 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 404 us MUL_MAT_ID(ffn_moe_down-55): 493 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 396 us MUL_MAT_ID(ffn_moe_down-56): 376 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 409 us MUL_MAT_ID(ffn_moe_down-57): 334 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 410 us MUL_MAT_ID(ffn_moe_down-58): 194 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 421 us MUL_MAT_ID(ffn_moe_down-59): 212 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 362 us MUL_MAT_ID(ffn_moe_down-60): 231 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 391 us MUL_MAT_ID(ffn_moe_down-61): 334 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 368 us MUL_MAT_ID(ffn_moe_down-62): 202 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 410 us MUL_MAT_ID(ffn_moe_down-63): 266 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 380 us MUL_MAT_ID(ffn_moe_down-64): 518 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 391 us MUL_MAT_ID(ffn_moe_down-65): 221 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 396 us MUL_MAT_ID(ffn_moe_down-66): 202 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 392 us MUL_MAT_ID(ffn_moe_down-67): 229 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 404 us MUL_MAT_ID(ffn_moe_down-68): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 361 us MUL_MAT_ID(ffn_moe_down-69): 398 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 399 us MUL_MAT_ID(ffn_moe_down-70): 208 us ggml_barrier(...): 304 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 424 us MUL_MAT_ID(ffn_moe_down-71): 190 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 375 us MUL_MAT_ID(ffn_moe_down-72): 205 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 386 us MUL_MAT_ID(ffn_moe_down-73): 204 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 362 us MUL_MAT_ID(ffn_moe_down-74): 214 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 359 us MUL_MAT_ID(ffn_moe_down-75): 200 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 374 us MUL_MAT_ID(ffn_moe_down-76): 226 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 434 us MUL_MAT_ID(ffn_moe_down-77): 211 us ggml_barrier(...): 154 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 405 us MUL_MAT_ID(ffn_moe_down-78): 273 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 374 us MUL_MAT_ID(ffn_moe_down-79): 203 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 392 us MUL_MAT_ID(ffn_moe_down-80): 196 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 359 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 370 us MUL_MAT_ID(ffn_moe_down-82): 200 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 379 us MUL_MAT_ID(ffn_moe_down-83): 207 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 356 us MUL_MAT_ID(ffn_moe_down-84): 249 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 392 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 356 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 413 us MUL_MAT_ID(ffn_moe_down-87): 250 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 357 us MUL_MAT_ID(ffn_moe_down-88): 192 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 364 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 345 us MUL_MAT_ID(ffn_moe_down-90): 248 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 368 us MUL_MAT_ID(ffn_moe_down-91): 744 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 390 us MUL_MAT_ID(ffn_moe_down-25): 205 us ggml_barrier(...): 189 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 403 us MUL_MAT_ID(ffn_moe_down-26): 526 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 375 us MUL_MAT_ID(ffn_moe_down-27): 200 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 362 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 385 us MUL_MAT_ID(ffn_moe_down-29): 298 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 370 us MUL_MAT_ID(ffn_moe_down-30): 312 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 377 us MUL_MAT_ID(ffn_moe_down-31): 204 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 374 us MUL_MAT_ID(ffn_moe_down-32): 198 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 410 us MUL_MAT_ID(ffn_moe_down-33): 205 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 396 us MUL_MAT_ID(ffn_moe_down-34): 195 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 372 us MUL_MAT_ID(ffn_moe_down-35): 192 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 397 us MUL_MAT_ID(ffn_moe_down-36): 404 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 345 us MUL_MAT_ID(ffn_moe_down-37): 229 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 392 us MUL_MAT_ID(ffn_moe_down-38): 191 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 381 us MUL_MAT_ID(ffn_moe_down-39): 256 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 409 us MUL_MAT_ID(ffn_moe_down-40): 229 us ggml_barrier(...): 288 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 371 us MUL_MAT_ID(ffn_moe_down-41): 209 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 336 us MUL_MAT_ID(ffn_moe_down-42): 184 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 380 us MUL_MAT_ID(ffn_moe_down-43): 213 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 362 us MUL_MAT_ID(ffn_moe_down-44): 197 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 434 us MUL_MAT_ID(ffn_moe_down-45): 203 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 346 us MUL_MAT_ID(ffn_moe_down-46): 204 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 373 us MUL_MAT_ID(ffn_moe_down-47): 205 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 377 us MUL_MAT_ID(ffn_moe_down-48): 351 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 420 us MUL_MAT_ID(ffn_moe_down-49): 236 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 376 us MUL_MAT_ID(ffn_moe_down-50): 190 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 377 us MUL_MAT_ID(ffn_moe_down-51): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 371 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 382 us MUL_MAT_ID(ffn_moe_down-53): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 377 us MUL_MAT_ID(ffn_moe_down-54): 242 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 389 us MUL_MAT_ID(ffn_moe_down-55): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 365 us MUL_MAT_ID(ffn_moe_down-56): 184 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 402 us MUL_MAT_ID(ffn_moe_down-57): 200 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 405 us MUL_MAT_ID(ffn_moe_down-58): 182 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 370 us MUL_MAT_ID(ffn_moe_down-59): 192 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 373 us MUL_MAT_ID(ffn_moe_down-60): 199 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 362 us MUL_MAT_ID(ffn_moe_down-61): 187 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 398 us MUL_MAT_ID(ffn_moe_down-62): 216 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 377 us MUL_MAT_ID(ffn_moe_down-63): 195 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 398 us MUL_MAT_ID(ffn_moe_down-64): 191 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 382 us MUL_MAT_ID(ffn_moe_down-65): 211 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 369 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 390 us MUL_MAT_ID(ffn_moe_down-67): 198 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 372 us MUL_MAT_ID(ffn_moe_down-68): 203 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 419 us MUL_MAT_ID(ffn_moe_down-69): 239 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 398 us MUL_MAT_ID(ffn_moe_down-70): 201 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 421 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 478 us MUL_MAT_ID(ffn_moe_down-72): 198 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 389 us MUL_MAT_ID(ffn_moe_down-73): 200 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 393 us MUL_MAT_ID(ffn_moe_down-74): 212 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 417 us MUL_MAT_ID(ffn_moe_down-75): 202 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 395 us MUL_MAT_ID(ffn_moe_down-76): 386 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 192 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 385 us MUL_MAT_ID(ffn_moe_down-78): 193 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 402 us MUL_MAT_ID(ffn_moe_down-79): 204 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 389 us MUL_MAT_ID(ffn_moe_down-80): 219 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 368 us MUL_MAT_ID(ffn_moe_down-81): 305 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 385 us MUL_MAT_ID(ffn_moe_down-82): 199 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 361 us MUL_MAT_ID(ffn_moe_down-83): 345 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 394 us MUL_MAT_ID(ffn_moe_down-84): 236 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 391 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 409 us MUL_MAT_ID(ffn_moe_down-86): 336 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 362 us MUL_MAT_ID(ffn_moe_down-87): 202 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 384 us MUL_MAT_ID(ffn_moe_down-88): 380 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 395 us MUL_MAT_ID(ffn_moe_down-89): 208 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 395 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 347 us MUL_MAT_ID(ffn_moe_down-91): 266 us ggml_barrier(...): 62 us GET_ROWS(inp_embd): 27 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 422 us MUL_MAT_ID(ffn_moe_down-25): 210 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 362 us MUL_MAT_ID(ffn_moe_down-26): 405 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 358 us MUL_MAT_ID(ffn_moe_down-27): 216 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 388 us MUL_MAT_ID(ffn_moe_down-28): 205 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 539 us MUL_MAT_ID(ffn_moe_down-29): 205 us ggml_barrier(...): 156 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 377 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 409 us MUL_MAT_ID(ffn_moe_down-31): 186 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 368 us MUL_MAT_ID(ffn_moe_down-32): 194 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 409 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 424 us MUL_MAT_ID(ffn_moe_down-34): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 386 us MUL_MAT_ID(ffn_moe_down-35): 193 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 399 us MUL_MAT_ID(ffn_moe_down-36): 194 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 418 us MUL_MAT_ID(ffn_moe_down-37): 225 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 373 us MUL_MAT_ID(ffn_moe_down-38): 201 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 373 us MUL_MAT_ID(ffn_moe_down-39): 230 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 375 us MUL_MAT_ID(ffn_moe_down-40): 223 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 386 us MUL_MAT_ID(ffn_moe_down-41): 387 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 378 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 375 us MUL_MAT_ID(ffn_moe_down-43): 185 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 344 us MUL_MAT_ID(ffn_moe_down-44): 194 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 372 us MUL_MAT_ID(ffn_moe_down-45): 194 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 353 us MUL_MAT_ID(ffn_moe_down-46): 194 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 384 us MUL_MAT_ID(ffn_moe_down-47): 233 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 383 us MUL_MAT_ID(ffn_moe_down-48): 209 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 366 us MUL_MAT_ID(ffn_moe_down-49): 234 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 375 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 406 us MUL_MAT_ID(ffn_moe_down-51): 214 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 369 us MUL_MAT_ID(ffn_moe_down-52): 205 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 403 us MUL_MAT_ID(ffn_moe_down-53): 185 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 353 us MUL_MAT_ID(ffn_moe_down-54): 185 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 345 us MUL_MAT_ID(ffn_moe_down-55): 271 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 407 us MUL_MAT_ID(ffn_moe_down-56): 184 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 385 us MUL_MAT_ID(ffn_moe_down-57): 195 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 400 us MUL_MAT_ID(ffn_moe_down-58): 198 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 572 us MUL_MAT_ID(ffn_moe_down-59): 225 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 374 us MUL_MAT_ID(ffn_moe_down-60): 211 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 395 us MUL_MAT_ID(ffn_moe_down-61): 207 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 369 us MUL_MAT_ID(ffn_moe_down-62): 200 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 358 us MUL_MAT_ID(ffn_moe_down-63): 398 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 366 us MUL_MAT_ID(ffn_moe_down-64): 431 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 356 us MUL_MAT_ID(ffn_moe_down-65): 215 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 386 us MUL_MAT_ID(ffn_moe_down-66): 223 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 370 us MUL_MAT_ID(ffn_moe_down-67): 189 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 396 us MUL_MAT_ID(ffn_moe_down-68): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 361 us MUL_MAT_ID(ffn_moe_down-69): 196 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 394 us MUL_MAT_ID(ffn_moe_down-70): 335 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 388 us MUL_MAT_ID(ffn_moe_down-71): 204 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 374 us MUL_MAT_ID(ffn_moe_down-72): 210 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 380 us MUL_MAT_ID(ffn_moe_down-73): 352 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 355 us MUL_MAT_ID(ffn_moe_down-74): 275 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 391 us MUL_MAT_ID(ffn_moe_down-75): 409 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 327 us MUL_MAT_ID(ffn_moe_down-76): 186 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 370 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 397 us MUL_MAT_ID(ffn_moe_down-78): 200 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 382 us MUL_MAT_ID(ffn_moe_down-79): 196 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 376 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 409 us MUL_MAT_ID(ffn_moe_down-81): 198 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 393 us MUL_MAT_ID(ffn_moe_down-82): 358 us ggml_barrier(...): 254 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 402 us MUL_MAT_ID(ffn_moe_down-83): 226 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 359 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 358 us MUL_MAT_ID(ffn_moe_down-85): 249 us ggml_barrier(...): 76 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 386 us MUL_MAT_ID(ffn_moe_down-86): 204 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 376 us MUL_MAT_ID(ffn_moe_down-87): 213 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 407 us MUL_MAT_ID(ffn_moe_down-88): 204 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 386 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 389 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 372 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 49 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 382 us MUL_MAT_ID(ffn_moe_down-25): 189 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 360 us MUL_MAT_ID(ffn_moe_down-26): 199 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 364 us MUL_MAT_ID(ffn_moe_down-27): 200 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 396 us MUL_MAT_ID(ffn_moe_down-28): 190 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 454 us MUL_MAT_ID(ffn_moe_down-29): 288 us ggml_barrier(...): 127 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 379 us MUL_MAT_ID(ffn_moe_down-30): 205 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 398 us MUL_MAT_ID(ffn_moe_down-31): 288 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 518 us MUL_MAT_ID(ffn_moe_down-32): 167 us ggml_barrier(...): 208 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 361 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 373 us MUL_MAT_ID(ffn_moe_down-34): 187 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 393 us MUL_MAT_ID(ffn_moe_down-35): 420 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 369 us MUL_MAT_ID(ffn_moe_down-36): 189 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 393 us MUL_MAT_ID(ffn_moe_down-37): 243 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 394 us MUL_MAT_ID(ffn_moe_down-38): 224 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 348 us MUL_MAT_ID(ffn_moe_down-39): 240 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 394 us MUL_MAT_ID(ffn_moe_down-40): 227 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 391 us MUL_MAT_ID(ffn_moe_down-41): 193 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 384 us MUL_MAT_ID(ffn_moe_down-42): 198 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 384 us MUL_MAT_ID(ffn_moe_down-43): 206 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 410 us MUL_MAT_ID(ffn_moe_down-44): 193 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 437 us MUL_MAT_ID(ffn_moe_down-45): 200 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 367 us MUL_MAT_ID(ffn_moe_down-46): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 368 us MUL_MAT_ID(ffn_moe_down-47): 195 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 405 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 434 us MUL_MAT_ID(ffn_moe_down-49): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 383 us MUL_MAT_ID(ffn_moe_down-50): 202 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 390 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 367 us MUL_MAT_ID(ffn_moe_down-52): 188 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 359 us MUL_MAT_ID(ffn_moe_down-53): 409 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 356 us MUL_MAT_ID(ffn_moe_down-54): 187 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 380 us MUL_MAT_ID(ffn_moe_down-55): 196 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 413 us MUL_MAT_ID(ffn_moe_down-56): 187 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 402 us MUL_MAT_ID(ffn_moe_down-57): 181 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 401 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 385 us MUL_MAT_ID(ffn_moe_down-59): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 364 us MUL_MAT_ID(ffn_moe_down-60): 200 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 389 us MUL_MAT_ID(ffn_moe_down-61): 186 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 365 us MUL_MAT_ID(ffn_moe_down-62): 201 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 367 us MUL_MAT_ID(ffn_moe_down-63): 185 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 356 us MUL_MAT_ID(ffn_moe_down-64): 182 us ggml_barrier(...): 100 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 385 us MUL_MAT_ID(ffn_moe_down-65): 194 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 363 us MUL_MAT_ID(ffn_moe_down-66): 188 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 415 us MUL_MAT_ID(ffn_moe_down-67): 213 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 395 us MUL_MAT_ID(ffn_moe_down-68): 190 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 365 us MUL_MAT_ID(ffn_moe_down-69): 199 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 391 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 361 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 389 us MUL_MAT_ID(ffn_moe_down-72): 186 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 394 us MUL_MAT_ID(ffn_moe_down-73): 186 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 395 us MUL_MAT_ID(ffn_moe_down-74): 256 us ggml_barrier(...): 277 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 375 us MUL_MAT_ID(ffn_moe_down-75): 182 us ggml_barrier(...): 186 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 408 us MUL_MAT_ID(ffn_moe_down-76): 178 us ggml_barrier(...): 273 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 396 us MUL_MAT_ID(ffn_moe_down-77): 225 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 381 us MUL_MAT_ID(ffn_moe_down-78): 191 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 392 us MUL_MAT_ID(ffn_moe_down-79): 199 us ggml_barrier(...): 112 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 355 us MUL_MAT_ID(ffn_moe_down-80): 202 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 410 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 368 us MUL_MAT_ID(ffn_moe_down-82): 184 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 377 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 400 us MUL_MAT_ID(ffn_moe_down-84): 240 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 391 us MUL_MAT_ID(ffn_moe_down-85): 243 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 395 us MUL_MAT_ID(ffn_moe_down-86): 199 us ggml_barrier(...): 160 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 358 us MUL_MAT_ID(ffn_moe_down-87): 195 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 379 us MUL_MAT_ID(ffn_moe_down-88): 283 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 367 us MUL_MAT_ID(ffn_moe_down-89): 198 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 385 us MUL_MAT_ID(ffn_moe_down-90): 234 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 381 us MUL_MAT_ID(ffn_moe_down-91): 248 us ggml_barrier(...): 28 us GET_ROWS(inp_embd): 5 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 391 us MUL_MAT_ID(ffn_moe_down-25): 179 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 383 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 394 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 352 us MUL_MAT_ID(ffn_moe_down-28): 193 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 382 us MUL_MAT_ID(ffn_moe_down-29): 190 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 361 us MUL_MAT_ID(ffn_moe_down-30): 191 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 371 us MUL_MAT_ID(ffn_moe_down-31): 303 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 388 us MUL_MAT_ID(ffn_moe_down-32): 204 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 385 us MUL_MAT_ID(ffn_moe_down-33): 200 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 376 us MUL_MAT_ID(ffn_moe_down-34): 200 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 367 us MUL_MAT_ID(ffn_moe_down-35): 190 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 397 us MUL_MAT_ID(ffn_moe_down-36): 207 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 392 us MUL_MAT_ID(ffn_moe_down-37): 236 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 487 us MUL_MAT_ID(ffn_moe_down-38): 197 us ggml_barrier(...): 221 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 371 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 402 us MUL_MAT_ID(ffn_moe_down-40): 244 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 413 us MUL_MAT_ID(ffn_moe_down-41): 341 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 368 us MUL_MAT_ID(ffn_moe_down-42): 183 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 370 us MUL_MAT_ID(ffn_moe_down-43): 378 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 344 us MUL_MAT_ID(ffn_moe_down-44): 188 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 398 us MUL_MAT_ID(ffn_moe_down-45): 223 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 349 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 403 us MUL_MAT_ID(ffn_moe_down-47): 230 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 374 us MUL_MAT_ID(ffn_moe_down-48): 193 us ggml_barrier(...): 126 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 373 us MUL_MAT_ID(ffn_moe_down-49): 246 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 375 us MUL_MAT_ID(ffn_moe_down-50): 191 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 395 us MUL_MAT_ID(ffn_moe_down-51): 188 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 362 us MUL_MAT_ID(ffn_moe_down-52): 209 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 361 us MUL_MAT_ID(ffn_moe_down-53): 257 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 395 us MUL_MAT_ID(ffn_moe_down-54): 233 us ggml_barrier(...): 272 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 409 us MUL_MAT_ID(ffn_moe_down-55): 188 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 370 us MUL_MAT_ID(ffn_moe_down-56): 283 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 369 us MUL_MAT_ID(ffn_moe_down-57): 357 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 351 us MUL_MAT_ID(ffn_moe_down-58): 187 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 359 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 363 us MUL_MAT_ID(ffn_moe_down-60): 210 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 414 us MUL_MAT_ID(ffn_moe_down-61): 201 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 383 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 338 us MUL_MAT_ID(ffn_moe_down-63): 189 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 354 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 403 us MUL_MAT_ID(ffn_moe_down-65): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 359 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 351 us MUL_MAT_ID(ffn_moe_down-67): 193 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 339 us MUL_MAT_ID(ffn_moe_down-68): 199 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 329 us MUL_MAT_ID(ffn_moe_down-69): 210 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 356 us MUL_MAT_ID(ffn_moe_down-70): 191 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 390 us MUL_MAT_ID(ffn_moe_down-71): 200 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 367 us MUL_MAT_ID(ffn_moe_down-72): 205 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 378 us MUL_MAT_ID(ffn_moe_down-73): 339 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 404 us MUL_MAT_ID(ffn_moe_down-74): 217 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 377 us MUL_MAT_ID(ffn_moe_down-75): 270 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 350 us MUL_MAT_ID(ffn_moe_down-76): 261 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 351 us MUL_MAT_ID(ffn_moe_down-77): 190 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 379 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 400 us MUL_MAT_ID(ffn_moe_down-79): 229 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 388 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 373 us MUL_MAT_ID(ffn_moe_down-81): 305 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 360 us MUL_MAT_ID(ffn_moe_down-82): 219 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 390 us MUL_MAT_ID(ffn_moe_down-83): 193 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 379 us MUL_MAT_ID(ffn_moe_down-84): 244 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 362 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 367 us MUL_MAT_ID(ffn_moe_down-86): 193 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 401 us MUL_MAT_ID(ffn_moe_down-87): 493 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 390 us MUL_MAT_ID(ffn_moe_down-88): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 404 us MUL_MAT_ID(ffn_moe_down-89): 184 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 367 us MUL_MAT_ID(ffn_moe_down-90): 227 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 360 us MUL_MAT_ID(ffn_moe_down-91): 235 us ggml_barrier(...): 67 us GET_ROWS(inp_embd): 16 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 395 us MUL_MAT_ID(ffn_moe_down-25): 177 us ggml_barrier(...): 191 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 388 us MUL_MAT_ID(ffn_moe_down-26): 186 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 369 us MUL_MAT_ID(ffn_moe_down-27): 206 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 361 us MUL_MAT_ID(ffn_moe_down-28): 194 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 368 us MUL_MAT_ID(ffn_moe_down-29): 188 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 366 us MUL_MAT_ID(ffn_moe_down-30): 204 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 399 us MUL_MAT_ID(ffn_moe_down-31): 217 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 400 us MUL_MAT_ID(ffn_moe_down-32): 204 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 382 us MUL_MAT_ID(ffn_moe_down-33): 204 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 394 us MUL_MAT_ID(ffn_moe_down-34): 230 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 413 us MUL_MAT_ID(ffn_moe_down-35): 203 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 374 us MUL_MAT_ID(ffn_moe_down-36): 309 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 371 us MUL_MAT_ID(ffn_moe_down-37): 228 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 371 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 400 us MUL_MAT_ID(ffn_moe_down-39): 234 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 390 us MUL_MAT_ID(ffn_moe_down-40): 262 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 407 us MUL_MAT_ID(ffn_moe_down-41): 197 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 372 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 421 us MUL_MAT_ID(ffn_moe_down-43): 238 us ggml_barrier(...): 266 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 355 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 417 us MUL_MAT_ID(ffn_moe_down-45): 369 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 346 us MUL_MAT_ID(ffn_moe_down-46): 408 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 367 us MUL_MAT_ID(ffn_moe_down-47): 201 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 376 us MUL_MAT_ID(ffn_moe_down-48): 195 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 372 us MUL_MAT_ID(ffn_moe_down-49): 190 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 359 us MUL_MAT_ID(ffn_moe_down-50): 214 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 540 us MUL_MAT_ID(ffn_moe_down-51): 183 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 370 us MUL_MAT_ID(ffn_moe_down-52): 302 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 400 us MUL_MAT_ID(ffn_moe_down-53): 185 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 404 us MUL_MAT_ID(ffn_moe_down-54): 200 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 354 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 389 us MUL_MAT_ID(ffn_moe_down-56): 210 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 367 us MUL_MAT_ID(ffn_moe_down-57): 195 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 366 us MUL_MAT_ID(ffn_moe_down-58): 208 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 370 us MUL_MAT_ID(ffn_moe_down-59): 188 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 383 us MUL_MAT_ID(ffn_moe_down-60): 195 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 388 us MUL_MAT_ID(ffn_moe_down-61): 213 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 368 us MUL_MAT_ID(ffn_moe_down-62): 185 us ggml_barrier(...): 319 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 351 us MUL_MAT_ID(ffn_moe_down-63): 355 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 383 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 379 us MUL_MAT_ID(ffn_moe_down-65): 195 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 383 us MUL_MAT_ID(ffn_moe_down-66): 217 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 382 us MUL_MAT_ID(ffn_moe_down-67): 191 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 443 us MUL_MAT_ID(ffn_moe_down-68): 180 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 392 us MUL_MAT_ID(ffn_moe_down-69): 205 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 353 us MUL_MAT_ID(ffn_moe_down-70): 204 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 395 us MUL_MAT_ID(ffn_moe_down-71): 207 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 383 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 402 us MUL_MAT_ID(ffn_moe_down-73): 190 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 389 us MUL_MAT_ID(ffn_moe_down-74): 202 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 401 us MUL_MAT_ID(ffn_moe_down-75): 183 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 404 us MUL_MAT_ID(ffn_moe_down-76): 208 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 372 us MUL_MAT_ID(ffn_moe_down-77): 227 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 432 us MUL_MAT_ID(ffn_moe_down-78): 202 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 368 us MUL_MAT_ID(ffn_moe_down-79): 278 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 388 us MUL_MAT_ID(ffn_moe_down-80): 207 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 383 us MUL_MAT_ID(ffn_moe_down-81): 195 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 404 us MUL_MAT_ID(ffn_moe_down-82): 182 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 407 us MUL_MAT_ID(ffn_moe_down-83): 195 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 400 us MUL_MAT_ID(ffn_moe_down-84): 233 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 363 us MUL_MAT_ID(ffn_moe_down-85): 230 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 361 us MUL_MAT_ID(ffn_moe_down-86): 436 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 397 us MUL_MAT_ID(ffn_moe_down-87): 196 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 380 us MUL_MAT_ID(ffn_moe_down-88): 539 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 339 us MUL_MAT_ID(ffn_moe_down-89): 380 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 363 us MUL_MAT_ID(ffn_moe_down-90): 222 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 363 us MUL_MAT_ID(ffn_moe_down-91): 234 us ggml_barrier(...): 52 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 1 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 381 us MUL_MAT_ID(ffn_moe_down-25): 188 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 397 us MUL_MAT_ID(ffn_moe_down-26): 188 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 387 us MUL_MAT_ID(ffn_moe_down-27): 193 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 376 us MUL_MAT_ID(ffn_moe_down-28): 182 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 384 us MUL_MAT_ID(ffn_moe_down-29): 189 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 365 us MUL_MAT_ID(ffn_moe_down-30): 197 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 382 us MUL_MAT_ID(ffn_moe_down-31): 216 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 407 us MUL_MAT_ID(ffn_moe_down-32): 196 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 409 us MUL_MAT_ID(ffn_moe_down-33): 384 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 404 us MUL_MAT_ID(ffn_moe_down-34): 221 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 384 us MUL_MAT_ID(ffn_moe_down-35): 335 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 382 us MUL_MAT_ID(ffn_moe_down-36): 195 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 371 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 372 us MUL_MAT_ID(ffn_moe_down-38): 188 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 366 us MUL_MAT_ID(ffn_moe_down-39): 254 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 407 us MUL_MAT_ID(ffn_moe_down-40): 237 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 389 us MUL_MAT_ID(ffn_moe_down-41): 246 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 385 us MUL_MAT_ID(ffn_moe_down-42): 186 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 371 us MUL_MAT_ID(ffn_moe_down-43): 361 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 362 us MUL_MAT_ID(ffn_moe_down-44): 182 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 369 us MUL_MAT_ID(ffn_moe_down-45): 362 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 375 us MUL_MAT_ID(ffn_moe_down-46): 205 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 406 us MUL_MAT_ID(ffn_moe_down-47): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 357 us MUL_MAT_ID(ffn_moe_down-48): 203 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 374 us MUL_MAT_ID(ffn_moe_down-49): 214 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 427 us MUL_MAT_ID(ffn_moe_down-50): 307 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 407 us MUL_MAT_ID(ffn_moe_down-51): 198 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 394 us MUL_MAT_ID(ffn_moe_down-52): 209 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 369 us MUL_MAT_ID(ffn_moe_down-53): 195 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 358 us MUL_MAT_ID(ffn_moe_down-54): 375 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 433 us MUL_MAT_ID(ffn_moe_down-55): 291 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 419 us MUL_MAT_ID(ffn_moe_down-56): 198 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 397 us MUL_MAT_ID(ffn_moe_down-57): 197 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 361 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 379 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 312 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 412 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 388 us MUL_MAT_ID(ffn_moe_down-61): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 352 us MUL_MAT_ID(ffn_moe_down-62): 196 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 198 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 374 us MUL_MAT_ID(ffn_moe_down-64): 208 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 408 us MUL_MAT_ID(ffn_moe_down-65): 237 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 374 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 390 us MUL_MAT_ID(ffn_moe_down-67): 196 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 385 us MUL_MAT_ID(ffn_moe_down-68): 214 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 389 us MUL_MAT_ID(ffn_moe_down-69): 230 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 398 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 422 us MUL_MAT_ID(ffn_moe_down-71): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 348 us MUL_MAT_ID(ffn_moe_down-72): 195 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 381 us MUL_MAT_ID(ffn_moe_down-73): 294 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 362 us MUL_MAT_ID(ffn_moe_down-74): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 391 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 360 us MUL_MAT_ID(ffn_moe_down-76): 224 us ggml_barrier(...): 151 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 387 us MUL_MAT_ID(ffn_moe_down-77): 198 us ggml_barrier(...): 117 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 379 us MUL_MAT_ID(ffn_moe_down-78): 198 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 388 us MUL_MAT_ID(ffn_moe_down-79): 194 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 385 us MUL_MAT_ID(ffn_moe_down-80): 199 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 371 us MUL_MAT_ID(ffn_moe_down-81): 290 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 359 us MUL_MAT_ID(ffn_moe_down-82): 181 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 362 us MUL_MAT_ID(ffn_moe_down-83): 505 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 398 us MUL_MAT_ID(ffn_moe_down-84): 242 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 410 us MUL_MAT_ID(ffn_moe_down-85): 229 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 374 us MUL_MAT_ID(ffn_moe_down-86): 370 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 372 us MUL_MAT_ID(ffn_moe_down-87): 284 us ggml_barrier(...): 290 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 400 us MUL_MAT_ID(ffn_moe_down-88): 209 us ggml_barrier(...): 121 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 404 us MUL_MAT_ID(ffn_moe_down-89): 430 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 378 us MUL_MAT_ID(ffn_moe_down-90): 438 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 374 us MUL_MAT_ID(ffn_moe_down-91): 249 us ggml_barrier(...): 45 us GET_ROWS(inp_embd): 21 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 410 us MUL_MAT_ID(ffn_moe_down-25): 203 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 354 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 395 us MUL_MAT_ID(ffn_moe_down-27): 210 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 355 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 378 us MUL_MAT_ID(ffn_moe_down-29): 197 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 388 us MUL_MAT_ID(ffn_moe_down-30): 188 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 392 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 374 us MUL_MAT_ID(ffn_moe_down-32): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 357 us MUL_MAT_ID(ffn_moe_down-33): 245 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 421 us MUL_MAT_ID(ffn_moe_down-34): 242 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 366 us MUL_MAT_ID(ffn_moe_down-35): 201 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 378 us MUL_MAT_ID(ffn_moe_down-36): 552 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 395 us MUL_MAT_ID(ffn_moe_down-37): 242 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 398 us MUL_MAT_ID(ffn_moe_down-38): 185 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 401 us MUL_MAT_ID(ffn_moe_down-39): 241 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 397 us MUL_MAT_ID(ffn_moe_down-40): 236 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 417 us MUL_MAT_ID(ffn_moe_down-41): 195 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 358 us MUL_MAT_ID(ffn_moe_down-42): 228 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 387 us MUL_MAT_ID(ffn_moe_down-43): 191 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 401 us MUL_MAT_ID(ffn_moe_down-44): 205 us ggml_barrier(...): 158 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 359 us MUL_MAT_ID(ffn_moe_down-45): 209 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 369 us MUL_MAT_ID(ffn_moe_down-46): 473 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 387 us MUL_MAT_ID(ffn_moe_down-47): 193 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 343 us MUL_MAT_ID(ffn_moe_down-48): 189 us ggml_barrier(...): 84 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 405 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 461 us MUL_MAT_ID(ffn_moe_down-50): 273 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 350 us MUL_MAT_ID(ffn_moe_down-51): 329 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 363 us MUL_MAT_ID(ffn_moe_down-52): 192 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 369 us MUL_MAT_ID(ffn_moe_down-53): 203 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 423 us MUL_MAT_ID(ffn_moe_down-54): 194 us ggml_barrier(...): 236 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 392 us MUL_MAT_ID(ffn_moe_down-55): 200 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 407 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 345 us MUL_MAT_ID(ffn_moe_down-57): 188 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 400 us MUL_MAT_ID(ffn_moe_down-58): 199 us ggml_barrier(...): 141 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 386 us MUL_MAT_ID(ffn_moe_down-59): 187 us ggml_barrier(...): 153 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 406 us MUL_MAT_ID(ffn_moe_down-60): 423 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 381 us MUL_MAT_ID(ffn_moe_down-61): 194 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 354 us MUL_MAT_ID(ffn_moe_down-62): 194 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 354 us MUL_MAT_ID(ffn_moe_down-63): 221 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 364 us MUL_MAT_ID(ffn_moe_down-64): 195 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 362 us MUL_MAT_ID(ffn_moe_down-65): 273 us ggml_barrier(...): 337 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 368 us MUL_MAT_ID(ffn_moe_down-66): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 395 us MUL_MAT_ID(ffn_moe_down-67): 202 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 328 us MUL_MAT_ID(ffn_moe_down-68): 223 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 399 us MUL_MAT_ID(ffn_moe_down-69): 205 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 352 us MUL_MAT_ID(ffn_moe_down-70): 206 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 379 us MUL_MAT_ID(ffn_moe_down-71): 429 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 359 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 92 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 375 us MUL_MAT_ID(ffn_moe_down-73): 194 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 357 us MUL_MAT_ID(ffn_moe_down-74): 294 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 377 us MUL_MAT_ID(ffn_moe_down-75): 220 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 352 us MUL_MAT_ID(ffn_moe_down-76): 312 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 386 us MUL_MAT_ID(ffn_moe_down-77): 242 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 361 us MUL_MAT_ID(ffn_moe_down-78): 196 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 361 us MUL_MAT_ID(ffn_moe_down-79): 275 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 382 us MUL_MAT_ID(ffn_moe_down-80): 186 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 372 us MUL_MAT_ID(ffn_moe_down-81): 188 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 391 us MUL_MAT_ID(ffn_moe_down-82): 187 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 397 us MUL_MAT_ID(ffn_moe_down-83): 191 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 366 us MUL_MAT_ID(ffn_moe_down-84): 227 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 399 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 362 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 392 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 387 us MUL_MAT_ID(ffn_moe_down-88): 183 us ggml_barrier(...): 136 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 393 us MUL_MAT_ID(ffn_moe_down-89): 186 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 383 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 123 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 375 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 34 us GET_ROWS(inp_embd): 5 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 389 us MUL_MAT_ID(ffn_moe_down-25): 184 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 383 us MUL_MAT_ID(ffn_moe_down-26): 196 us ggml_barrier(...): 96 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 384 us MUL_MAT_ID(ffn_moe_down-27): 192 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 403 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 386 us MUL_MAT_ID(ffn_moe_down-29): 510 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 366 us MUL_MAT_ID(ffn_moe_down-30): 207 us ggml_barrier(...): 143 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 395 us MUL_MAT_ID(ffn_moe_down-31): 193 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 478 us MUL_MAT_ID(ffn_moe_down-32): 203 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 366 us MUL_MAT_ID(ffn_moe_down-33): 195 us ggml_barrier(...): 140 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 391 us MUL_MAT_ID(ffn_moe_down-34): 238 us ggml_barrier(...): 314 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 396 us MUL_MAT_ID(ffn_moe_down-35): 210 us ggml_barrier(...): 142 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 387 us MUL_MAT_ID(ffn_moe_down-36): 273 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 391 us MUL_MAT_ID(ffn_moe_down-37): 224 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 404 us MUL_MAT_ID(ffn_moe_down-38): 200 us ggml_barrier(...): 107 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 376 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 473 us MUL_MAT_ID(ffn_moe_down-40): 430 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 364 us MUL_MAT_ID(ffn_moe_down-41): 207 us ggml_barrier(...): 119 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 387 us MUL_MAT_ID(ffn_moe_down-42): 192 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 360 us MUL_MAT_ID(ffn_moe_down-43): 196 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 361 us MUL_MAT_ID(ffn_moe_down-44): 195 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 386 us MUL_MAT_ID(ffn_moe_down-45): 213 us ggml_barrier(...): 289 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 340 us MUL_MAT_ID(ffn_moe_down-46): 190 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 422 us MUL_MAT_ID(ffn_moe_down-47): 244 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 399 us MUL_MAT_ID(ffn_moe_down-48): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 360 us MUL_MAT_ID(ffn_moe_down-49): 188 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 390 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 366 us MUL_MAT_ID(ffn_moe_down-51): 405 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 367 us MUL_MAT_ID(ffn_moe_down-52): 218 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 415 us MUL_MAT_ID(ffn_moe_down-53): 203 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 369 us MUL_MAT_ID(ffn_moe_down-54): 368 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 366 us MUL_MAT_ID(ffn_moe_down-55): 195 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 363 us MUL_MAT_ID(ffn_moe_down-56): 208 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 388 us MUL_MAT_ID(ffn_moe_down-57): 188 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 378 us MUL_MAT_ID(ffn_moe_down-58): 191 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 392 us MUL_MAT_ID(ffn_moe_down-59): 212 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 397 us MUL_MAT_ID(ffn_moe_down-60): 289 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 394 us MUL_MAT_ID(ffn_moe_down-61): 206 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 391 us MUL_MAT_ID(ffn_moe_down-62): 553 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 374 us MUL_MAT_ID(ffn_moe_down-63): 210 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 547 us MUL_MAT_ID(ffn_moe_down-64): 167 us ggml_barrier(...): 165 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 347 us MUL_MAT_ID(ffn_moe_down-65): 216 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 380 us MUL_MAT_ID(ffn_moe_down-66): 184 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 387 us MUL_MAT_ID(ffn_moe_down-67): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 420 us MUL_MAT_ID(ffn_moe_down-68): 191 us ggml_barrier(...): 323 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 430 us MUL_MAT_ID(ffn_moe_down-69): 487 us ggml_barrier(...): 144 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 401 us MUL_MAT_ID(ffn_moe_down-70): 207 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 409 us MUL_MAT_ID(ffn_moe_down-71): 198 us ggml_barrier(...): 106 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 197 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 465 us MUL_MAT_ID(ffn_moe_down-73): 197 us ggml_barrier(...): 190 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 380 us MUL_MAT_ID(ffn_moe_down-74): 200 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 359 us MUL_MAT_ID(ffn_moe_down-75): 245 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 402 us MUL_MAT_ID(ffn_moe_down-76): 198 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 378 us MUL_MAT_ID(ffn_moe_down-77): 253 us ggml_barrier(...): 120 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 393 us MUL_MAT_ID(ffn_moe_down-78): 203 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 404 us MUL_MAT_ID(ffn_moe_down-79): 193 us ggml_barrier(...): 128 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 376 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 118 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 369 us MUL_MAT_ID(ffn_moe_down-81): 268 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 408 us MUL_MAT_ID(ffn_moe_down-82): 185 us ggml_barrier(...): 99 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 446 us MUL_MAT_ID(ffn_moe_down-83): 196 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 354 us MUL_MAT_ID(ffn_moe_down-84): 243 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 420 us MUL_MAT_ID(ffn_moe_down-85): 228 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 369 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 372 us MUL_MAT_ID(ffn_moe_down-87): 504 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 355 us MUL_MAT_ID(ffn_moe_down-88): 283 us ggml_barrier(...): 209 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 356 us MUL_MAT_ID(ffn_moe_down-89): 181 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 373 us MUL_MAT_ID(ffn_moe_down-90): 239 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 369 us MUL_MAT_ID(ffn_moe_down-91): 239 us ggml_barrier(...): 59 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 406 us MUL_MAT_ID(ffn_moe_down-25): 186 us ggml_barrier(...): 166 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 375 us MUL_MAT_ID(ffn_moe_down-26): 281 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 503 us MUL_MAT_ID(ffn_moe_down-27): 223 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 376 us MUL_MAT_ID(ffn_moe_down-28): 211 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 358 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 346 us MUL_MAT_ID(ffn_moe_down-30): 444 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 374 us MUL_MAT_ID(ffn_moe_down-31): 401 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 384 us MUL_MAT_ID(ffn_moe_down-32): 204 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 384 us MUL_MAT_ID(ffn_moe_down-33): 205 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 383 us MUL_MAT_ID(ffn_moe_down-34): 194 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 406 us MUL_MAT_ID(ffn_moe_down-35): 220 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 393 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 438 us MUL_MAT_ID(ffn_moe_down-37): 243 us ggml_barrier(...): 228 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 393 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 364 us MUL_MAT_ID(ffn_moe_down-39): 225 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 368 us MUL_MAT_ID(ffn_moe_down-40): 221 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 368 us MUL_MAT_ID(ffn_moe_down-41): 379 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 375 us MUL_MAT_ID(ffn_moe_down-42): 203 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 372 us MUL_MAT_ID(ffn_moe_down-43): 226 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 388 us MUL_MAT_ID(ffn_moe_down-44): 208 us ggml_barrier(...): 138 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 377 us MUL_MAT_ID(ffn_moe_down-45): 242 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 366 us MUL_MAT_ID(ffn_moe_down-46): 200 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 446 us MUL_MAT_ID(ffn_moe_down-47): 170 us ggml_barrier(...): 231 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 351 us MUL_MAT_ID(ffn_moe_down-48): 199 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 395 us MUL_MAT_ID(ffn_moe_down-49): 187 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 370 us MUL_MAT_ID(ffn_moe_down-50): 209 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 358 us MUL_MAT_ID(ffn_moe_down-51): 190 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 386 us MUL_MAT_ID(ffn_moe_down-52): 198 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 361 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 398 us MUL_MAT_ID(ffn_moe_down-54): 494 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 371 us MUL_MAT_ID(ffn_moe_down-55): 206 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 363 us MUL_MAT_ID(ffn_moe_down-56): 348 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 365 us MUL_MAT_ID(ffn_moe_down-57): 191 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 394 us MUL_MAT_ID(ffn_moe_down-58): 193 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 348 us MUL_MAT_ID(ffn_moe_down-59): 183 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 415 us MUL_MAT_ID(ffn_moe_down-60): 188 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 380 us MUL_MAT_ID(ffn_moe_down-61): 283 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 366 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 387 us MUL_MAT_ID(ffn_moe_down-63): 256 us ggml_barrier(...): 262 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 388 us MUL_MAT_ID(ffn_moe_down-64): 208 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 355 us MUL_MAT_ID(ffn_moe_down-65): 280 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 358 us MUL_MAT_ID(ffn_moe_down-66): 197 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 370 us MUL_MAT_ID(ffn_moe_down-67): 200 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 354 us MUL_MAT_ID(ffn_moe_down-68): 285 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 357 us MUL_MAT_ID(ffn_moe_down-69): 195 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 368 us MUL_MAT_ID(ffn_moe_down-70): 385 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 372 us MUL_MAT_ID(ffn_moe_down-71): 368 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 358 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 364 us MUL_MAT_ID(ffn_moe_down-73): 509 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 407 us MUL_MAT_ID(ffn_moe_down-74): 198 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 393 us MUL_MAT_ID(ffn_moe_down-75): 233 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 370 us MUL_MAT_ID(ffn_moe_down-76): 202 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 532 us MUL_MAT_ID(ffn_moe_down-77): 177 us ggml_barrier(...): 175 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 421 us MUL_MAT_ID(ffn_moe_down-78): 318 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 348 us MUL_MAT_ID(ffn_moe_down-79): 220 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 406 us MUL_MAT_ID(ffn_moe_down-80): 190 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 350 us MUL_MAT_ID(ffn_moe_down-81): 210 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 399 us MUL_MAT_ID(ffn_moe_down-82): 188 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 399 us MUL_MAT_ID(ffn_moe_down-83): 335 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 404 us MUL_MAT_ID(ffn_moe_down-84): 228 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 415 us MUL_MAT_ID(ffn_moe_down-85): 221 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 401 us MUL_MAT_ID(ffn_moe_down-86): 181 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 415 us MUL_MAT_ID(ffn_moe_down-87): 194 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 389 us MUL_MAT_ID(ffn_moe_down-88): 180 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 366 us MUL_MAT_ID(ffn_moe_down-89): 203 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 426 us MUL_MAT_ID(ffn_moe_down-90): 233 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 402 us MUL_MAT_ID(ffn_moe_down-91): 240 us ggml_barrier(...): 36 us GET_ROWS(inp_embd): 19 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 401 us MUL_MAT_ID(ffn_moe_down-25): 186 us ggml_barrier(...): 181 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 398 us MUL_MAT_ID(ffn_moe_down-26): 189 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 391 us MUL_MAT_ID(ffn_moe_down-27): 202 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 403 us MUL_MAT_ID(ffn_moe_down-28): 256 us ggml_barrier(...): 225 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 397 us MUL_MAT_ID(ffn_moe_down-29): 201 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 390 us MUL_MAT_ID(ffn_moe_down-30): 207 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 412 us MUL_MAT_ID(ffn_moe_down-31): 209 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 368 us MUL_MAT_ID(ffn_moe_down-32): 556 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 352 us MUL_MAT_ID(ffn_moe_down-33): 248 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 444 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 406 us MUL_MAT_ID(ffn_moe_down-35): 369 us ggml_barrier(...): 256 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 381 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 362 us MUL_MAT_ID(ffn_moe_down-37): 234 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 400 us MUL_MAT_ID(ffn_moe_down-38): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 378 us MUL_MAT_ID(ffn_moe_down-39): 253 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 413 us MUL_MAT_ID(ffn_moe_down-40): 294 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 424 us MUL_MAT_ID(ffn_moe_down-41): 188 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 531 us MUL_MAT_ID(ffn_moe_down-42): 425 us ggml_barrier(...): 152 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 400 us MUL_MAT_ID(ffn_moe_down-43): 193 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 365 us MUL_MAT_ID(ffn_moe_down-44): 201 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 391 us MUL_MAT_ID(ffn_moe_down-45): 197 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 347 us MUL_MAT_ID(ffn_moe_down-46): 224 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 367 us MUL_MAT_ID(ffn_moe_down-47): 202 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 405 us MUL_MAT_ID(ffn_moe_down-48): 531 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 392 us MUL_MAT_ID(ffn_moe_down-49): 195 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 380 us MUL_MAT_ID(ffn_moe_down-50): 354 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 375 us MUL_MAT_ID(ffn_moe_down-51): 195 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 564 us MUL_MAT_ID(ffn_moe_down-52): 173 us ggml_barrier(...): 133 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 395 us MUL_MAT_ID(ffn_moe_down-53): 347 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 408 us MUL_MAT_ID(ffn_moe_down-54): 182 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 393 us MUL_MAT_ID(ffn_moe_down-55): 186 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 401 us MUL_MAT_ID(ffn_moe_down-56): 201 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 405 us MUL_MAT_ID(ffn_moe_down-57): 185 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 375 us MUL_MAT_ID(ffn_moe_down-58): 186 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 377 us MUL_MAT_ID(ffn_moe_down-59): 196 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 395 us MUL_MAT_ID(ffn_moe_down-60): 330 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 429 us MUL_MAT_ID(ffn_moe_down-61): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 400 us MUL_MAT_ID(ffn_moe_down-62): 226 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 393 us MUL_MAT_ID(ffn_moe_down-63): 193 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 486 us MUL_MAT_ID(ffn_moe_down-64): 412 us ggml_barrier(...): 210 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 422 us MUL_MAT_ID(ffn_moe_down-65): 256 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 401 us MUL_MAT_ID(ffn_moe_down-66): 200 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 416 us MUL_MAT_ID(ffn_moe_down-67): 354 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 389 us MUL_MAT_ID(ffn_moe_down-68): 193 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 430 us MUL_MAT_ID(ffn_moe_down-69): 213 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 374 us MUL_MAT_ID(ffn_moe_down-70): 199 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 374 us MUL_MAT_ID(ffn_moe_down-71): 194 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 378 us MUL_MAT_ID(ffn_moe_down-72): 200 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 449 us MUL_MAT_ID(ffn_moe_down-73): 211 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 369 us MUL_MAT_ID(ffn_moe_down-74): 209 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 356 us MUL_MAT_ID(ffn_moe_down-75): 300 us ggml_barrier(...): 177 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 376 us MUL_MAT_ID(ffn_moe_down-76): 236 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 447 us MUL_MAT_ID(ffn_moe_down-77): 275 us ggml_barrier(...): 93 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 403 us MUL_MAT_ID(ffn_moe_down-78): 207 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 383 us MUL_MAT_ID(ffn_moe_down-79): 226 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 401 us MUL_MAT_ID(ffn_moe_down-80): 260 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 375 us MUL_MAT_ID(ffn_moe_down-81): 495 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 364 us MUL_MAT_ID(ffn_moe_down-82): 258 us ggml_barrier(...): 167 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 482 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 197 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 389 us MUL_MAT_ID(ffn_moe_down-84): 258 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 396 us MUL_MAT_ID(ffn_moe_down-85): 287 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 407 us MUL_MAT_ID(ffn_moe_down-86): 210 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 384 us MUL_MAT_ID(ffn_moe_down-87): 193 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 403 us MUL_MAT_ID(ffn_moe_down-88): 190 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 359 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 395 us MUL_MAT_ID(ffn_moe_down-90): 247 us ggml_barrier(...): 240 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 501 us MUL_MAT_ID(ffn_moe_down-91): 222 us ggml_barrier(...): 121 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 426 us MUL_MAT_ID(ffn_moe_down-25): 191 us ggml_barrier(...): 9 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 386 us MUL_MAT_ID(ffn_moe_down-26): 340 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 388 us MUL_MAT_ID(ffn_moe_down-27): 196 us ggml_barrier(...): 101 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 411 us MUL_MAT_ID(ffn_moe_down-28): 192 us ggml_barrier(...): 346 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 359 us MUL_MAT_ID(ffn_moe_down-29): 199 us ggml_barrier(...): 148 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 504 us MUL_MAT_ID(ffn_moe_down-30): 179 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 394 us MUL_MAT_ID(ffn_moe_down-31): 193 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 371 us MUL_MAT_ID(ffn_moe_down-32): 204 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 377 us MUL_MAT_ID(ffn_moe_down-33): 209 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 399 us MUL_MAT_ID(ffn_moe_down-34): 205 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 386 us MUL_MAT_ID(ffn_moe_down-35): 209 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 376 us MUL_MAT_ID(ffn_moe_down-36): 209 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 365 us MUL_MAT_ID(ffn_moe_down-37): 231 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 380 us MUL_MAT_ID(ffn_moe_down-38): 209 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 382 us MUL_MAT_ID(ffn_moe_down-39): 235 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 382 us MUL_MAT_ID(ffn_moe_down-40): 722 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 363 us MUL_MAT_ID(ffn_moe_down-41): 197 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 383 us MUL_MAT_ID(ffn_moe_down-42): 191 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 385 us MUL_MAT_ID(ffn_moe_down-43): 350 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 345 us MUL_MAT_ID(ffn_moe_down-44): 190 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 391 us MUL_MAT_ID(ffn_moe_down-45): 387 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 386 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 384 us MUL_MAT_ID(ffn_moe_down-47): 271 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 385 us MUL_MAT_ID(ffn_moe_down-48): 206 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 375 us MUL_MAT_ID(ffn_moe_down-49): 196 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 381 us MUL_MAT_ID(ffn_moe_down-50): 203 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 427 us MUL_MAT_ID(ffn_moe_down-51): 193 us ggml_barrier(...): 280 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 355 us MUL_MAT_ID(ffn_moe_down-52): 345 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 391 us MUL_MAT_ID(ffn_moe_down-53): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 389 us MUL_MAT_ID(ffn_moe_down-54): 217 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 375 us MUL_MAT_ID(ffn_moe_down-55): 192 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 379 us MUL_MAT_ID(ffn_moe_down-56): 191 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 390 us MUL_MAT_ID(ffn_moe_down-57): 185 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 346 us MUL_MAT_ID(ffn_moe_down-58): 201 us ggml_barrier(...): 74 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 403 us MUL_MAT_ID(ffn_moe_down-59): 278 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 391 us MUL_MAT_ID(ffn_moe_down-60): 370 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 379 us MUL_MAT_ID(ffn_moe_down-61): 316 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 391 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 340 us MUL_MAT_ID(ffn_moe_down-63): 272 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 386 us MUL_MAT_ID(ffn_moe_down-64): 190 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 402 us MUL_MAT_ID(ffn_moe_down-65): 204 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 349 us MUL_MAT_ID(ffn_moe_down-66): 219 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 365 us MUL_MAT_ID(ffn_moe_down-67): 204 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 378 us MUL_MAT_ID(ffn_moe_down-68): 182 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 409 us MUL_MAT_ID(ffn_moe_down-69): 212 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 391 us MUL_MAT_ID(ffn_moe_down-70): 220 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 373 us MUL_MAT_ID(ffn_moe_down-71): 227 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 388 us MUL_MAT_ID(ffn_moe_down-72): 202 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 400 us MUL_MAT_ID(ffn_moe_down-73): 192 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 352 us MUL_MAT_ID(ffn_moe_down-74): 226 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 406 us MUL_MAT_ID(ffn_moe_down-75): 201 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 379 us MUL_MAT_ID(ffn_moe_down-76): 185 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 367 us MUL_MAT_ID(ffn_moe_down-77): 570 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 390 us MUL_MAT_ID(ffn_moe_down-78): 220 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 385 us MUL_MAT_ID(ffn_moe_down-79): 192 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 369 us MUL_MAT_ID(ffn_moe_down-80): 203 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 412 us MUL_MAT_ID(ffn_moe_down-81): 213 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 370 us MUL_MAT_ID(ffn_moe_down-82): 190 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 354 us MUL_MAT_ID(ffn_moe_down-83): 197 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 400 us MUL_MAT_ID(ffn_moe_down-84): 222 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 369 us MUL_MAT_ID(ffn_moe_down-85): 232 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 377 us MUL_MAT_ID(ffn_moe_down-86): 188 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 391 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 372 us MUL_MAT_ID(ffn_moe_down-88): 195 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 380 us MUL_MAT_ID(ffn_moe_down-89): 410 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 370 us MUL_MAT_ID(ffn_moe_down-90): 224 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 418 us MUL_MAT_ID(ffn_moe_down-91): 248 us ggml_barrier(...): 18 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 426 us MUL_MAT_ID(ffn_moe_down-25): 179 us ggml_barrier(...): 226 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 401 us MUL_MAT_ID(ffn_moe_down-26): 198 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 398 us MUL_MAT_ID(ffn_moe_down-27): 197 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 397 us MUL_MAT_ID(ffn_moe_down-28): 204 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 376 us MUL_MAT_ID(ffn_moe_down-29): 212 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 357 us MUL_MAT_ID(ffn_moe_down-30): 236 us ggml_barrier(...): 249 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 383 us MUL_MAT_ID(ffn_moe_down-31): 185 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 397 us MUL_MAT_ID(ffn_moe_down-32): 186 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 384 us MUL_MAT_ID(ffn_moe_down-33): 212 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 403 us MUL_MAT_ID(ffn_moe_down-34): 204 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 405 us MUL_MAT_ID(ffn_moe_down-35): 199 us ggml_barrier(...): 3 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 417 us MUL_MAT_ID(ffn_moe_down-36): 201 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 374 us MUL_MAT_ID(ffn_moe_down-37): 233 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 402 us MUL_MAT_ID(ffn_moe_down-38): 195 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 380 us MUL_MAT_ID(ffn_moe_down-39): 237 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 517 us MUL_MAT_ID(ffn_moe_down-40): 210 us ggml_barrier(...): 163 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 377 us MUL_MAT_ID(ffn_moe_down-41): 175 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 392 us MUL_MAT_ID(ffn_moe_down-42): 197 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 409 us MUL_MAT_ID(ffn_moe_down-43): 187 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 371 us MUL_MAT_ID(ffn_moe_down-44): 185 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 394 us MUL_MAT_ID(ffn_moe_down-45): 204 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 348 us MUL_MAT_ID(ffn_moe_down-46): 197 us ggml_barrier(...): 70 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 393 us MUL_MAT_ID(ffn_moe_down-47): 191 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 395 us MUL_MAT_ID(ffn_moe_down-48): 192 us ggml_barrier(...): 62 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 424 us MUL_MAT_ID(ffn_moe_down-49): 186 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 387 us MUL_MAT_ID(ffn_moe_down-50): 541 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 359 us MUL_MAT_ID(ffn_moe_down-51): 338 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 379 us MUL_MAT_ID(ffn_moe_down-52): 186 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 384 us MUL_MAT_ID(ffn_moe_down-53): 210 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 417 us MUL_MAT_ID(ffn_moe_down-54): 219 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 395 us MUL_MAT_ID(ffn_moe_down-55): 186 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 397 us MUL_MAT_ID(ffn_moe_down-56): 192 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 511 us MUL_MAT_ID(ffn_moe_down-57): 213 us ggml_barrier(...): 86 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 399 us MUL_MAT_ID(ffn_moe_down-58): 200 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 473 us MUL_MAT_ID(ffn_moe_down-59): 411 us ggml_barrier(...): 194 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 489 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 365 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 387 us MUL_MAT_ID(ffn_moe_down-62): 197 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 377 us MUL_MAT_ID(ffn_moe_down-63): 194 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 362 us MUL_MAT_ID(ffn_moe_down-64): 202 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 395 us MUL_MAT_ID(ffn_moe_down-65): 222 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 415 us MUL_MAT_ID(ffn_moe_down-66): 191 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 407 us MUL_MAT_ID(ffn_moe_down-67): 186 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 379 us MUL_MAT_ID(ffn_moe_down-68): 192 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 369 us MUL_MAT_ID(ffn_moe_down-69): 199 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 352 us MUL_MAT_ID(ffn_moe_down-70): 202 us ggml_barrier(...): 308 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 364 us MUL_MAT_ID(ffn_moe_down-71): 210 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 480 us MUL_MAT_ID(ffn_moe_down-72): 299 us ggml_barrier(...): 75 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 399 us MUL_MAT_ID(ffn_moe_down-73): 196 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 420 us MUL_MAT_ID(ffn_moe_down-74): 395 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 392 us MUL_MAT_ID(ffn_moe_down-75): 203 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 423 us MUL_MAT_ID(ffn_moe_down-76): 218 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 419 us MUL_MAT_ID(ffn_moe_down-77): 189 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 366 us MUL_MAT_ID(ffn_moe_down-78): 354 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 389 us MUL_MAT_ID(ffn_moe_down-79): 302 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 405 us MUL_MAT_ID(ffn_moe_down-80): 198 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 396 us MUL_MAT_ID(ffn_moe_down-81): 354 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 388 us MUL_MAT_ID(ffn_moe_down-82): 197 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 421 us MUL_MAT_ID(ffn_moe_down-83): 291 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 353 us MUL_MAT_ID(ffn_moe_down-84): 238 us ggml_barrier(...): 213 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 363 us MUL_MAT_ID(ffn_moe_down-85): 622 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 417 us MUL_MAT_ID(ffn_moe_down-86): 207 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 356 us MUL_MAT_ID(ffn_moe_down-87): 199 us ggml_barrier(...): 88 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 392 us MUL_MAT_ID(ffn_moe_down-88): 390 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 386 us MUL_MAT_ID(ffn_moe_down-89): 190 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 416 us MUL_MAT_ID(ffn_moe_down-90): 634 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 373 us MUL_MAT_ID(ffn_moe_down-91): 253 us ggml_barrier(...): 59 us GET_ROWS(inp_embd): 17 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 401 us MUL_MAT_ID(ffn_moe_down-25): 201 us ggml_barrier(...): 178 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 399 us MUL_MAT_ID(ffn_moe_down-26): 190 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 366 us MUL_MAT_ID(ffn_moe_down-27): 205 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 395 us MUL_MAT_ID(ffn_moe_down-28): 188 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 406 us MUL_MAT_ID(ffn_moe_down-29): 404 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 385 us MUL_MAT_ID(ffn_moe_down-30): 192 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 368 us MUL_MAT_ID(ffn_moe_down-31): 196 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 401 us MUL_MAT_ID(ffn_moe_down-32): 180 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 372 us MUL_MAT_ID(ffn_moe_down-33): 212 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 409 us MUL_MAT_ID(ffn_moe_down-34): 201 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 372 us MUL_MAT_ID(ffn_moe_down-35): 187 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 354 us MUL_MAT_ID(ffn_moe_down-36): 213 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 360 us MUL_MAT_ID(ffn_moe_down-37): 230 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 355 us MUL_MAT_ID(ffn_moe_down-38): 193 us ggml_barrier(...): 71 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 379 us MUL_MAT_ID(ffn_moe_down-39): 236 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 376 us MUL_MAT_ID(ffn_moe_down-40): 237 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 380 us MUL_MAT_ID(ffn_moe_down-41): 198 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 425 us MUL_MAT_ID(ffn_moe_down-42): 234 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 387 us MUL_MAT_ID(ffn_moe_down-43): 189 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 380 us MUL_MAT_ID(ffn_moe_down-44): 225 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 385 us MUL_MAT_ID(ffn_moe_down-45): 200 us ggml_barrier(...): 170 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 369 us MUL_MAT_ID(ffn_moe_down-46): 227 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 397 us MUL_MAT_ID(ffn_moe_down-47): 209 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 367 us MUL_MAT_ID(ffn_moe_down-48): 360 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 330 us MUL_MAT_ID(ffn_moe_down-49): 191 us ggml_barrier(...): 87 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 392 us MUL_MAT_ID(ffn_moe_down-50): 195 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 367 us MUL_MAT_ID(ffn_moe_down-51): 201 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 383 us MUL_MAT_ID(ffn_moe_down-52): 187 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 370 us MUL_MAT_ID(ffn_moe_down-53): 199 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 414 us MUL_MAT_ID(ffn_moe_down-54): 186 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 364 us MUL_MAT_ID(ffn_moe_down-55): 223 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 393 us MUL_MAT_ID(ffn_moe_down-56): 191 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 382 us MUL_MAT_ID(ffn_moe_down-57): 408 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 372 us MUL_MAT_ID(ffn_moe_down-58): 197 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 403 us MUL_MAT_ID(ffn_moe_down-59): 199 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 388 us MUL_MAT_ID(ffn_moe_down-60): 194 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 381 us MUL_MAT_ID(ffn_moe_down-61): 277 us ggml_barrier(...): 40 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 406 us MUL_MAT_ID(ffn_moe_down-62): 195 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 370 us MUL_MAT_ID(ffn_moe_down-63): 237 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 382 us MUL_MAT_ID(ffn_moe_down-64): 241 us ggml_barrier(...): 72 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 397 us MUL_MAT_ID(ffn_moe_down-65): 259 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 389 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 382 us MUL_MAT_ID(ffn_moe_down-67): 197 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 392 us MUL_MAT_ID(ffn_moe_down-68): 504 us ggml_barrier(...): 108 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 408 us MUL_MAT_ID(ffn_moe_down-69): 415 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 385 us MUL_MAT_ID(ffn_moe_down-70): 399 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 366 us MUL_MAT_ID(ffn_moe_down-71): 195 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 353 us MUL_MAT_ID(ffn_moe_down-72): 206 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 341 us MUL_MAT_ID(ffn_moe_down-73): 223 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 366 us MUL_MAT_ID(ffn_moe_down-74): 454 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 373 us MUL_MAT_ID(ffn_moe_down-75): 413 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 362 us MUL_MAT_ID(ffn_moe_down-76): 194 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 393 us MUL_MAT_ID(ffn_moe_down-77): 204 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 414 us MUL_MAT_ID(ffn_moe_down-78): 197 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 395 us MUL_MAT_ID(ffn_moe_down-79): 186 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 395 us MUL_MAT_ID(ffn_moe_down-80): 554 us ggml_barrier(...): 66 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 362 us MUL_MAT_ID(ffn_moe_down-81): 236 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 356 us MUL_MAT_ID(ffn_moe_down-82): 182 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 410 us MUL_MAT_ID(ffn_moe_down-83): 251 us ggml_barrier(...): 18 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 346 us MUL_MAT_ID(ffn_moe_down-84): 241 us ggml_barrier(...): 85 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 376 us MUL_MAT_ID(ffn_moe_down-85): 256 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 378 us MUL_MAT_ID(ffn_moe_down-86): 201 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 394 us MUL_MAT_ID(ffn_moe_down-87): 192 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 420 us MUL_MAT_ID(ffn_moe_down-88): 211 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 366 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 375 us MUL_MAT_ID(ffn_moe_down-90): 663 us ggml_barrier(...): 307 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 396 us MUL_MAT_ID(ffn_moe_down-91): 242 us ggml_barrier(...): 39 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 418 us MUL_MAT_ID(ffn_moe_down-25): 190 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 400 us MUL_MAT_ID(ffn_moe_down-26): 209 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 364 us MUL_MAT_ID(ffn_moe_down-27): 541 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 382 us MUL_MAT_ID(ffn_moe_down-28): 201 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 386 us MUL_MAT_ID(ffn_moe_down-29): 188 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 369 us MUL_MAT_ID(ffn_moe_down-30): 187 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 366 us MUL_MAT_ID(ffn_moe_down-31): 206 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 390 us MUL_MAT_ID(ffn_moe_down-32): 346 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 389 us MUL_MAT_ID(ffn_moe_down-33): 202 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 340 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 385 us MUL_MAT_ID(ffn_moe_down-35): 189 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 362 us MUL_MAT_ID(ffn_moe_down-36): 204 us ggml_barrier(...): 13 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 402 us MUL_MAT_ID(ffn_moe_down-37): 315 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 374 us MUL_MAT_ID(ffn_moe_down-38): 211 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 353 us MUL_MAT_ID(ffn_moe_down-39): 232 us ggml_barrier(...): 90 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 368 us MUL_MAT_ID(ffn_moe_down-40): 198 us ggml_barrier(...): 306 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 360 us MUL_MAT_ID(ffn_moe_down-41): 190 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 388 us MUL_MAT_ID(ffn_moe_down-42): 205 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 359 us MUL_MAT_ID(ffn_moe_down-43): 205 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 378 us MUL_MAT_ID(ffn_moe_down-44): 192 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 360 us MUL_MAT_ID(ffn_moe_down-45): 209 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 353 us MUL_MAT_ID(ffn_moe_down-46): 187 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 377 us MUL_MAT_ID(ffn_moe_down-47): 180 us ggml_barrier(...): 362 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 397 us MUL_MAT_ID(ffn_moe_down-48): 205 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 359 us MUL_MAT_ID(ffn_moe_down-49): 189 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 376 us MUL_MAT_ID(ffn_moe_down-50): 455 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 429 us MUL_MAT_ID(ffn_moe_down-51): 204 us ggml_barrier(...): 352 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 519 us MUL_MAT_ID(ffn_moe_down-52): 199 us ggml_barrier(...): 122 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 417 us MUL_MAT_ID(ffn_moe_down-53): 306 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 377 us MUL_MAT_ID(ffn_moe_down-54): 178 us ggml_barrier(...): 339 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 410 us MUL_MAT_ID(ffn_moe_down-55): 191 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 412 us MUL_MAT_ID(ffn_moe_down-56): 392 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 395 us MUL_MAT_ID(ffn_moe_down-57): 207 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 395 us MUL_MAT_ID(ffn_moe_down-58): 195 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 413 us MUL_MAT_ID(ffn_moe_down-59): 340 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 394 us MUL_MAT_ID(ffn_moe_down-60): 351 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 387 us MUL_MAT_ID(ffn_moe_down-61): 195 us ggml_barrier(...): 16 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 380 us MUL_MAT_ID(ffn_moe_down-62): 199 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 381 us MUL_MAT_ID(ffn_moe_down-63): 224 us ggml_barrier(...): 77 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 412 us MUL_MAT_ID(ffn_moe_down-64): 197 us ggml_barrier(...): 17 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 381 us MUL_MAT_ID(ffn_moe_down-65): 336 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 368 us MUL_MAT_ID(ffn_moe_down-66): 199 us ggml_barrier(...): 63 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 549 us MUL_MAT_ID(ffn_moe_down-67): 326 us ggml_barrier(...): 174 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 353 us MUL_MAT_ID(ffn_moe_down-68): 214 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 408 us MUL_MAT_ID(ffn_moe_down-69): 190 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 373 us MUL_MAT_ID(ffn_moe_down-70): 182 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 384 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 404 us MUL_MAT_ID(ffn_moe_down-72): 191 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 370 us MUL_MAT_ID(ffn_moe_down-73): 226 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 348 us MUL_MAT_ID(ffn_moe_down-74): 195 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 375 us MUL_MAT_ID(ffn_moe_down-75): 194 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 438 us MUL_MAT_ID(ffn_moe_down-76): 191 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 381 us MUL_MAT_ID(ffn_moe_down-77): 221 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 377 us MUL_MAT_ID(ffn_moe_down-78): 192 us ggml_barrier(...): 67 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 362 us MUL_MAT_ID(ffn_moe_down-79): 216 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 373 us MUL_MAT_ID(ffn_moe_down-80): 205 us ggml_barrier(...): 48 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 384 us MUL_MAT_ID(ffn_moe_down-81): 197 us ggml_barrier(...): 59 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 391 us MUL_MAT_ID(ffn_moe_down-82): 186 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 383 us MUL_MAT_ID(ffn_moe_down-83): 198 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 389 us MUL_MAT_ID(ffn_moe_down-84): 224 us ggml_barrier(...): 61 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 365 us MUL_MAT_ID(ffn_moe_down-85): 238 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 388 us MUL_MAT_ID(ffn_moe_down-86): 525 us ggml_barrier(...): 35 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 382 us MUL_MAT_ID(ffn_moe_down-87): 197 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 401 us MUL_MAT_ID(ffn_moe_down-88): 204 us ggml_barrier(...): 27 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 435 us MUL_MAT_ID(ffn_moe_down-89): 352 us ggml_barrier(...): 241 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 346 us MUL_MAT_ID(ffn_moe_down-90): 216 us ggml_barrier(...): 268 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 401 us MUL_MAT_ID(ffn_moe_down-91): 242 us ggml_barrier(...): 115 us GET_ROWS(inp_embd): 14 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 428 us MUL_MAT_ID(ffn_moe_down-25): 191 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 371 us MUL_MAT_ID(ffn_moe_down-26): 192 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 390 us MUL_MAT_ID(ffn_moe_down-27): 195 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 379 us MUL_MAT_ID(ffn_moe_down-28): 211 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 385 us MUL_MAT_ID(ffn_moe_down-29): 193 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 357 us MUL_MAT_ID(ffn_moe_down-30): 193 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 376 us MUL_MAT_ID(ffn_moe_down-31): 192 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 414 us MUL_MAT_ID(ffn_moe_down-32): 286 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 435 us MUL_MAT_ID(ffn_moe_down-33): 203 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 404 us MUL_MAT_ID(ffn_moe_down-34): 208 us ggml_barrier(...): 15 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 355 us MUL_MAT_ID(ffn_moe_down-35): 198 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 389 us MUL_MAT_ID(ffn_moe_down-36): 177 us ggml_barrier(...): 329 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 417 us MUL_MAT_ID(ffn_moe_down-37): 246 us ggml_barrier(...): 206 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 391 us MUL_MAT_ID(ffn_moe_down-38): 211 us ggml_barrier(...): 247 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 381 us MUL_MAT_ID(ffn_moe_down-39): 727 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 358 us MUL_MAT_ID(ffn_moe_down-40): 230 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 415 us MUL_MAT_ID(ffn_moe_down-41): 266 us ggml_barrier(...): 171 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 358 us MUL_MAT_ID(ffn_moe_down-42): 239 us ggml_barrier(...): 64 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 398 us MUL_MAT_ID(ffn_moe_down-43): 194 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 358 us MUL_MAT_ID(ffn_moe_down-44): 184 us ggml_barrier(...): 19 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 373 us MUL_MAT_ID(ffn_moe_down-45): 190 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 364 us MUL_MAT_ID(ffn_moe_down-46): 211 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 383 us MUL_MAT_ID(ffn_moe_down-47): 194 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 357 us MUL_MAT_ID(ffn_moe_down-48): 196 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 379 us MUL_MAT_ID(ffn_moe_down-49): 358 us ggml_barrier(...): 56 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 392 us MUL_MAT_ID(ffn_moe_down-50): 207 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 368 us MUL_MAT_ID(ffn_moe_down-51): 213 us ggml_barrier(...): 69 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 405 us MUL_MAT_ID(ffn_moe_down-52): 194 us ggml_barrier(...): 20 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 390 us MUL_MAT_ID(ffn_moe_down-53): 194 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 422 us MUL_MAT_ID(ffn_moe_down-54): 278 us ggml_barrier(...): 137 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 382 us MUL_MAT_ID(ffn_moe_down-55): 229 us ggml_barrier(...): 38 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 376 us MUL_MAT_ID(ffn_moe_down-56): 235 us ggml_barrier(...): 33 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 392 us MUL_MAT_ID(ffn_moe_down-57): 184 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 379 us MUL_MAT_ID(ffn_moe_down-58): 204 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 391 us MUL_MAT_ID(ffn_moe_down-59): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 412 us MUL_MAT_ID(ffn_moe_down-60): 185 us ggml_barrier(...): 25 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 392 us MUL_MAT_ID(ffn_moe_down-61): 190 us ggml_barrier(...): 324 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 409 us MUL_MAT_ID(ffn_moe_down-62): 208 us ggml_barrier(...): 8 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 380 us MUL_MAT_ID(ffn_moe_down-63): 257 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 375 us MUL_MAT_ID(ffn_moe_down-64): 196 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 369 us MUL_MAT_ID(ffn_moe_down-65): 195 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 372 us MUL_MAT_ID(ffn_moe_down-66): 189 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 403 us MUL_MAT_ID(ffn_moe_down-67): 318 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 371 us MUL_MAT_ID(ffn_moe_down-68): 201 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 385 us MUL_MAT_ID(ffn_moe_down-69): 193 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 363 us MUL_MAT_ID(ffn_moe_down-70): 257 us ggml_barrier(...): 528 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 380 us MUL_MAT_ID(ffn_moe_down-71): 196 us ggml_barrier(...): 248 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 354 us MUL_MAT_ID(ffn_moe_down-72): 205 us ggml_barrier(...): 79 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 368 us MUL_MAT_ID(ffn_moe_down-73): 209 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 366 us MUL_MAT_ID(ffn_moe_down-74): 201 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 362 us MUL_MAT_ID(ffn_moe_down-75): 198 us ggml_barrier(...): 55 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 373 us MUL_MAT_ID(ffn_moe_down-76): 195 us ggml_barrier(...): 43 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 378 us MUL_MAT_ID(ffn_moe_down-77): 196 us ggml_barrier(...): 49 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 383 us MUL_MAT_ID(ffn_moe_down-78): 194 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 378 us MUL_MAT_ID(ffn_moe_down-79): 188 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 399 us MUL_MAT_ID(ffn_moe_down-80): 197 us ggml_barrier(...): 42 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 401 us MUL_MAT_ID(ffn_moe_down-81): 188 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 402 us MUL_MAT_ID(ffn_moe_down-82): 183 us ggml_barrier(...): 150 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 378 us MUL_MAT_ID(ffn_moe_down-83): 493 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 356 us MUL_MAT_ID(ffn_moe_down-84): 250 us ggml_barrier(...): 73 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 457 us MUL_MAT_ID(ffn_moe_down-85): 282 us ggml_barrier(...): 169 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 377 us MUL_MAT_ID(ffn_moe_down-86): 203 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 370 us MUL_MAT_ID(ffn_moe_down-87): 190 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 373 us MUL_MAT_ID(ffn_moe_down-88): 199 us ggml_barrier(...): 31 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 379 us MUL_MAT_ID(ffn_moe_down-89): 358 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 360 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 80 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 359 us MUL_MAT_ID(ffn_moe_down-91): 254 us ggml_barrier(...): 53 us GET_ROWS(inp_embd): 15 us ggml_barrier(...): 0 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-25): 378 us MUL_MAT_ID(ffn_moe_down-25): 187 us ggml_barrier(...): 172 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-26): 387 us MUL_MAT_ID(ffn_moe_down-26): 555 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-27): 387 us MUL_MAT_ID(ffn_moe_down-27): 412 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-28): 381 us MUL_MAT_ID(ffn_moe_down-28): 189 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-29): 370 us MUL_MAT_ID(ffn_moe_down-29): 388 us ggml_barrier(...): 54 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-30): 386 us MUL_MAT_ID(ffn_moe_down-30): 178 us ggml_barrier(...): 34 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-31): 350 us MUL_MAT_ID(ffn_moe_down-31): 268 us ggml_barrier(...): 53 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-32): 386 us MUL_MAT_ID(ffn_moe_down-32): 186 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-33): 398 us MUL_MAT_ID(ffn_moe_down-33): 194 us ggml_barrier(...): 5 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-34): 355 us MUL_MAT_ID(ffn_moe_down-34): 212 us ggml_barrier(...): 78 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-35): 383 us MUL_MAT_ID(ffn_moe_down-35): 200 us ggml_barrier(...): 44 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-36): 384 us MUL_MAT_ID(ffn_moe_down-36): 200 us ggml_barrier(...): 102 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-37): 413 us MUL_MAT_ID(ffn_moe_down-37): 227 us ggml_barrier(...): 58 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-38): 374 us MUL_MAT_ID(ffn_moe_down-38): 196 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-39): 378 us MUL_MAT_ID(ffn_moe_down-39): 238 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-40): 388 us MUL_MAT_ID(ffn_moe_down-40): 407 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-41): 400 us MUL_MAT_ID(ffn_moe_down-41): 194 us ggml_barrier(...): 12 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-42): 415 us MUL_MAT_ID(ffn_moe_down-42): 192 us ggml_barrier(...): 50 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-43): 467 us MUL_MAT_ID(ffn_moe_down-43): 380 us ggml_barrier(...): 239 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-44): 373 us MUL_MAT_ID(ffn_moe_down-44): 302 us ggml_barrier(...): 45 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-45): 393 us MUL_MAT_ID(ffn_moe_down-45): 404 us ggml_barrier(...): 14 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-46): 406 us MUL_MAT_ID(ffn_moe_down-46): 220 us ggml_barrier(...): 11 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-47): 411 us MUL_MAT_ID(ffn_moe_down-47): 215 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-48): 386 us MUL_MAT_ID(ffn_moe_down-48): 200 us ggml_barrier(...): 271 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-49): 367 us MUL_MAT_ID(ffn_moe_down-49): 184 us ggml_barrier(...): 41 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-50): 379 us MUL_MAT_ID(ffn_moe_down-50): 200 us ggml_barrier(...): 242 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-51): 397 us MUL_MAT_ID(ffn_moe_down-51): 214 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-52): 400 us MUL_MAT_ID(ffn_moe_down-52): 202 us ggml_barrier(...): 10 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-53): 403 us MUL_MAT_ID(ffn_moe_down-53): 381 us ggml_barrier(...): 37 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-54): 386 us MUL_MAT_ID(ffn_moe_down-54): 170 us ggml_barrier(...): 365 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-55): 369 us MUL_MAT_ID(ffn_moe_down-55): 225 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-56): 370 us MUL_MAT_ID(ffn_moe_down-56): 185 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-57): 370 us MUL_MAT_ID(ffn_moe_down-57): 185 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-58): 363 us MUL_MAT_ID(ffn_moe_down-58): 196 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-59): 352 us MUL_MAT_ID(ffn_moe_down-59): 202 us ggml_barrier(...): 47 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-60): 387 us MUL_MAT_ID(ffn_moe_down-60): 210 us ggml_barrier(...): 51 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-61): 401 us MUL_MAT_ID(ffn_moe_down-61): 279 us ggml_barrier(...): 30 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-62): 380 us MUL_MAT_ID(ffn_moe_down-62): 193 us ggml_barrier(...): 46 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-63): 373 us MUL_MAT_ID(ffn_moe_down-63): 188 us ggml_barrier(...): 60 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-64): 376 us MUL_MAT_ID(ffn_moe_down-64): 182 us ggml_barrier(...): 217 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-65): 401 us MUL_MAT_ID(ffn_moe_down-65): 189 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-66): 394 us MUL_MAT_ID(ffn_moe_down-66): 190 us ggml_barrier(...): 203 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-67): 377 us MUL_MAT_ID(ffn_moe_down-67): 201 us ggml_barrier(...): 274 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-68): 375 us MUL_MAT_ID(ffn_moe_down-68): 216 us ggml_barrier(...): 97 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-69): 360 us MUL_MAT_ID(ffn_moe_down-69): 357 us ggml_barrier(...): 82 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-70): 397 us MUL_MAT_ID(ffn_moe_down-70): 228 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-71): 466 us MUL_MAT_ID(ffn_moe_down-71): 240 us ggml_barrier(...): 130 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-72): 390 us MUL_MAT_ID(ffn_moe_down-72): 188 us ggml_barrier(...): 39 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-73): 388 us MUL_MAT_ID(ffn_moe_down-73): 195 us ggml_barrier(...): 28 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-74): 383 us MUL_MAT_ID(ffn_moe_down-74): 192 us ggml_barrier(...): 26 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-75): 399 us MUL_MAT_ID(ffn_moe_down-75): 191 us ggml_barrier(...): 32 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-76): 399 us MUL_MAT_ID(ffn_moe_down-76): 193 us ggml_barrier(...): 22 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-77): 368 us MUL_MAT_ID(ffn_moe_down-77): 162 us ggml_barrier(...): 325 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-78): 402 us MUL_MAT_ID(ffn_moe_down-78): 199 us ggml_barrier(...): 345 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-79): 381 us MUL_MAT_ID(ffn_moe_down-79): 190 us ggml_barrier(...): 24 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-80): 377 us MUL_MAT_ID(ffn_moe_down-80): 188 us ggml_barrier(...): 105 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-81): 363 us MUL_MAT_ID(ffn_moe_down-81): 204 us ggml_barrier(...): 68 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-82): 409 us MUL_MAT_ID(ffn_moe_down-82): 372 us ggml_barrier(...): 23 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-83): 358 us MUL_MAT_ID(ffn_moe_down-83): 376 us ggml_barrier(...): 52 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-84): 399 us MUL_MAT_ID(ffn_moe_down-84): 288 us ggml_barrier(...): 21 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-85): 368 us MUL_MAT_ID(ffn_moe_down-85): 236 us ggml_barrier(...): 36 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-86): 395 us MUL_MAT_ID(ffn_moe_down-86): 325 us ggml_barrier(...): 29 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-87): 395 us MUL_MAT_ID(ffn_moe_down-87): 184 us ggml_barrier(...): 65 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-88): 417 us MUL_MAT_ID(ffn_moe_down-88): 221 us ggml_barrier(...): 283 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-89): 357 us MUL_MAT_ID(ffn_moe_down-89): 192 us ggml_barrier(...): 278 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-90): 350 us MUL_MAT_ID(ffn_moe_down-90): 228 us ggml_barrier(...): 57 us MOE_FUSED_UP_GATE(ffn_moe_gate_par-91): 376 us MUL_MAT_ID(ffn_moe_down-91): 230 us ggml_barrier(...): 32 us | glm4moe 355B.A32B Q3_K - Medium | 155.60 GiB | 369.08 B | CUDA | 94 | 8192 | 8192 | q8_0 | q8_0 | graph | 0.00 | tg128 | 7.43 ± 0.00 | build: f4a28688 (4135)