bash GGML_VK_PERF_LOGGER=1 GGML_VK_PERF_LOGGER_FREQUENCY=99999 ./llama-server -m /home/tipu/AI/models/ggml-org/Qwen3-Coder-30B-A3B/Qwen3-Coder-30B-A3B-Instruct-Q8_0.gguf --port 8888 --jinja --n-predict -1 --n-gpu-layers 99 --flash-attn off --temp 0.7 --top-k 20 --top-p 0.8 --min-p 0.01 --ctx-size 30768 --mlock --ubatch-size 512 --batch-size 2048 --cache-reuse 512 --cache-ram -1 --ctx-checkpoints 16 --offline --parallel 1 --kv-unified --context-shift --prio-batch 3 --prio 3 --alias Qwen3-Coder-30B-A3B --no-mmap -dio
ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = AMD Radeon Graphics (RADV RENOIR) (radv) | uma: 1 | fp16: 1 | bf16: 0 | warp size: 64 | shared memory: 65536 | int dot: 0 | matrix cores: none
build: 7554 (9be29e68d) with GNU 13.3.0 for Linux x86_64
system info: n_threads = 8, n_threads_batch = 8, total_threads = 16
system_info: n_threads = 8 (n_threads_batch = 8) / 16 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | OPENMP = 1 | REPACK = 1 |
Running without SSL
init: using 15 threads for HTTP server
start: binding port with default address family
main: loading model
srv load_model: loading model '/home/tipu/AI/models/ggml-org/Qwen3-Coder-30B-A3B/Qwen3-Coder-30B-A3B-Instruct-Q8_0.gguf'
common_init_result: fitting params to device memory, for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on
llama_params_fit_impl: projected to use 37522 MiB of device memory vs. 57344 MiB of free device memory
llama_params_fit_impl: will leave 18776 >= 1024 MiB of free device memory, no changes needed
llama_params_fit: successfully fit params to free device memory
llama_params_fit: fitting params to free memory took 0.25 seconds
llama_model_load_from_file_impl: using device Vulkan0 (AMD Radeon Graphics (RADV RENOIR)) (0000:03:00.0) - 56299 MiB free
llama_model_loader: loaded meta data with 44 key-value pairs and 579 tensors from /home/tipu/AI/models/ggml-org/Qwen3-Coder-30B-A3B/Qwen3-Coder-30B-A3B-Instruct-Q8_0.gguf (version GGUF V3 (latest))
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv 0: general.architecture str = qwen3moe
llama_model_loader: - kv 1: general.type str = model
llama_model_loader: - kv 2: general.name str = Qwen3-Coder-30B-A3B-Instruct
llama_model_loader: - kv 3: general.finetune str = Instruct
llama_model_loader: - kv 4: general.basename str = Qwen3-Coder-30B-A3B-Instruct
llama_model_loader: - kv 5: general.quantized_by str = Unsloth
llama_model_loader: - kv 6: general.size_label str = 30B-A3B
llama_model_loader: - kv 7: general.license str = apache-2.0
llama_model_loader: - kv 8: general.license.link str = https://huggingface.co/Qwen/Qwen3-Cod...
llama_model_loader: - kv 9: general.repo_url str = https://huggingface.co/unsloth
llama_model_loader: - kv 10: general.base_model.count u32 = 1
llama_model_loader: - kv 11: general.base_model.0.name str = Qwen3 Coder 30B A3B Instruct
llama_model_loader: - kv 12: general.base_model.0.organization str = Qwen
llama_model_loader: - kv 13: general.base_model.0.repo_url str = https://huggingface.co/Qwen/Qwen3-Cod...
llama_model_loader: - kv 14: general.tags arr[str,2] = ["unsloth", "text-generation"]
llama_model_loader: - kv 15: qwen3moe.block_count u32 = 48
llama_model_loader: - kv 16: qwen3moe.context_length u32 = 262144
llama_model_loader: - kv 17: qwen3moe.embedding_length u32 = 2048
llama_model_loader: - kv 18: qwen3moe.feed_forward_length u32 = 5472
llama_model_loader: - kv 19: qwen3moe.attention.head_count u32 = 32
llama_model_loader: - kv 20: qwen3moe.attention.head_count_kv u32 = 4
llama_model_loader: - kv 21: qwen3moe.rope.freq_base f32 = 10000000.000000
llama_model_loader: - kv 22: qwen3moe.attention.layer_norm_rms_epsilon f32 = 0.000001
llama_model_loader: - kv 23: qwen3moe.expert_used_count u32 = 8
llama_model_loader: - kv 24: qwen3moe.attention.key_length u32 = 128
llama_model_loader: - kv 25: qwen3moe.attention.value_length u32 = 128
llama_model_loader: - kv 26: qwen3moe.expert_count u32 = 128
llama_model_loader: - kv 27: qwen3moe.expert_feed_forward_length u32 = 768
llama_model_loader: - kv 28: qwen3moe.expert_shared_feed_forward_length u32 = 0
llama_model_loader: - kv 29: tokenizer.ggml.model str = gpt2
llama_model_loader: - kv 30: tokenizer.ggml.pre str = qwen2
llama_model_loader: - kv 31: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv 32: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv 33: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
llama_model_loader: - kv 34: tokenizer.ggml.eos_token_id u32 = 151645
llama_model_loader: - kv 35: tokenizer.ggml.padding_token_id u32 = 151654
llama_model_loader: - kv 36: tokenizer.ggml.add_bos_token bool = false
llama_model_loader: - kv 37: tokenizer.chat_template str = {# Copyright 2025-present Unsloth. Ap...
llama_model_loader: - kv 38: general.quantization_version u32 = 2
llama_model_loader: - kv 39: general.file_type u32 = 7
llama_model_loader: - kv 40: quantize.imatrix.file str = Qwen3-Coder-30B-A3B-Instruct-GGUF/ima...
llama_model_loader: - kv 41: quantize.imatrix.dataset str = unsloth_calibration_Qwen3-Coder-30B-A...
llama_model_loader: - kv 42: quantize.imatrix.entries_count u32 = 384
llama_model_loader: - kv 43: quantize.imatrix.chunks_count u32 = 154
llama_model_loader: - type f32: 241 tensors
llama_model_loader: - type q8_0: 338 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type = Q8_0
print_info: file size = 30.25 GiB (8.51 BPW)
load: printing all EOG tokens:
load: - 151643 ('<|endoftext|>')
load: - 151645 ('<|im_end|>')
load: - 151662 ('<|fim_pad|>')
load: - 151663 ('<|repo_name|>')
load: - 151664 ('<|file_sep|>')
load: special tokens cache size = 26
load: token to piece cache size = 0.9311 MB
print_info: arch = qwen3moe
print_info: vocab_only = 0
print_info: no_alloc = 0
print_info: n_ctx_train = 262144
print_info: n_embd = 2048
print_info: n_embd_inp = 2048
print_info: n_layer = 48
print_info: n_head = 32
print_info: n_head_kv = 4
print_info: n_rot = 128
print_info: n_swa = 0
print_info: is_swa_any = 0
print_info: n_embd_head_k = 128
print_info: n_embd_head_v = 128
print_info: n_gqa = 8
print_info: n_embd_k_gqa = 512
print_info: n_embd_v_gqa = 512
print_info: f_norm_eps = 0.0e+00
print_info: f_norm_rms_eps = 1.0e-06
print_info: f_clamp_kqv = 0.0e+00
print_info: f_max_alibi_bias = 0.0e+00
print_info: f_logit_scale = 0.0e+00
print_info: f_attn_scale = 0.0e+00
print_info: n_ff = 5472
print_info: n_expert = 128
print_info: n_expert_used = 8
print_info: n_expert_groups = 0
print_info: n_group_used = 0
print_info: causal attn = 1
print_info: pooling type = 0
print_info: rope type = 2
print_info: rope scaling = linear
print_info: freq_base_train = 10000000.0
print_info: freq_scale_train = 1
print_info: n_ctx_orig_yarn = 262144
print_info: rope_yarn_log_mul= 0.0000
print_info: rope_finetuned = unknown
print_info: model type = 30B.A3B
print_info: model params = 30.53 B
print_info: general.name = Qwen3-Coder-30B-A3B-Instruct
print_info: n_ff_exp = 768
print_info: vocab type = BPE
print_info: n_vocab = 151936
print_info: n_merges = 151387
print_info: BOS token = 11 ','
print_info: EOS token = 151645 '<|im_end|>'
print_info: EOT token = 151645 '<|im_end|>'
print_info: PAD token = 151654 '<|vision_pad|>'
print_info: LF token = 198 'Ċ'
print_info: FIM PRE token = 151659 '<|fim_prefix|>'
print_info: FIM SUF token = 151661 '<|fim_suffix|>'
print_info: FIM MID token = 151660 '<|fim_middle|>'
print_info: FIM PAD token = 151662 '<|fim_pad|>'
print_info: FIM REP token = 151663 '<|repo_name|>'
print_info: FIM SEP token = 151664 '<|file_sep|>'
print_info: EOG token = 151643 '<|endoftext|>'
print_info: EOG token = 151645 '<|im_end|>'
print_info: EOG token = 151662 '<|fim_pad|>'
print_info: EOG token = 151663 '<|repo_name|>'
print_info: EOG token = 151664 '<|file_sep|>'
print_info: max token length = 256
load_tensors: loading model tensors, this can take a while... (mmap = false, direct_io = true)
load_tensors: offloading output layer to GPU
load_tensors: offloading 47 repeating layers to GPU
load_tensors: offloaded 49/49 layers to GPU
load_tensors: Vulkan0 model buffer size = 30658.10 MiB
load_tensors: Vulkan_Host model buffer size = 315.30 MiB
...................................................................................................
common_init_result: added <|endoftext|> logit bias = -inf
common_init_result: added <|im_end|> logit bias = -inf
common_init_result: added <|fim_pad|> logit bias = -inf
common_init_result: added <|repo_name|> logit bias = -inf
common_init_result: added <|file_sep|> logit bias = -inf
llama_context: constructing llama_context
llama_context: n_seq_max = 1
llama_context: n_ctx = 30976
llama_context: n_ctx_seq = 30976
llama_context: n_batch = 2048
llama_context: n_ubatch = 512
llama_context: causal_attn = 1
llama_context: flash_attn = disabled
llama_context: kv_unified = true
llama_context: freq_base = 10000000.0
llama_context: freq_scale = 1
llama_context: n_ctx_seq (30976) < n_ctx_train (262144) -- the full capacity of the model will not be utilized
llama_context: Vulkan_Host output buffer size = 0.58 MiB
llama_kv_cache: Vulkan0 KV buffer size = 2904.00 MiB
llama_kv_cache: size = 2904.00 MiB ( 30976 cells, 48 layers, 1/1 seqs), K (f16): 1452.00 MiB, V (f16): 1452.00 MiB
llama_context: Vulkan0 compute buffer size = 3960.51 MiB
llama_context: Vulkan_Host compute buffer size = 66.52 MiB
llama_context: graph nodes = 3270
llama_context: graph splits = 2
common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
srv load_model: initializing slots, n_slots = 1
slot load_model: id 0 | task -1 | new slot, n_ctx = 30976
srv load_model: prompt cache is enabled, size limit: no limit
srv load_model: use `--cache-ram 0` to disable the prompt cache
srv load_model: for more info see https://github.com/ggml-org/llama.cpp/pull/16391
srv load_model: thinking = 0
load_model: chat template, chat_template: {# Copyright 2025-present Unsloth. Apache 2.0 License. Unsloth Chat template fixes #}
{% macro render_item_list(item_list, tag_name='required') %}
{%- if item_list is defined and item_list is iterable and item_list | length > 0 %}
{%- if tag_name %}{{- '\n<' ~ tag_name ~ '>' -}}{% endif %}
{{- '[' }}
{%- for item in item_list -%}
{%- if loop.index > 1 %}{{- ", "}}{% endif -%}
{%- if item is string -%}
{{ "`" ~ item ~ "`" }}
{%- else -%}
{{ item }}
{%- endif -%}
{%- endfor -%}
{{- ']' }}
{%- if tag_name %}{{- '' -}}{% endif %}
{%- endif %}
{% endmacro %}
{%- if messages[0]["role"] == "system" %}
{%- set system_message = messages[0]["content"] %}
{%- set loop_messages = messages[1:] %}
{%- else %}
{%- set loop_messages = messages %}
{%- endif %}
{%- if not tools is defined %}
{%- set tools = [] %}
{%- endif %}
{%- if system_message is defined %}
{{- "<|im_start|>system\n" + system_message }}
{%- else %}
{%- if tools is iterable and tools | length > 0 %}
{{- "<|im_start|>system\nYou are Qwen, a helpful AI assistant that can interact with a computer to solve tasks." }}
{%- endif %}
{%- endif %}
{%- if tools is iterable and tools | length > 0 %}
{{- "\n\nYou have access to the following functions:\n\n" }}
{{- "" }}
{%- for tool in tools %}
{%- if tool.function is defined %}
{%- set tool = tool.function %}
{%- endif %}
{{- "\n\n" ~ tool.name ~ "" }}
{{- '\n' ~ (tool.description | trim) ~ '' }}
{{- '\n' }}
{%- for param_name, param_fields in tool.parameters.properties|items %}
{{- '\n' }}
{{- '\n' ~ param_name ~ '' }}
{%- if param_fields.type is defined %}
{{- '\n' ~ (param_fields.type | string) ~ '' }}
{%- endif %}
{%- if param_fields.description is defined %}
{{- '\n' ~ (param_fields.description | trim) ~ '' }}
{%- endif %}
{{- render_item_list(param_fields.enum, 'enum') }}
{%- set handled_keys = ['type', 'description', 'enum', 'required'] %}
{%- for json_key, json_value in param_fields|items %}
{%- if json_key not in handled_keys %}
{%- set normed_json_key = json_key|string %}
{%- if json_value is mapping %}
{{- '\n<' ~ normed_json_key ~ '>' ~ (json_value | tojson | safe) ~ '' }}
{%- else %}
{{- '\n<' ~ normed_json_key ~ '>' ~ (json_value | string) ~ '' }}
{%- endif %}
{%- endif %}
{%- endfor %}
{{- render_item_list(param_fields.required, 'required') }}
{{- '\n' }}
{%- endfor %}
{{- render_item_list(tool.parameters.required, 'required') }}
{{- '\n' }}
{%- if tool.return is defined %}
{%- if tool.return is mapping %}
{{- '\n' ~ (tool.return | tojson | safe) ~ '' }}
{%- else %}
{{- '\n' ~ (tool.return | string) ~ '' }}
{%- endif %}
{%- endif %}
{{- '\n' }}
{%- endfor %}
{{- "\n" }}
{{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n\n\n\nvalue_1\n\n\nThis is the value for the second parameter\nthat can span\nmultiple lines\n\n\n\n\n\nReminder:\n- Function calls MUST follow the specified format: an inner block must be nested within XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n' }}
{%- endif %}
{%- if system_message is defined %}
{{- '<|im_end|>\n' }}
{%- else %}
{%- if tools is iterable and tools | length > 0 %}
{{- '<|im_end|>\n' }}
{%- endif %}
{%- endif %}
{%- for message in loop_messages %}
{%- if message.role == "assistant" and message.tool_calls is defined and message.tool_calls is iterable and message.tool_calls | length > 0 %}
{{- '<|im_start|>' + message.role }}
{%- if message.content is defined and message.content is string and message.content | trim | length > 0 %}
{{- '\n' + message.content | trim + '\n' }}
{%- endif %}
{%- for tool_call in message.tool_calls %}
{%- if tool_call.function is defined %}
{%- set tool_call = tool_call.function %}
{%- endif %}
{{- '\n\n\n' }}
{%- if tool_call.arguments is defined %}
{%- for args_name, args_value in tool_call.arguments|items %}
{{- '\n' }}
{%- set args_value = args_value if args_value is string else args_value | string %}
{{- args_value }}
{{- '\n\n' }}
{%- endfor %}
{%- endif %}
{{- '\n' }}
{%- endfor %}
{{- '<|im_end|>\n' }}
{%- elif message.role == "user" or message.role == "system" or message.role == "assistant" %}
{{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
{%- elif message.role == "tool" %}
{%- if loop.previtem and loop.previtem.role != "tool" %}
{{- '<|im_start|>user\n' }}
{%- endif %}
{{- '\n' }}
{{- message.content }}
{{- '\n\n' }}
{%- if not loop.last and loop.nextitem.role != "tool" %}
{{- '<|im_end|>\n' }}
{%- elif loop.last %}
{{- '<|im_end|>\n' }}
{%- endif %}
{%- else %}
{{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>\n' }}
{%- endif %}
{%- endfor %}
{%- if add_generation_prompt %}
{{- '<|im_start|>assistant\n' }}
{%- endif %}
{# Copyright 2025-present Unsloth. Apache 2.0 License. Unsloth Chat template fixes #}, example_format: '<|im_start|>system
You are a helpful assistant<|im_end|>
<|im_start|>user
Hello<|im_end|>
<|im_start|>assistant
Hi there<|im_end|>
<|im_start|>user
How are you?<|im_end|>
<|im_start|>assistant
'
main: model loaded
main: server is listening on http://127.0.0.1:8888
main: starting the main loop...
srv update_slots: all slots are idle
srv params_from_: Chat format: Qwen3 Coder
slot get_availabl: id 0 | task -1 | selected slot by LRU, t_last = -1
slot launch_slot_: id 0 | task -1 | sampler chain: logits -> penalties -> dry -> top-n-sigma -> top-k -> typical -> top-p -> min-p -> xtc -> temp-ext -> dist
slot launch_slot_: id 0 | task 0 | processing task
slot update_slots: id 0 | task 0 | new prompt, n_ctx_slot = 30976, n_keep = 0, task.n_tokens = 16
slot update_slots: id 0 | task 0 | n_tokens = 0, memory_seq_rm [0, end)
slot update_slots: id 0 | task 0 | prompt processing progress, n_tokens = 16, batch.n_tokens = 16, progress = 1.000000
slot update_slots: id 0 | task 0 | prompt done, n_tokens = 16, batch.n_tokens = 16
slot print_timing: id 0 | task 0 |
prompt eval time = 734.46 ms / 16 tokens ( 45.90 ms per token, 21.78 tokens per second)
eval time = 60560.01 ms / 685 tokens ( 88.41 ms per token, 11.31 tokens per second)
total time = 61294.48 ms / 701 tokens
slot release: id 0 | task 0 | stop processing: n_tokens = 700, truncated = 0
srv update_slots: all slots are idle
srv log_server_r: request: POST /v1/chat/completions 127.0.0.1 200
^Csrv operator(): operator(): cleaning up before exit...
llama_memory_breakdown_print: | memory breakdown [MiB] | total free self model context compute unaccounted |
llama_memory_breakdown_print: | - Vulkan0 (Graphics (RADV RENOIR)) | 57344 = 18375 + (37522 = 30658 + 2904 + 3960) + 1445 |
llama_memory_breakdown_print: | - Host | 381 = 315 + 0 + 66 |
----------------
Vulkan Timings:
ADD: 780 x 4.393 us = 3427.08 us
CONT: 32928 x 2.97 us = 97816.6 us
GET_ROWS: 1372 x 4.293 us = 5890.4 us
GLU: 32928 x 4.277 us = 140842 us
MUL: 94 x 133.467 us = 12545.9 us
MULTI_ADD ADD: 32928 x 4.406 us = 145098 us
MUL_MAT f16 m=128 n=16 k=256 batch=32: 48 x 100.056 us = 4802.72 us (334.699 GFLOPS/s)
MUL_MAT f16 m=256 n=16 k=128 batch=32: 48 x 124.13 us = 5958.24 us (269.261 GFLOPS/s)
MUL_MAT f32 m=128 n=16 k=2048: 47 x 181.981 us = 8553.12 us (46.0847 GFLOPS/s)
MUL_MAT q8_0 m=2048 n=16 k=4096: 48 x 668.689 us = 32097.1 us (401.386 GFLOPS/s)
MUL_MAT q8_0 m=4096 n=16 k=2048: 48 x 659.058 us = 31634.8 us (407.202 GFLOPS/s)
MUL_MAT q8_0 m=512 n=16 k=2048: 96 x 0.334 us = 32.12 us (100263 GFLOPS/s)
MUL_MAT_ADD MUL_MAT_VEC q8_0 m=2048 n=1 k=4096: 32148 x 222.829 us = 7.16353e+06 us (75.2824 GFLOPS/s)
MUL_MAT_ID q8_0 m=2048 n=128 k=768 n_expert=128 batch=2: 47 x 10624.6 us = 499356 us (75.7471 GFLOPS/s)
MUL_MAT_ID q8_0 m=2048 n=8 k=768 n_expert=128 batch=16: 47 x 4394.88 us = 206559 us (91.5591 GFLOPS/s)
MUL_MAT_ID q8_0 m=768 n=128 k=2048 n_expert=128 batch=2: 94 x 10526.6 us = 989500 us (76.4834 GFLOPS/s)
MUL_MAT_ID q8_0 m=768 n=8 k=2048 n_expert=128 batch=16: 94 x 4349.88 us = 408889 us (92.5439 GFLOPS/s)
MUL_MAT_ID_MUL MUL_MAT_ID_VEC q8_0 m=2048 n=128 k=768 n_expert=128: 1 x 4685.92 us = 4685.92 us (85.8724 GFLOPS/s)
MUL_MAT_ID_MUL MUL_MAT_ID_VEC q8_0 m=2048 n=8 k=768 n_expert=128: 32833 x 316.897 us = 1.04047e+07 us (79.3615 GFLOPS/s)
MUL_MAT_ID_VEC q8_0 m=768 n=128 k=2048 n_expert=128: 2 x 4861.38 us = 9722.76 us (82.8067 GFLOPS/s)
MUL_MAT_ID_VEC q8_0 m=768 n=8 k=2048 n_expert=128: 65666 x 323.983 us = 2.12747e+07 us (77.6573 GFLOPS/s)
MUL_MAT_VEC f16 m=128 n=1 k=256 batch=32: 11520 x 30.612 us = 352656 us (68.3727 GFLOPS/s)
MUL_MAT_VEC f16 m=128 n=1 k=512 batch=32: 12288 x 46.716 us = 574057 us (89.6936 GFLOPS/s)
MUL_MAT_VEC f16 m=128 n=1 k=768 batch=32: 9024 x 66.831 us = 603087 us (94.0779 GFLOPS/s)
MUL_MAT_VEC f16 m=128 n=2 k=256 batch=32: 48 x 127.955 us = 6141.84 us (32.7155 GFLOPS/s)
MUL_MAT_VEC f16 m=256 n=1 k=128 batch=32: 11520 x 24.281 us = 279721 us (86.0314 GFLOPS/s)
MUL_MAT_VEC f16 m=256 n=2 k=128 batch=32: 48 x 188.656 us = 9055.52 us (22.1456 GFLOPS/s)
MUL_MAT_VEC f16 m=512 n=1 k=128 batch=32: 12288 x 39.648 us = 487201 us (105.374 GFLOPS/s)
MUL_MAT_VEC f16 m=768 n=1 k=128 batch=32: 9024 x 55.247 us = 498551 us (113.433 GFLOPS/s)
MUL_MAT_VEC f32 m=128 n=1 k=2048: 32834 x 28.037 us = 920576 us (18.6951 GFLOPS/s)
MUL_MAT_VEC f32 m=128 n=2 k=2048: 47 x 27.9 us = 1311.32 us (37.5736 GFLOPS/s)
MUL_MAT_VEC q8_0 m=151936 n=1 k=2048: 686 x 7977.48 us = 5.47255e+06 us (77.9918 GFLOPS/s)
MUL_MAT_VEC q8_0 m=2048 n=1 k=4096: 684 x 221.86 us = 151752 us (75.6114 GFLOPS/s)
MUL_MAT_VEC q8_0 m=2048 n=2 k=4096: 48 x 223.284 us = 10717.6 us (150.258 GFLOPS/s)
MUL_MAT_VEC q8_0 m=4096 n=1 k=2048: 32832 x 229.88 us = 7.54743e+06 us (72.9646 GFLOPS/s)
MUL_MAT_VEC q8_0 m=4096 n=2 k=2048: 48 x 223.579 us = 10731.8 us (150.042 GFLOPS/s)
MUL_MAT_VEC q8_0 m=512 n=1 k=2048: 65664 x 22.038 us = 1.44716e+06 us (95.1336 GFLOPS/s)
MUL_MAT_VEC q8_0 m=512 n=2 k=2048: 96 x 21.27 us = 2041.92 us (197.145 GFLOPS/s)
RMS_NORM_MUL RMS_NORM(2048,1,1,1): 66352 x 5.374 us = 356620 us
RMS_NORM_MUL RMS_NORM(2048,16,1,1): 95 x 14.966 us = 1421.84 us
RMS_NORM_MUL RMS_NORM(2048,2,1,1): 95 x 7.029 us = 667.76 us
RMS_NORM_MUL_ROPE RMS_NORM(128,32,1,1): 32832 x 11.591 us = 380571 us
RMS_NORM_MUL_ROPE RMS_NORM(128,32,16,1): 48 x 65.856 us = 3161.12 us
RMS_NORM_MUL_ROPE RMS_NORM(128,32,2,1): 48 x 14.922 us = 716.28 us
RMS_NORM_MUL_ROPE_VIEW_SET_ROWS RMS_NORM(128,4,1,1): 32832 x 7.34 us = 241009 us
RMS_NORM_MUL_ROPE_VIEW_SET_ROWS RMS_NORM(128,4,16,1): 48 x 14.377 us = 690.12 us
RMS_NORM_MUL_ROPE_VIEW_SET_ROWS RMS_NORM(128,4,2,1): 48 x 11.031 us = 529.52 us
SET_ROWS: 32928 x 0.486 us = 16033.5 us
SOFT_MAX: 32928 x 10.865 us = 357788 us
TOPK_MOE_EARLY_SOFTMAX_NORM SOFT_MAX: 32928 x 6.266 us = 206346 us
Total time: 6.13906e+07 us.