[Bug]: Qwen3-Coder-480B-FP8 running on 8*H20,deepgemm warmup OOM #29360
Description
Your current environment
The output of python collect_env.py
Collecting environment information...
==============================
System Info
==============================
OS : Ubuntu 22.04.5 LTS (x86_64)
GCC version : (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version : Could not collect
CMake version : version 4.1.2
Libc version : glibc-2.35
==============================
PyTorch Info
==============================
PyTorch version : 2.9.0+cu129
Is debug build : False
CUDA used to build PyTorch : 12.9
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.12.12 (main, Oct 10 2025, 08:52:57) [GCC 11.4.0] (64-bit runtime)
Python platform : Linux-6.8.0-60-generic-x86_64-with-glibc2.35
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : 12.9.86
CUDA_MODULE_LOADING set to :
GPU models and configuration :
GPU 0: NVIDIA H20
GPU 1: NVIDIA H20
GPU 2: NVIDIA H20
GPU 3: NVIDIA H20
GPU 4: NVIDIA H20
GPU 5: NVIDIA H20
GPU 6: NVIDIA H20
GPU 7: NVIDIA H20
Nvidia driver version : 560.35.03
cuDNN version : Could not collect
HIP runtime version : N/A
MIOpen runtime version : N/A
Is XNNPACK available : True
==============================
CPU Info
==============================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 46 bits physical, 57 bits virtual
Byte Order: Little Endian
CPU(s): 192
On-line CPU(s) list: 0-191
Vendor ID: GenuineIntel
Model name: Intel(R) Xeon(R) Platinum 8468
CPU family: 6
Model: 143
Thread(s) per core: 2
Core(s) per socket: 48
Socket(s): 2
Stepping: 8
CPU max MHz: 3800.0000
CPU min MHz: 800.0000
BogoMIPS: 4200.00
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr
sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aper
fmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic mo
vbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 cdp_l2 ssbd mba i
brs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx5
12f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_ll
c cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect user_shstk avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts vnmi avx5
12vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid bus_lock_detec
t cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr ibt amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l
1d arch_capabilities
Virtualization: VT-x
L1d cache: 4.5 MiB (96 instances)
L1i cache: 3 MiB (96 instances)
L2 cache: 192 MiB (96 instances)
L3 cache: 210 MiB (2 instances)
NUMA node(s): 2
NUMA node0 CPU(s): 0,2,4,6,8,10,12,14,16,18,20,22,24,26,28,30,32,34,36,38,40,42,44,46,48,50,52,54,56,58,60,62,64,66,6
8,70,72,74,76,78,80,82,84,86,88,90,92,94,96,98,100,102,104,106,108,110,112,114,116,118,120,122,124,126,128,130,132,134,136,138,140,142,1
44,146,148,150,152,154,156,158,160,162,164,166,168,170,172,174,176,178,180,182,184,186,188,190
NUMA node1 CPU(s): 1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47,49,51,53,55,57,59,61,63,65,67,6
9,71,73,75,77,79,81,83,85,87,89,91,93,95,97,99,101,103,105,107,109,111,113,115,117,119,121,123,125,127,129,131,133,135,137,139,141,143,1
45,147,149,151,153,155,157,159,161,163,165,167,169,171,173,175,177,179,181,183,185,187,189,191
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS SW sequence; BHI
BHI_DIS_S
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
==============================
Versions of relevant libraries
==============================
[pip3] flashinfer-python==0.5.2
[pip3] numpy==2.2.0
[pip3] nvidia-cublas-cu12==12.9.1.4
[pip3] nvidia-cuda-cupti-cu12==12.9.79
[pip3] nvidia-cuda-nvrtc-cu12==12.9.86
[pip3] nvidia-cuda-runtime-cu12==12.9.79
[pip3] nvidia-cudnn-cu12==9.10.2.21
[pip3] nvidia-cudnn-frontend==1.16.0
[pip3] nvidia-cufft-cu12==11.4.1.4
[pip3] nvidia-cufile-cu12==1.14.1.1
[pip3] nvidia-curand-cu12==10.3.10.19
[pip3] nvidia-cusolver-cu12==11.7.5.82
[pip3] nvidia-cusparse-cu12==12.5.10.65
[pip3] nvidia-cusparselt-cu12==0.7.1
[pip3] nvidia-cutlass-dsl==4.2.1
[pip3] nvidia-ml-py==13.580.82
[pip3] nvidia-nccl-cu12==2.27.5
[pip3] nvidia-nvjitlink-cu12==12.9.86
[pip3] nvidia-nvshmem-cu12==3.3.20
[pip3] nvidia-nvtx-cu12==12.9.79
[pip3] pyzmq==27.1.0
[pip3] torch==2.9.0+cu129
[pip3] torchaudio==2.9.0+cu129
[pip3] torchvision==0.24.0+cu129
[pip3] transformers==4.57.1
[pip3] triton==3.5.0
[conda] Could not collect
==============================
vLLM Info
==============================
ROCM Version : Could not collect
vLLM Version : 0.11.1
vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled
GPU Topology:
GPU0 GPU1 GPU2 GPU3 GPU4 GPU5 GPU6 GPU7 CPU Affinity NUMA Affinity GPU NUMA ID
GPU0 X NV18 NV18 NV18 NV18 NV18 NV18 NV18 0,2,4,6,8,10 0 N/A
GPU1 NV18 X NV18 NV18 NV18 NV18 NV18 NV18 0,2,4,6,8,10 0 N/A
GPU2 NV18 NV18 X NV18 NV18 NV18 NV18 NV18 0,2,4,6,8,10 0 N/A
GPU3 NV18 NV18 NV18 X NV18 NV18 NV18 NV18 0,2,4,6,8,10 0 N/A
GPU4 NV18 NV18 NV18 NV18 X NV18 NV18 NV18 1,3,5,7,9,11 1 N/A
GPU5 NV18 NV18 NV18 NV18 NV18 X NV18 NV18 1,3,5,7,9,11 1 N/A
GPU6 NV18 NV18 NV18 NV18 NV18 NV18 X NV18 1,3,5,7,9,11 1 N/A
GPU7 NV18 NV18 NV18 NV18 NV18 NV18 NV18 X 1,3,5,7,9,11 1 N/A
Legend:
X = Self
SYS = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
PHB = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
PXB = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
PIX = Connection traversing at most a single PCIe bridge
NV# = Connection traversing a bonded set of # NVLinks
==============================
Environment Variables
==============================
NCCL_VERSION=2.27.3-1
NVIDIA_PRODUCT_NAME=CUDA
VLLM_USAGE_SOURCE=production-docker-image
LD_LIBRARY_PATH=/usr/local/nvidia/lib64:/usr/local/cuda/lib64
NVIDIA_VISIBLE_DEVICES=GPU-cc85819e-2755-7177-7d17-b72a8a43f494,GPU-5d97824c-a6d6-40e0-7b64-fe7a0f616062,GPU-092a1c67-4fce-ebfc-61c2-f25
bac439bd7,GPU-e346dd6d-4a6c-85f9-92d9-3e79cecca803,GPU-5ba91686-0304-8fad-7c5d-c26adcf43852,GPU-457b31ce-43c4-8ca1-89e9-131bd9a4dc52,GPU
-28dee1d9-af6e-4479-1343-1fe5dd039720,GPU-ffdbed5e-dc33-f1de-448b-eb4cfb23ca84
NVIDIA_DRIVER_CAPABILITIES=compute,utility
CUDA_VERSION=12.9.1
PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True
NVIDIA_REQUIRE_CUDA=cuda>=12.9 brand=unknown,driver>=535,driver<536 brand=grid,driver>=535,driver<536 brand=tesla,driver>=535,driver<536
brand=nvidia,driver>=535,driver<536 brand=quadro,driver>=535,driver<536 brand=quadrortx,driver>=535,driver<536 brand=nvidiartx,driver>=
535,driver<536 brand=vapps,driver>=535,driver<536 brand=vpc,driver>=535,driver<536 brand=vcs,driver>=535,driver<536 brand=vws,driver>=53
5,driver<536 brand=cloudgaming,driver>=535,driver<536 brand=unknown,driver>=550,driver<551 brand=grid,driver>=550,driver<551 brand=tesla
,driver>=550,driver<551 brand=nvidia,driver>=550,driver<551 brand=quadro,driver>=550,driver<551 brand=quadrortx,driver>=550,driver<551 b
rand=nvidiartx,driver>=550,driver<551 brand=vapps,driver>=550,driver<551 brand=vpc,driver>=550,driver<551 brand=vcs,driver>=550,driver<5
51 brand=vws,driver>=550,driver<551 brand=cloudgaming,driver>=550,driver<551 brand=unknown,driver>=560,driver<561 brand=grid,driver>=560
,driver<561 brand=tesla,driver>=560,driver<561 brand=nvidia,driver>=560,driver<561 brand=quadro,driver>=560,driver<561 brand=quadrortx,d
river>=560,driver<561 brand=nvidiartx,driver>=560,driver<561 brand=vapps,driver>=560,driver<561 brand=vpc,driver>=560,driver<561 brand=v
cs,driver>=560,driver<561 brand=vws,driver>=560,driver<561 brand=cloudgaming,driver>=560,driver<561 brand=unknown,driver>=565,driver<566
brand=grid,driver>=565,driver<566 brand=tesla,driver>=565,driver<566 brand=nvidia,driver>=565,driver<566 brand=quadro,driver>=565,drive
r<566 brand=quadrortx,driver>=565,driver<566 brand=nvidiartx,driver>=565,driver<566 brand=vapps,driver>=565,driver<566 brand=vpc,driver>
=565,driver<566 brand=vcs,driver>=565,driver<566 brand=vws,driver>=565,driver<566 brand=cloudgaming,driver>=565,driver<566 brand=unknown
,driver>=570,driver<571 brand=grid,driver>=570,driver<571 brand=tesla,driver>=570,driver<571 brand=nvidia,driver>=570,driver<571 brand=q
uadro,driver>=570,driver<571 brand=quadrortx,driver>=570,driver<571 brand=nvidiartx,driver>=570,driver<571 brand=vapps,driver>=570,drive
r<571 brand=vpc,driver>=570,driver<571 brand=vcs,driver>=570,driver<571 brand=vws,driver>=570,driver<571 brand=cloudgaming,driver>=570,d
river<571
CUDA_HOME=/usr/local/cuda
CUDA_HOME=/usr/local/cuda
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
[W1125 09:14:44.842352278 AllocatorConfig.cpp:28] Warning: PYTORCH_CUDA_ALLOC_CONF is deprecated, use PYTORCH_ALLOC_CONF instead (functi
on operator())
🐛 Describe the bug
INFO 11-25 08:00:36 [scheduler.py:216] Chunked prefill is enabled with max_num_batched_tokens=2048.
(APIServer pid=1) INFO 11-25 08:00:36 [api_server.py:1977] vLLM API server version 0.11.1
(APIServer pid=1) INFO 11-25 08:00:36 [utils.py:253] non-default args: {'model_tag': '/models', 'port': 8003, 'uvicorn_log_level': '
debug', 'enable_auto_tool_choice': True, 'tool_call_parser': 'qwen3_coder', 'model': '/models', 'max_model_len': 131072, 'enforce_ea
ger': True, 'served_model_name': ['qwen3-coder-480b'], 'tensor_parallel_size': 8, 'enable_expert_parallel': True, 'gpu_memory_utili
zation': 0.98, 'max_num_seqs': 30, 'enable_chunked_prefill': True, 'enable_log_requests': True}
(APIServer pid=1) INFO 11-25 08:00:42 [model.py:631] Resolved architecture: Qwen3MoeForCausalLM
(APIServer pid=1) INFO 11-25 08:00:42 [model.py:1745] Using max model len 131072
(APIServer pid=1) INFO 11-25 08:00:42 [scheduler.py:216] Chunked prefill is enabled with max_num_batched_tokens=8192.
(APIServer pid=1) INFO 11-25 08:00:42 [vllm.py:500] Cudagraph is disabled under eager mode
(EngineCore_DP0 pid=270) INFO 11-25 08:00:48 [core.py:93] Initializing a V1 LLM engine (v0.11.1) with config: model='/models', specu
lative_config=None, tokenizer='/models', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, tokenizer_revision=None, tru
st_remote_code=False, dtype=torch.bfloat16, max_seq_len=131072, download_dir=None, load_format=auto, tensor_parallel_size=8, pipeline_pa
rallel_size=1, data_parallel_size=1, disable_custom_all_reduce=False, quantization=fp8, enforce_eager=True, kv_cache_dtype=auto, device_
config=cuda, structured_outputs_config=StructuredOutputsConfig(backend='auto', disable_fallback=False, disable_any_whitespace=False, dis
able_additional_properties=False, reasoning_parser='', reasoning_parser_plugin='', enable_in_reasoning=False), observability_config=Obse
rvabilityConfig(show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None), seed=0, served_model_nam
e=qwen3-coder-480b, enable_prefix_caching=True, enable_chunked_prefill=True, pooler_config=None, compilation_config={'level': None,
'mode': <CompilationMode.NONE: 0>, 'debug_dump_path': None, 'cache_dir': '', 'compile_cache_save_format': 'binary', 'backend': 'inducto
r', 'custom_ops': ['all', '+quant_fp8'], 'splitting_ops': None, 'compile_mm_encoder': False, 'use_inductor': None, 'compile_sizes': [],
'inductor_compile_config': {'enable_auto_functionalized_v2': False, 'combo_kernels': True, 'benchmark_combo_kernel': True}, 'inductor_pa
sses': {}, 'cudagraph_mode': <CUDAGraphMode.NONE: 0>, 'cudagraph_num_of_warmups': 0, 'cudagraph_capture_sizes': [], 'cudagraph_copy_inpu
ts': False, 'cudagraph_specialize_lora': True, 'use_inductor_graph_partition': False, 'pass_config': {}, 'max_cudagraph_capture_size': 0
, 'local_cache_dir': None}
(EngineCore_DP0 pid=270) WARNING 11-25 08:00:48 [multiproc_executor.py:869] Reducing Torch parallelism from 96 threads to 1 to avoid unn
ecessary CPU contention. Set OMP_NUM_THREADS in the external environment to tune this value as needed.
INFO 11-25 08:01:01 [parallel_state.py:1208] world_size=8 rank=2 local_rank=2 distributed_init_method=tcp://127.0.0.1:34667 backend=nccl
INFO 11-25 08:01:01 [parallel_state.py:1208] world_size=8 rank=5 local_rank=5 distributed_init_method=tcp://127.0.0.1:34667 backend=nccl
INFO 11-25 08:01:01 [parallel_state.py:1208] world_size=8 rank=0 local_rank=0 distributed_init_method=tcp://127.0.0.1:34667 backend=nccl
INFO 11-25 08:01:01 [parallel_state.py:1208] world_size=8 rank=6 local_rank=6 distributed_init_method=tcp://127.0.0.1:34667 backend=nccl
INFO 11-25 08:01:01 [parallel_state.py:1208] world_size=8 rank=3 local_rank=3 distributed_init_method=tcp://127.0.0.1:34667 backend=nccl
INFO 11-25 08:01:01 [parallel_state.py:1208] world_size=8 rank=7 local_rank=7 distributed_init_method=tcp://127.0.0.1:34667 backend=nccl
INFO 11-25 08:01:01 [parallel_state.py:1208] world_size=8 rank=1 local_rank=1 distributed_init_method=tcp://127.0.0.1:34667 backend=nccl
INFO 11-25 08:01:01 [parallel_state.py:1208] world_size=8 rank=4 local_rank=4 distributed_init_method=tcp://127.0.0.1:34667 backend=nccl
[Gloo] Rank 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 2 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 3 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 1 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 4 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 5 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 7 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 6 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 1 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 5 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 2 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 4 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 3 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 7 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 6 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
INFO 11-25 08:01:02 [pynccl.py:111] vLLM is using nccl==2.27.5
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
[Gloo] Rank 1 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 2 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 3 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 4 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 6 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 5 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
[Gloo] Rank 7 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
INFO 11-25 08:01:07 [parallel_state.py:1394] rank 4 in world size 8 is assigned as DP rank 0, PP rank 0, TP rank 4, EP rank 4
INFO 11-25 08:01:07 [parallel_state.py:1394] rank 6 in world size 8 is assigned as DP rank 0, PP rank 0, TP rank 6, EP rank 6
INFO 11-25 08:01:07 [parallel_state.py:1394] rank 0 in world size 8 is assigned as DP rank 0, PP rank 0, TP rank 0, EP rank 0
INFO 11-25 08:01:07 [parallel_state.py:1394] rank 3 in world size 8 is assigned as DP rank 0, PP rank 0, TP rank 3, EP rank 3
INFO 11-25 08:01:07 [parallel_state.py:1394] rank 5 in world size 8 is assigned as DP rank 0, PP rank 0, TP rank 5, EP rank 5
INFO 11-25 08:01:07 [parallel_state.py:1394] rank 7 in world size 8 is assigned as DP rank 0, PP rank 0, TP rank 7, EP rank 7
INFO 11-25 08:01:07 [parallel_state.py:1394] rank 2 in world size 8 is assigned as DP rank 0, PP rank 0, TP rank 2, EP rank 2
INFO 11-25 08:01:07 [parallel_state.py:1394] rank 1 in world size 8 is assigned as DP rank 0, PP rank 0, TP rank 1, EP rank 1
(Worker_TP0_EP0 pid=403) INFO 11-25 08:01:07 [gpu_model_runner.py:3255] Starting to load model /app/models...
(Worker_TP6_EP6 pid=409) INFO 11-25 08:01:08 [deep_gemm.py:76] DeepGEMM E8M0 enabled on current platform.
(Worker_TP7_EP7 pid=410) INFO 11-25 08:01:08 [deep_gemm.py:76] DeepGEMM E8M0 enabled on current platform.
(Worker_TP5_EP5 pid=408) INFO 11-25 08:01:08 [deep_gemm.py:76] DeepGEMM E8M0 enabled on current platform.
(Worker_TP4_EP4 pid=407) INFO 11-25 08:01:08 [deep_gemm.py:76] DeepGEMM E8M0 enabled on current platform.
(Worker_TP2_EP2 pid=405) INFO 11-25 08:01:08 [deep_gemm.py:76] DeepGEMM E8M0 enabled on current platform.
(Worker_TP0_EP0 pid=403) INFO 11-25 08:01:08 [deep_gemm.py:76] DeepGEMM E8M0 enabled on current platform.
(Worker_TP3_EP3 pid=406) INFO 11-25 08:01:08 [deep_gemm.py:76] DeepGEMM E8M0 enabled on current platform.
(Worker_TP1_EP1 pid=404) INFO 11-25 08:01:08 [deep_gemm.py:76] DeepGEMM E8M0 enabled on current platform.
[W1125 08:01:24.672657737 AllocatorConfig.cpp:28] Warning: PYTORCH_CUDA_ALLOC_CONF is deprecated, use PYTORCH_ALLOC_CONF instead (functi
on operator())
[W1125 08:01:24.751603340 AllocatorConfig.cpp:28] Warning: PYTORCH_CUDA_ALLOC_CONF is deprecated, use PYTORCH_ALLOC_CONF instead (functi
on operator())
[W1125 08:01:24.790234425 AllocatorConfig.cpp:28] Warning: PYTORCH_CUDA_ALLOC_CONF is deprecated, use PYTORCH_ALLOC_CONF instead (functi
on operator())
[W1125 08:01:24.795054355 AllocatorConfig.cpp:28] Warning: PYTORCH_CUDA_ALLOC_CONF is deprecated, use PYTORCH_ALLOC_CONF instead (functi
on operator())
[W1125 08:01:24.798832445 AllocatorConfig.cpp:28] Warning: PYTORCH_CUDA_ALLOC_CONF is deprecated, use PYTORCH_ALLOC_CONF instead (functi
on operator())
[W1125 08:01:24.799420173 AllocatorConfig.cpp:28] Warning: PYTORCH_CUDA_ALLOC_CONF is deprecated, use PYTORCH_ALLOC_CONF instead (functi
on operator())
[W1125 08:01:24.799967632 AllocatorConfig.cpp:28] Warning: PYTORCH_CUDA_ALLOC_CONF is deprecated, use PYTORCH_ALLOC_CONF instead (functi
on operator())
[W1125 08:01:24.803973851 AllocatorConfig.cpp:28] Warning: PYTORCH_CUDA_ALLOC_CONF is deprecated, use PYTORCH_ALLOC_CONF instead (functi
on operator())
(Worker_TP6_EP6 pid=409) INFO 11-25 08:01:24 [cuda.py:418] Valid backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION']
(Worker_TP6_EP6 pid=409) INFO 11-25 08:01:24 [cuda.py:427] Using FLASH_ATTN backend.
(Worker_TP6_EP6 pid=409) INFO 11-25 08:01:24 [layer.py:342] Enabled separate cuda stream for MoE shared_experts
(Worker_TP6_EP6 pid=409) INFO 11-25 08:01:25 [layer.py:468] [EP Rank 6/8] Expert parallelism is enabled. Expert placement strategy: line
ar. Local/global number of experts: 20/160. Experts local to global index map: 0->120, 1->121, 2->122, 3->123, 4->124, 5->125, 6->126, 7
->127, 8->128, 9->129, 10->130, 11->131, 12->132, 13->133, 14->134, 15->135, 16->136, 17->137, 18->138, 19->139.
(Worker_TP6_EP6 pid=409) INFO 11-25 08:01:25 [fp8.py:167] Using DeepGEMM backend for FP8 MoE
(Worker_TP4_EP4 pid=407) INFO 11-25 08:01:25 [cuda.py:418] Valid backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION']
(Worker_TP4_EP4 pid=407) INFO 11-25 08:01:25 [cuda.py:427] Using FLASH_ATTN backend.
(Worker_TP4_EP4 pid=407) INFO 11-25 08:01:25 [layer.py:342] Enabled separate cuda stream for MoE shared_experts
(Worker_TP4_EP4 pid=407) INFO 11-25 08:01:25 [layer.py:468] [EP Rank 4/8] Expert parallelism is enabled. Expert placement strategy: line
ar. Local/global number of experts: 20/160. Experts local to global index map: 0->80, 1->81, 2->82, 3->83, 4->84, 5->85, 6->86, 7->87, 8
->88, 9->89, 10->90, 11->91, 12->92, 13->93, 14->94, 15->95, 16->96, 17->97, 18->98, 19->99.
(Worker_TP4_EP4 pid=407) INFO 11-25 08:01:25 [fp8.py:167] Using DeepGEMM backend for FP8 MoE
(Worker_TP5_EP5 pid=408) INFO 11-25 08:01:25 [cuda.py:418] Valid backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION']
(Worker_TP5_EP5 pid=408) INFO 11-25 08:01:25 [cuda.py:427] Using FLASH_ATTN backend.
(Worker_TP5_EP5 pid=408) INFO 11-25 08:01:25 [layer.py:342] Enabled separate cuda stream for MoE shared_experts
(Worker_TP2_EP2 pid=405) INFO 11-25 08:01:25 [cuda.py:418] Valid backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION']
(Worker_TP2_EP2 pid=405) INFO 11-25 08:01:25 [cuda.py:427] Using FLASH_ATTN backend.
(Worker_TP2_EP2 pid=405) INFO 11-25 08:01:25 [layer.py:342] Enabled separate cuda stream for MoE shared_experts
(Worker_TP3_EP3 pid=406) INFO 11-25 08:01:25 [cuda.py:418] Valid backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION']
(Worker_TP3_EP3 pid=406) INFO 11-25 08:01:25 [cuda.py:427] Using FLASH_ATTN backend.
(Worker_TP3_EP3 pid=406) INFO 11-25 08:01:25 [layer.py:342] Enabled separate cuda stream for MoE shared_experts
(Worker_TP0_EP0 pid=403) INFO 11-25 08:01:25 [cuda.py:418] Valid backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION']
(Worker_TP0_EP0 pid=403) INFO 11-25 08:01:25 [cuda.py:427] Using FLASH_ATTN backend.
(Worker_TP0_EP0 pid=403) INFO 11-25 08:01:25 [layer.py:342] Enabled separate cuda stream for MoE shared_experts
(Worker_TP7_EP7 pid=410) INFO 11-25 08:01:25 [cuda.py:418] Valid backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION']
(Worker_TP7_EP7 pid=410) INFO 11-25 08:01:25 [cuda.py:427] Using FLASH_ATTN backend.
(Worker_TP7_EP7 pid=410) INFO 11-25 08:01:25 [layer.py:342] Enabled separate cuda stream for MoE shared_experts
(Worker_TP1_EP1 pid=404) INFO 11-25 08:01:25 [cuda.py:418] Valid backends: ['FLASH_ATTN', 'FLASHINFER', 'TRITON_ATTN', 'FLEX_ATTENTION']
(Worker_TP1_EP1 pid=404) INFO 11-25 08:01:25 [cuda.py:427] Using FLASH_ATTN backend.
(Worker_TP1_EP1 pid=404) INFO 11-25 08:01:25 [layer.py:342] Enabled separate cuda stream for MoE shared_experts
(Worker_TP5_EP5 pid=408) INFO 11-25 08:01:25 [layer.py:468] [EP Rank 5/8] Expert parallelism is enabled. Expert placement strategy: line
ar. Local/global number of experts: 20/160. Experts local to global index map: 0->100, 1->101, 2->102, 3->103, 4->104, 5->105, 6->106, 7
->107, 8->108, 9->109, 10->110, 11->111, 12->112, 13->113, 14->114, 15->115, 16->116, 17->117, 18->118, 19->119.
(Worker_TP5_EP5 pid=408) INFO 11-25 08:01:25 [fp8.py:167] Using DeepGEMM backend for FP8 MoE
(Worker_TP2_EP2 pid=405) INFO 11-25 08:01:25 [layer.py:468] [EP Rank 2/8] Expert parallelism is enabled. Expert placement strategy: line
ar. Local/global number of experts: 20/160. Experts local to global index map: 0->40, 1->41, 2->42, 3->43, 4->44, 5->45, 6->46, 7->47, 8
->48, 9->49, 10->50, 11->51, 12->52, 13->53, 14->54, 15->55, 16->56, 17->57, 18->58, 19->59.
(Worker_TP2_EP2 pid=405) INFO 11-25 08:01:25 [fp8.py:167] Using DeepGEMM backend for FP8 MoE
(Worker_TP3_EP3 pid=406) INFO 11-25 08:01:25 [layer.py:468] [EP Rank 3/8] Expert parallelism is enabled. Expert placement strategy: line
ar. Local/global number of experts: 20/160. Experts local to global index map: 0->60, 1->61, 2->62, 3->63, 4->64, 5->65, 6->66, 7->67, 8
->68, 9->69, 10->70, 11->71, 12->72, 13->73, 14->74, 15->75, 16->76, 17->77, 18->78, 19->79.
(Worker_TP3_EP3 pid=406) INFO 11-25 08:01:25 [fp8.py:167] Using DeepGEMM backend for FP8 MoE
(Worker_TP0_EP0 pid=403) INFO 11-25 08:01:25 [layer.py:468] [EP Rank 0/8] Expert parallelism is enabled. Expert placement strategy: line
ar. Local/global number of experts: 20/160. Experts local to global index map: 0->0, 1->1, 2->2, 3->3, 4->4, 5->5, 6->6, 7->7, 8->8, 9->
9, 10->10, 11->11, 12->12, 13->13, 14->14, 15->15, 16->16, 17->17, 18->18, 19->19.
(Worker_TP0_EP0 pid=403) INFO 11-25 08:01:25 [fp8.py:167] Using DeepGEMM backend for FP8 MoE
(Worker_TP7_EP7 pid=410) INFO 11-25 08:01:25 [layer.py:468] [EP Rank 7/8] Expert parallelism is enabled. Expert placement strategy: line
ar. Local/global number of experts: 20/160. Experts local to global index map: 0->140, 1->141, 2->142, 3->143, 4->144, 5->145, 6->146, 7
->147, 8->148, 9->149, 10->150, 11->151, 12->152, 13->153, 14->154, 15->155, 16->156, 17->157, 18->158, 19->159.
(Worker_TP7_EP7 pid=410) INFO 11-25 08:01:25 [fp8.py:167] Using DeepGEMM backend for FP8 MoE
(Worker_TP1_EP1 pid=404) INFO 11-25 08:01:25 [layer.py:468] [EP Rank 1/8] Expert parallelism is enabled. Expert placement strategy: line
ar. Local/global number of experts: 20/160. Experts local to global index map: 0->20, 1->21, 2->22, 3->23, 4->24, 5->25, 6->26, 7->27, 8
->28, 9->29, 10->30, 11->31, 12->32, 13->33, 14->34, 15->35, 16->36, 17->37, 18->38, 19->39.
(Worker_TP1_EP1 pid=404) INFO 11-25 08:01:25 [fp8.py:167] Using DeepGEMM backend for FP8 MoE
Loading safetensors checkpoint shards: 0% Completed | 0/49 [00:00<?, ?it/s]
Loading safetensors checkpoint shards: 2% Completed | 1/49 [00:01<01:21, 1.71s/it]
Loading safetensors checkpoint shards: 4% Completed | 2/49 [00:02<00:59, 1.27s/it]
Loading safetensors checkpoint shards: 6% Completed | 3/49 [00:03<00:51, 1.13s/it]
Loading safetensors checkpoint shards: 8% Completed | 4/49 [00:05<00:55, 1.22s/it]
Loading safetensors checkpoint shards: 10% Completed | 5/49 [00:06<00:55, 1.26s/it]
Loading safetensors checkpoint shards: 12% Completed | 6/49 [00:07<00:48, 1.12s/it]
Loading safetensors checkpoint shards: 14% Completed | 7/49 [00:08<00:45, 1.08s/it]
Loading safetensors checkpoint shards: 16% Completed | 8/49 [00:09<00:49, 1.21s/it]
Loading safetensors checkpoint shards: 18% Completed | 9/49 [00:10<00:48, 1.21s/it]
Loading safetensors checkpoint shards: 20% Completed | 10/49 [00:12<00:52, 1.34s/it]
Loading safetensors checkpoint shards: 22% Completed | 11/49 [00:13<00:46, 1.22s/it]
Loading safetensors checkpoint shards: 24% Completed | 12/49 [00:14<00:43, 1.17s/it]
Loading safetensors checkpoint shards: 27% Completed | 13/49 [00:15<00:42, 1.18s/it]
Loading safetensors checkpoint shards: 29% Completed | 14/49 [00:17<00:42, 1.22s/it]
Loading safetensors checkpoint shards: 31% Completed | 15/49 [00:18<00:40, 1.20s/it]
Loading safetensors checkpoint shards: 33% Completed | 16/49 [00:19<00:38, 1.17s/it]
Loading safetensors checkpoint shards: 35% Completed | 17/49 [00:20<00:39, 1.25s/it]
Loading safetensors checkpoint shards: 37% Completed | 18/49 [00:22<00:40, 1.29s/it]
Loading safetensors checkpoint shards: 39% Completed | 19/49 [00:23<00:42, 1.41s/it]
Loading safetensors checkpoint shards: 41% Completed | 20/49 [00:24<00:36, 1.24s/it]
Loading safetensors checkpoint shards: 43% Completed | 21/49 [00:25<00:29, 1.07s/it]
Loading safetensors checkpoint shards: 45% Completed | 22/49 [00:25<00:23, 1.15it/s]
Loading safetensors checkpoint shards: 47% Completed | 23/49 [00:26<00:22, 1.15it/s]
Loading safetensors checkpoint shards: 49% Completed | 24/49 [00:27<00:22, 1.12it/s]
Loading safetensors checkpoint shards: 51% Completed | 25/49 [00:28<00:22, 1.08it/s]
Loading safetensors checkpoint shards: 53% Completed | 26/49 [00:29<00:23, 1.03s/it]
Loading safetensors checkpoint shards: 55% Completed | 27/49 [00:31<00:28, 1.28s/it]
Loading safetensors checkpoint shards: 57% Completed | 28/49 [00:32<00:24, 1.18s/it]
Loading safetensors checkpoint shards: 59% Completed | 29/49 [00:34<00:26, 1.31s/it]
Loading safetensors checkpoint shards: 61% Completed | 30/49 [00:35<00:22, 1.18s/it]
Loading safetensors checkpoint shards: 63% Completed | 31/49 [00:36<00:21, 1.19s/it]
Loading safetensors checkpoint shards: 65% Completed | 32/49 [00:37<00:20, 1.18s/it]
Loading safetensors checkpoint shards: 67% Completed | 33/49 [00:39<00:21, 1.33s/it]
Loading safetensors checkpoint shards: 69% Completed | 34/49 [00:40<00:20, 1.39s/it]
Loading safetensors checkpoint shards: 71% Completed | 35/49 [00:41<00:16, 1.20s/it]
Loading safetensors checkpoint shards: 73% Completed | 36/49 [00:42<00:13, 1.02s/it]
Loading safetensors checkpoint shards: 76% Completed | 37/49 [00:42<00:09, 1.23it/s]
Loading safetensors checkpoint shards: 78% Completed | 38/49 [00:42<00:07, 1.48it/s]
Loading safetensors checkpoint shards: 80% Completed | 39/49 [00:43<00:05, 1.79it/s]
Loading safetensors checkpoint shards: 82% Completed | 40/49 [00:43<00:04, 2.11it/s]
Loading safetensors checkpoint shards: 84% Completed | 41/49 [00:43<00:03, 2.39it/s]
Loading safetensors checkpoint shards: 88% Completed | 43/49 [00:44<00:02, 2.95it/s]
Loading safetensors checkpoint shards: 90% Completed | 44/49 [00:44<00:01, 2.83it/s]
Loading safetensors checkpoint shards: 92% Completed | 45/49 [00:44<00:01, 2.89it/s]
Loading safetensors checkpoint shards: 94% Completed | 46/49 [00:45<00:00, 3.03it/s]
Loading safetensors checkpoint shards: 96% Completed | 47/49 [00:45<00:00, 3.10it/s]
Loading safetensors checkpoint shards: 98% Completed | 48/49 [00:45<00:00, 2.78it/s]
Loading safetensors checkpoint shards: 100% Completed | 49/49 [00:46<00:00, 2.91it/s]
Loading safetensors checkpoint shards: 100% Completed | 49/49 [00:46<00:00, 1.06it/s]
(Worker_TP0_EP0 pid=403)
(Worker_TP0_EP0 pid=403) INFO 11-25 08:02:13 [default_loader.py:314] Loading weights took 46.16 seconds
(Worker_TP0_EP0 pid=403) INFO 11-25 08:02:37 [gpu_model_runner.py:3334] Model loading took 56.3302 GiB memory and 88.442082 seconds
(Worker_TP0_EP0 pid=403) INFO 11-25 08:03:06 [gpu_worker.py:359] Available KV cache memory: 34.03 GiB
(EngineCore_DP0 pid=270) INFO 11-25 08:03:07 [kv_cache_utils.py:1229] GPU KV cache size: 1,151,216 tokens
(EngineCore_DP0 pid=270) INFO 11-25 08:03:07 [kv_cache_utils.py:1234] Maximum concurrency for 131,072 tokens per request: 8.78x
(EngineCore_DP0 pid=270) INFO 11-25 08:03:07 [kv_cache_utils.py:1234] Maximum concurrency for 131,072 tokens per request: 8.78x
(EngineCore_DP0 pid=270) INFO 11-25 08:03:07 [kv_cache_utils.py:1234] Maximum concurrency for 131,072 tokens per request: 8.78x
(EngineCore_DP0 pid=270) INFO 11-25 08:03:07 [kv_cache_utils.py:1234] Maximum concurrency for 131,072 tokens per request: 8.78x
(EngineCore_DP0 pid=270) INFO 11-25 08:03:07 [kv_cache_utils.py:1234] Maximum concurrency for 131,072 tokens per request: 8.78x
(EngineCore_DP0 pid=270) INFO 11-25 08:03:07 [kv_cache_utils.py:1234] Maximum concurrency for 131,072 tokens per request: 8.78x
(EngineCore_DP0 pid=270) INFO 11-25 08:03:07 [kv_cache_utils.py:1234] Maximum concurrency for 131,072 tokens per request: 8.78x
(EngineCore_DP0 pid=270) INFO 11-25 08:03:07 [kv_cache_utils.py:1234] Maximum concurrency for 131,072 tokens per request: 8.78x
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([1792, 6144])) [relaxed]: 39%|███▊ | 98/253 [0(EngineCore_DP0 pid=270) INFO 11-25 08:04
:08 [shm_broadcast.py:501] No available shared memory broadcast block found in 60 seconds. This typically happens when some processes ar
e hanging or doing some time-consuming work (e.g. compilation, weight/kv cache quantization).
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([1792, 6144])) [relaxed]: 100%|██████████| 253/253 [01:11<00:00, 3.53it/s]
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([1792, 6144])) [relaxed]: 100%|██████████| 253/253 [01:11<00:00, 3.53it/s]
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([1792, 6144])) [relaxed]: 100%|██████████| 253/253 [01:11<00:00, 3.52it/s]
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([1792, 6144])) [relaxed]: 100%|██████████| 253/253 [01:11<00:00, 3.52it/s]
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([1792, 6144])) [relaxed]: 100%|██████████| 253/253 [01:11<00:00, 3.52it/s]
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([1792, 6144])) [relaxed]: 100%|██████████| 253/253 [01:11<00:00, 3.52it/s]
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([1792, 6144])) [relaxed]: 100%|██████████| 253/253 [01:11<00:00, 3.52it/s]
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([1792, 6144])) [relaxed]: 100%|██████████| 253/253 [01:11<00:00, 3.52it/s]
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([6144, 1536])) [relaxed]: 100%|██████████| 286/286 [00:46<00:00, 6.19it/s]
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] WorkerProc hit an exception.
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] Traceback (most recent call last):
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/execut
or/multiproc_executor.py", line 810, in worker_busy_loop
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] output = func(*args, **kwargs)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker
/gpu_worker.py", line 425, in compile_or_warm_up_model
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] kernel_warmup(self)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/kernel_warmup.py", line 37, in kernel_warmup
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deep_gemm_warmup(model, max_tokens)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 314, in deep_gemm_warmup
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deepgemm_grouped_fp8_gemm_nt_contiguous_warmup(model, max_
tokens)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 307, in deepgemm_grouped_fp8_gemm_nt_contiguous_warmup
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] _deepgemm_grouped_fp8_gemm_nt_contiguous_warmup(
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 284, in _deepgemm_grouped_fp8_gemm_nt_contiguous_warmup
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] _warmup(w, ws)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 262, in warmup
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] out = torch.empty((MAX_M, n), device=device, dtype=torch.b
float16)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
^^^^^^^^
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] torch.OutOfMemoryError: CUDA out of memory. Tried to allocate
666.00 MiB. GPU 1 has a total capacity of 95.10 GiB of which 145.94 MiB is free. Process 53395 has 94.95 GiB memory in use. Of the alloc
ated memory 92.33 GiB is allocated by PyTorch, and 32.77 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory
is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management
(https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] Traceback (most recent call last):
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/execut
or/multiproc_executor.py", line 810, in worker_busy_loop
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] output = func(*args, **kwargs)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker
/gpu_worker.py", line 425, in compile_or_warm_up_model
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] kernel_warmup(self)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/kernel_warmup.py", line 37, in kernel_warmup
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deep_gemm_warmup(model, max_tokens)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 314, in deep_gemm_warmup
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deepgemm_grouped_fp8_gemm_nt_contiguous_warmup(model, max
tokens)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 307, in deepgemm_grouped_fp8_gemm_nt_contiguous_warmup
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] _deepgemm_grouped_fp8_gemm_nt_contiguous_warmup(
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 284, in _deepgemm_grouped_fp8_gemm_nt_contiguous_warmup
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] _warmup(w, ws)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 262, in warmup
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] out = torch.empty((MAX_M, n), device=device, dtype=torch.b
float16)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
^^^^^^^^
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815] torch.OutOfMemoryError: CUDA out of memory. Tried to allocate
666.00 MiB. GPU 1 has a total capacity of 95.10 GiB of which 145.94 MiB is free. Process 53395 has 94.95 GiB memory in use. Of the alloc
ated memory 92.33 GiB is allocated by PyTorch, and 32.77 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory
is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management
(https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
(Worker_TP1_EP1 pid=404) ERROR 11-25 08:05:06 [multiproc_executor.py:815]
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([6144, 1536])) [relaxed]: 100%|██████████| 286/286 [00:46<00:00, 6.19it/s]
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([6144, 1536])) [relaxed]: 100%|██████████| 286/286 [00:46<00:00, 6.19it/s]
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] WorkerProc hit an exception.
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] Traceback (most recent call last):
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/execut
or/multiproc_executor.py", line 810, in worker_busy_loop
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] output = func(*args, **kwargs)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker
/gpu_worker.py", line 425, in compile_or_warm_up_model
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] kernel_warmup(self)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/kernel_warmup.py", line 37, in kernel_warmup
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deep_gemm_warmup(model, max_tokens)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 314, in deep_gemm_warmup
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deepgemm_grouped_fp8_gemm_nt_contiguous_warmup(model, max
tokens)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 307, in deepgemm_grouped_fp8_gemm_nt_contiguous_warmup
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] _deepgemm_grouped_fp8_gemm_nt_contiguous_warmup(
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 284, in _deepgemm_grouped_fp8_gemm_nt_contiguous_warmup
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] _warmup(w, ws)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 262, in warmup
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] out = torch.empty((MAX_M, n), device=device, dtype=torch.b
float16)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
^^^^^^^^
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] torch.OutOfMemoryError: CUDA out of memory. Tried to allocate
666.00 MiB. GPU 3 has a total capacity of 95.10 GiB of which 145.94 MiB is free. Process 53397 has 94.95 GiB memory in use. Of the alloc
ated memory 92.33 GiB is allocated by PyTorch, and 32.77 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory
is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management
(https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] Traceback (most recent call last):
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/execut
or/multiproc_executor.py", line 810, in worker_busy_loop
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] output = func(*args, **kwargs)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker
/gpu_worker.py", line 425, in compile_or_warm_up_model
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] kernel_warmup(self)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/kernel_warmup.py", line 37, in kernel_warmup
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deep_gemm_warmup(model, max_tokens)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 314, in deep_gemm_warmup
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deepgemm_grouped_fp8_gemm_nt_contiguous_warmup(model, max
tokens)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 307, in deepgemm_grouped_fp8_gemm_nt_contiguous_warmup
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] _deepgemm_grouped_fp8_gemm_nt_contiguous_warmup(
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 284, in _deepgemm_grouped_fp8_gemm_nt_contiguous_warmup
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] _warmup(w, ws)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 262, in warmup
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] out = torch.empty((MAX_M, n), device=device, dtype=torch.b
float16)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
^^^^^^^^
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815] torch.OutOfMemoryError: CUDA out of memory. Tried to allocate
666.00 MiB. GPU 3 has a total capacity of 95.10 GiB of which 145.94 MiB is free. Process 53397 has 94.95 GiB memory in use. Of the alloc
ated memory 92.33 GiB is allocated by PyTorch, and 32.77 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory
is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management
(https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
(Worker_TP3_EP3 pid=406) ERROR 11-25 08:05:06 [multiproc_executor.py:815]
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] WorkerProc hit an exception.
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] Traceback (most recent call last):
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/execut
or/multiproc_executor.py", line 810, in worker_busy_loop
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] output = func(*args, **kwargs)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker
/gpu_worker.py", line 425, in compile_or_warm_up_model
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] kernel_warmup(self)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/kernel_warmup.py", line 37, in kernel_warmup
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deep_gemm_warmup(model, max_tokens)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 314, in deep_gemm_warmup
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deepgemm_grouped_fp8_gemm_nt_contiguous_warmup(model, max
tokens)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 307, in deepgemm_grouped_fp8_gemm_nt_contiguous_warmup
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] _deepgemm_grouped_fp8_gemm_nt_contiguous_warmup(
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 284, in _deepgemm_grouped_fp8_gemm_nt_contiguous_warmup
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] _warmup(w, ws)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 262, in warmup
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] out = torch.empty((MAX_M, n), device=device, dtype=torch.b
float16)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
^^^^^^^^
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] torch.OutOfMemoryError: CUDA out of memory. Tried to allocate
666.00 MiB. GPU 2 has a total capacity of 95.10 GiB of which 145.94 MiB is free. Process 53396 has 94.95 GiB memory in use. Of the alloc
ated memory 92.33 GiB is allocated by PyTorch, and 32.77 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory
is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management
(https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] Traceback (most recent call last):
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/execut
or/multiproc_executor.py", line 810, in worker_busy_loop
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] output = func(*args, **kwargs)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker
/gpu_worker.py", line 425, in compile_or_warm_up_model
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] kernel_warmup(self)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/kernel_warmup.py", line 37, in kernel_warmup
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deep_gemm_warmup(model, max_tokens)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 314, in deep_gemm_warmup
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deepgemm_grouped_fp8_gemm_nt_contiguous_warmup(model, max
tokens)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 307, in deepgemm_grouped_fp8_gemm_nt_contiguous_warmup
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] _deepgemm_grouped_fp8_gemm_nt_contiguous_warmup(
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 284, in _deepgemm_grouped_fp8_gemm_nt_contiguous_warmup
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] _warmup(w, ws)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 262, in _warmup
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] out = torch.empty((MAX_M, n), device=device, dtype=torch.b
float16)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
^^^^^^^^
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815] torch.OutOfMemoryError: CUDA out of memory. Tried to allocate
666.00 MiB. GPU 2 has a total capacity of 95.10 GiB of which 145.94 MiB is free. Process 53396 has 94.95 GiB memory in use. Of the alloc
ated memory 92.33 GiB is allocated by PyTorch, and 32.77 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory
is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management
(https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
(Worker_TP2_EP2 pid=405) ERROR 11-25 08:05:06 [multiproc_executor.py:815]
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([6144, 1536])) [relaxed]: 100%|██████████| 286/286 [00:46<00:00, 6.18it/s]
DeepGemm(fp8_gemm_nt) warmup (W=torch.Size([6144, 1536])) [relaxed]: 100%|██████████| 286/286 [00:46<00:00, 6.18it/s]
(Worker_TP5_EP5 pid=408) ERROR 11-25 08:05:06 [multiproc_executor.py:815] WorkerProc hit an exception.
(Worker_TP5_EP5 pid=408) ERROR 11-25 08:05:06 [multiproc_executor.py:815] Traceback (most recent call last):
(Worker_TP5_EP5 pid=408) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/execut
or/multiproc_executor.py", line 810, in worker_busy_loop
(Worker_TP5_EP5 pid=408) ERROR 11-25 08:05:06 [multiproc_executor.py:815] output = func(*args, **kwargs)
(Worker_TP5_EP5 pid=408) ERROR 11-25 08:05:06 [multiproc_executor.py:815] ^^^^^^^^^^^^^^^^^^^^^
(Worker_TP5_EP5 pid=408) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker
/gpu_worker.py", line 425, in compile_or_warm_up_model
(Worker_TP5_EP5 pid=408) ERROR 11-25 08:05:06 [multiproc_executor.py:815] kernel_warmup(self)
(Worker_TP5_EP5 pid=408) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/kernel_warmup.py", line 37, in kernel_warmup
(Worker_TP5_EP5 pid=408) ERROR 11-25 08:05:06 [multiproc_executor.py:815] deep_gemm_warmup(model, max_tokens)
(Worker_TP5_EP5 pid=408) ERROR 11-25 08:05:06 [multiproc_executor.py:815] File "/usr/local/lib/python3.12/dist-packages/vllm/model_exe
cutor/warmup/deep_gemm_warmup.py", line 314, in deep_gemm_warmup
Before submitting a new issue...
- Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.