[Bug]: Not support device type: cpu

[princepride]

Your current environment

The output of python collect_env.py

INFO 05-17 08:15:35 [__init__.py:248] Automatically detected platform cuda.
Collecting environment information...
==============================
        System Info
==============================
OS                           : Ubuntu 22.04.3 LTS (x86_64)
GCC version                  : (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version                : Could not collect
CMake version                : Could not collect
Libc version                 : glibc-2.35

==============================
       PyTorch Info
==============================
PyTorch version              : 2.7.0+cu126
Is debug build               : False
CUDA used to build PyTorch   : 12.6
ROCM used to build PyTorch   : N/A

==============================
      Python Environment
==============================
Python version               : 3.10.12 (main, Jun 11 2023, 05:26:28) [GCC 11.4.0] (64-bit runtime)
Python platform              : Linux-6.8.0-57-generic-x86_64-with-glibc2.35

==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : 11.8.89
CUDA_MODULE_LOADING set to   : LAZY
GPU models and configuration : GPU 0: NVIDIA A40
Nvidia driver version        : 565.57.01
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:                        52 bits physical, 57 bits virtual
Byte Order:                           Little Endian
CPU(s):                               96
On-line CPU(s) list:                  0-95
Vendor ID:                            GenuineIntel
Model name:                           Intel(R) Xeon(R) Gold 6342 CPU @ 2.80GHz
CPU family:                           6
Model:                                106
Thread(s) per core:                   2
Core(s) per socket:                   24
Socket(s):                            2
Stepping:                             6
CPU max MHz:                          3500.0000
CPU min MHz:                          800.0000
BogoMIPS:                             5600.00
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 aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect wbnoinvd dtherm ida arat pln pts vnmi avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid fsrm md_clear pconfig flush_l1d arch_capabilities
Virtualization:                       VT-x
L1d cache:                            2.3 MiB (48 instances)
L1i cache:                            1.5 MiB (48 instances)
L2 cache:                             60 MiB (48 instances)
L3 cache:                             72 MiB (2 instances)
NUMA node(s):                         2
NUMA node0 CPU(s):                    0-23,48-71
NUMA node1 CPU(s):                    24-47,72-95
Vulnerability Gather data sampling:   Mitigation; Microcode
Vulnerability Itlb multihit:          Not affected
Vulnerability L1tf:                   Not affected
Vulnerability Mds:                    Not affected
Vulnerability Meltdown:               Not affected
Vulnerability Mmio stale data:        Mitigation; Clear CPU buffers; SMT vulnerable
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 SW loop, KVM SW loop
Vulnerability Srbds:                  Not affected
Vulnerability Tsx async abort:        Not affected

==============================
Versions of relevant libraries
==============================
[pip3] numpy==2.2.5
[pip3] nvidia-cublas-cu12==12.6.4.1
[pip3] nvidia-cuda-cupti-cu12==12.6.80
[pip3] nvidia-cuda-nvrtc-cu12==12.6.77
[pip3] nvidia-cuda-runtime-cu12==12.6.77
[pip3] nvidia-cudnn-cu12==9.5.1.17
[pip3] nvidia-cufft-cu12==11.3.0.4
[pip3] nvidia-cufile-cu12==1.11.1.6
[pip3] nvidia-curand-cu12==10.3.7.77
[pip3] nvidia-cusolver-cu12==11.7.1.2
[pip3] nvidia-cusparse-cu12==12.5.4.2
[pip3] nvidia-cusparselt-cu12==0.6.3
[pip3] nvidia-nccl-cu12==2.26.2
[pip3] nvidia-nvjitlink-cu12==12.6.85
[pip3] nvidia-nvtx-cu12==12.6.77
[pip3] pyzmq==26.4.0
[pip3] torch==2.7.0
[pip3] torchaudio==2.7.0
[pip3] torchvision==0.22.0
[pip3] transformers==4.51.3
[pip3] triton==3.3.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : Could not collect
Neuron SDK Version           : N/A
vLLM Version                 : 0.9.1.dev49+g3e0d43502 (git sha: 3e0d43502)
vLLM Build Flags:
  CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
        GPU0    NIC0    NIC1    CPU Affinity    NUMA Affinity   GPU NUMA ID
GPU0     X      SYS     SYS     0-23,48-71      0               N/A
NIC0    SYS      X      PIX
NIC1    SYS     PIX      X 

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

NIC Legend:

  NIC0: mlx5_0
  NIC1: mlx5_1

==============================
     Environment Variables
==============================
NVIDIA_VISIBLE_DEVICES=GPU-2f99657d-f945-6673-180d-d080f05e8f28
NVIDIA_REQUIRE_CUDA=cuda>=11.8 brand=tesla,driver>=450,driver<451 brand=tesla,driver>=470,driver<471 brand=unknown,driver>=470,driver<471 brand=nvidia,driver>=470,driver<471 brand=nvidiartx,driver>=470,driver<471 brand=geforce,driver>=470,driver<471 brand=geforcertx,driver>=470,driver<471 brand=quadro,driver>=470,driver<471 brand=quadrortx,driver>=470,driver<471 brand=titan,driver>=470,driver<471 brand=titanrtx,driver>=470,driver<471 brand=tesla,driver>=510,driver<511 brand=unknown,driver>=510,driver<511 brand=nvidia,driver>=510,driver<511 brand=nvidiartx,driver>=510,driver<511 brand=geforce,driver>=510,driver<511 brand=geforcertx,driver>=510,driver<511 brand=quadro,driver>=510,driver<511 brand=quadrortx,driver>=510,driver<511 brand=titan,driver>=510,driver<511 brand=titanrtx,driver>=510,driver<511 brand=tesla,driver>=515,driver<516 brand=unknown,driver>=515,driver<516 brand=nvidia,driver>=515,driver<516 brand=nvidiartx,driver>=515,driver<516 brand=geforce,driver>=515,driver<516 brand=geforcertx,driver>=515,driver<516 brand=quadro,driver>=515,driver<516 brand=quadrortx,driver>=515,driver<516 brand=titan,driver>=515,driver<516 brand=titanrtx,driver>=515,driver<516
NCCL_VERSION=2.15.5-1
NVIDIA_DRIVER_CAPABILITIES=compute,display,graphics,utility,video
NVIDIA_PRODUCT_NAME=CUDA
CUDA_VERSION=11.8.0
LD_LIBRARY_PATH=/usr/local/nvidia/lib:/usr/local/nvidia/lib64
NCCL_CUMEM_ENABLE=0
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
CUDA_MODULE_LOADING=LAZY

🐛 Describe the bug

As described in the documentation: https://docs.vllm.ai/en/latest/serving/serve_args.html#vllm.entrypoints.openai.cli_args-create_parser_for_docs-deviceconfig [--device {auto,cpu,cuda,hpu,neuron,tpu,xpu}], we can choose cpu as the device. However, after I ran

vllm serve facebook/opt-125m --device cpu

, the following error occurred:

Error Message:

ERROR 05-17 08:13:54 [core.py:489] EngineCore failed to start.
ERROR 05-17 08:13:54 [core.py:489] Traceback (most recent call last):
ERROR 05-17 08:13:54 [core.py:489]   File "/workspace/vllm/vllm/v1/engine/core.py", line 480, in run_engine_core
ERROR 05-17 08:13:54 [core.py:489]     engine_core = EngineCoreProc(*args, **kwargs)
ERROR 05-17 08:13:54 [core.py:489]   File "/workspace/vllm/vllm/v1/engine/core.py", line 379, in __init__
ERROR 05-17 08:13:54 [core.py:489]     super().__init__(vllm_config, executor_class, log_stats,
ERROR 05-17 08:13:54 [core.py:489]   File "/workspace/vllm/vllm/v1/engine/core.py", line 67, in __init__
ERROR 05-17 08:13:54 [core.py:489]     self.model_executor = executor_class(vllm_config)
ERROR 05-17 08:13:54 [core.py:489]   File "/workspace/vllm/vllm/executor/executor_base.py", line 52, in __init__
ERROR 05-17 08:13:54 [core.py:489]     self._init_executor()
ERROR 05-17 08:13:54 [core.py:489]   File "/workspace/vllm/vllm/executor/uniproc_executor.py", line 46, in _init_executor
ERROR 05-17 08:13:54 [core.py:489]     self.collective_rpc("init_device")
ERROR 05-17 08:13:54 [core.py:489]   File "/workspace/vllm/vllm/executor/uniproc_executor.py", line 56, in collective_rpc
ERROR 05-17 08:13:54 [core.py:489]     answer = run_method(self.driver_worker, method, args, kwargs)
ERROR 05-17 08:13:54 [core.py:489]   File "/workspace/vllm/vllm/utils.py", line 2598, in run_method
ERROR 05-17 08:13:54 [core.py:489]     return func(*args, **kwargs)
ERROR 05-17 08:13:54 [core.py:489]   File "/workspace/vllm/vllm/worker/worker_base.py", line 604, in init_device
ERROR 05-17 08:13:54 [core.py:489]     self.worker.init_device()  # type: ignore
ERROR 05-17 08:13:54 [core.py:489]   File "/workspace/vllm/vllm/v1/worker/gpu_worker.py", line 133, in init_device
ERROR 05-17 08:13:54 [core.py:489]     raise RuntimeError(
ERROR 05-17 08:13:54 [core.py:489] RuntimeError: Not support device type: cpu
Process EngineCore_0:
Traceback (most recent call last):
  File "/usr/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap
    self.run()
  File "/usr/lib/python3.10/multiprocessing/process.py", line 108, in run
    self._target(*self._args, **self._kwargs)
  File "/workspace/vllm/vllm/v1/engine/core.py", line 493, in run_engine_core
    raise e
  File "/workspace/vllm/vllm/v1/engine/core.py", line 480, in run_engine_core
    engine_core = EngineCoreProc(*args, **kwargs)
  File "/workspace/vllm/vllm/v1/engine/core.py", line 379, in __init__
    super().__init__(vllm_config, executor_class, log_stats,
  File "/workspace/vllm/vllm/v1/engine/core.py", line 67, in __init__
    self.model_executor = executor_class(vllm_config)
  File "/workspace/vllm/vllm/executor/executor_base.py", line 52, in __init__
    self._init_executor()
  File "/workspace/vllm/vllm/executor/uniproc_executor.py", line 46, in _init_executor
    self.collective_rpc("init_device")
  File "/workspace/vllm/vllm/executor/uniproc_executor.py", line 56, in collective_rpc
    answer = run_method(self.driver_worker, method, args, kwargs)
  File "/workspace/vllm/vllm/utils.py", line 2598, in run_method
    return func(*args, **kwargs)
  File "/workspace/vllm/vllm/worker/worker_base.py", line 604, in init_device
    self.worker.init_device()  # type: ignore
  File "/workspace/vllm/vllm/v1/worker/gpu_worker.py", line 133, in init_device
    raise RuntimeError(
RuntimeError: Not support device type: cpu
Traceback (most recent call last):
  File "/usr/local/bin/vllm", line 8, in <module>
    sys.exit(main())
  File "/workspace/vllm/vllm/entrypoints/cli/main.py", line 53, in main
    args.dispatch_function(args)
  File "/workspace/vllm/vllm/entrypoints/cli/serve.py", line 40, in cmd
    uvloop.run(run_server(args))
  File "/usr/local/lib/python3.10/dist-packages/uvloop/__init__.py", line 82, in run
    return loop.run_until_complete(wrapper())
  File "uvloop/loop.pyx", line 1518, in uvloop.loop.Loop.run_until_complete
  File "/usr/local/lib/python3.10/dist-packages/uvloop/__init__.py", line 61, in wrapper
    return await main
  File "/workspace/vllm/vllm/entrypoints/openai/api_server.py", line 1324, in run_server
    async with build_async_engine_client(args) as engine_client:
  File "/usr/lib/python3.10/contextlib.py", line 199, in __aenter__
    return await anext(self.gen)
  File "/workspace/vllm/vllm/entrypoints/openai/api_server.py", line 153, in build_async_engine_client
    async with build_async_engine_client_from_engine_args(
  File "/usr/lib/python3.10/contextlib.py", line 199, in __aenter__
    return await anext(self.gen)
  File "/workspace/vllm/vllm/entrypoints/openai/api_server.py", line 185, in build_async_engine_client_from_engine_args
    async_llm = AsyncLLM.from_vllm_config(
  File "/workspace/vllm/vllm/v1/engine/async_llm.py", line 152, in from_vllm_config
    return cls(
  File "/workspace/vllm/vllm/v1/engine/async_llm.py", line 118, in __init__
    self.engine_core = core_client_class(
  File "/workspace/vllm/vllm/v1/engine/core_client.py", line 734, in __init__
    super().__init__(
  File "/workspace/vllm/vllm/v1/engine/core_client.py", line 418, in __init__
    self._wait_for_engine_startup(output_address, parallel_config)
  File "/workspace/vllm/vllm/v1/engine/core_client.py", line 484, in _wait_for_engine_startup
    raise RuntimeError("Engine core initialization failed. "
RuntimeError: Engine core initialization failed. See root cause above. Failed core proc

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.

[princepride]

@DarkLight1337, I would like to know if this is a bug. If it is, can it be assigned to me to fix? Thank you.

[DarkLight1337]

Did you set VLLM_USE_V1 in a previous command? vLLM doesn't support V1 on CPU yet.

[princepride]

root@20ffe1404c1f:/workspace/vllm# VLLM_USE_V1=0 vllm serve facebook/opt-125m --device cpu
INFO 05-17 11:00:15 [__init__.py:248] Automatically detected platform cuda.
INFO 05-17 11:00:18 [__init__.py:30] Available plugins for group vllm.general_plugins:
INFO 05-17 11:00:18 [__init__.py:32] name=lora_filesystem_resolver, value=vllm.plugins.lora_resolvers.filesystem_resolver:register_filesystem_resolver
INFO 05-17 11:00:18 [__init__.py:34] all available plugins for group vllm.general_plugins will be loaded.
INFO 05-17 11:00:18 [__init__.py:36] set environment variable VLLM_PLUGINS to control which plugins to load.
INFO 05-17 11:00:18 [__init__.py:44] plugin lora_filesystem_resolver loaded.
INFO 05-17 11:00:20 [api_server.py:1289] vLLM API server version 0.9.1.dev53+g9214e6063
INFO 05-17 11:00:20 [cli_args.py:297] non-default args: {'device': 'cpu'}
INFO 05-17 11:00:27 [config.py:787] This model supports multiple tasks: {'score', 'embed', 'classify', 'reward', 'generate'}. Defaulting to 'generate'.
INFO 05-17 11:00:27 [api_server.py:257] Started engine process with PID 1447
INFO 05-17 11:00:30 [__init__.py:248] Automatically detected platform cuda.
INFO 05-17 11:00:33 [__init__.py:30] Available plugins for group vllm.general_plugins:
INFO 05-17 11:00:33 [__init__.py:32] name=lora_filesystem_resolver, value=vllm.plugins.lora_resolvers.filesystem_resolver:register_filesystem_resolver
INFO 05-17 11:00:33 [__init__.py:34] all available plugins for group vllm.general_plugins will be loaded.
INFO 05-17 11:00:33 [__init__.py:36] set environment variable VLLM_PLUGINS to control which plugins to load.
INFO 05-17 11:00:33 [__init__.py:44] plugin lora_filesystem_resolver loaded.
INFO 05-17 11:00:33 [llm_engine.py:240] Initializing a V0 LLM engine (v0.9.1.dev53+g9214e6063) with config: model='facebook/opt-125m', speculative_config=None, tokenizer='facebook/opt-125m', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, override_neuron_config={}, tokenizer_revision=None, trust_remote_code=False, dtype=torch.float16, max_seq_len=2048, download_dir=None, load_format=auto, tensor_parallel_size=1, pipeline_parallel_size=1, disable_custom_all_reduce=False, quantization=None, enforce_eager=False, kv_cache_dtype=auto,  device_config=cpu, decoding_config=DecodingConfig(backend='xgrammar', disable_fallback=False, disable_any_whitespace=False, disable_additional_properties=False, reasoning_backend=''), observability_config=ObservabilityConfig(show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None), seed=None, served_model_name=facebook/opt-125m, num_scheduler_steps=1, multi_step_stream_outputs=True, enable_prefix_caching=None, chunked_prefill_enabled=False, use_async_output_proc=True, pooler_config=None, compilation_config={"compile_sizes": [], "inductor_compile_config": {"enable_auto_functionalized_v2": false}, "cudagraph_capture_sizes": [256, 248, 240, 232, 224, 216, 208, 200, 192, 184, 176, 168, 160, 152, 144, 136, 128, 120, 112, 104, 96, 88, 80, 72, 64, 56, 48, 40, 32, 24, 16, 8, 4, 2, 1], "max_capture_size": 256}, use_cached_outputs=True, 
INFO 05-17 11:00:33 [cuda.py:291] Using Flash Attention backend.
ERROR 05-17 11:00:33 [engine.py:454] Not support device type: cpu
ERROR 05-17 11:00:33 [engine.py:454] Traceback (most recent call last):
ERROR 05-17 11:00:33 [engine.py:454]   File "/workspace/vllm/vllm/engine/multiprocessing/engine.py", line 442, in run_mp_engine
ERROR 05-17 11:00:33 [engine.py:454]     engine = MQLLMEngine.from_vllm_config(
ERROR 05-17 11:00:33 [engine.py:454]   File "/workspace/vllm/vllm/engine/multiprocessing/engine.py", line 129, in from_vllm_config
ERROR 05-17 11:00:33 [engine.py:454]     return cls(
ERROR 05-17 11:00:33 [engine.py:454]   File "/workspace/vllm/vllm/engine/multiprocessing/engine.py", line 83, in __init__
ERROR 05-17 11:00:33 [engine.py:454]     self.engine = LLMEngine(*args, **kwargs)
ERROR 05-17 11:00:33 [engine.py:454]   File "/workspace/vllm/vllm/engine/llm_engine.py", line 275, in __init__
ERROR 05-17 11:00:33 [engine.py:454]     self.model_executor = executor_class(vllm_config=vllm_config)
ERROR 05-17 11:00:33 [engine.py:454]   File "/workspace/vllm/vllm/executor/executor_base.py", line 52, in __init__
ERROR 05-17 11:00:33 [engine.py:454]     self._init_executor()
ERROR 05-17 11:00:33 [engine.py:454]   File "/workspace/vllm/vllm/executor/uniproc_executor.py", line 46, in _init_executor
ERROR 05-17 11:00:33 [engine.py:454]     self.collective_rpc("init_device")
ERROR 05-17 11:00:33 [engine.py:454]   File "/workspace/vllm/vllm/executor/uniproc_executor.py", line 56, in collective_rpc
ERROR 05-17 11:00:33 [engine.py:454]     answer = run_method(self.driver_worker, method, args, kwargs)
ERROR 05-17 11:00:33 [engine.py:454]   File "/workspace/vllm/vllm/utils.py", line 2598, in run_method
ERROR 05-17 11:00:33 [engine.py:454]     return func(*args, **kwargs)
ERROR 05-17 11:00:33 [engine.py:454]   File "/workspace/vllm/vllm/worker/worker_base.py", line 604, in init_device
ERROR 05-17 11:00:33 [engine.py:454]     self.worker.init_device()  # type: ignore
ERROR 05-17 11:00:33 [engine.py:454]   File "/workspace/vllm/vllm/worker/worker.py", line 187, in init_device
ERROR 05-17 11:00:33 [engine.py:454]     raise RuntimeError(
ERROR 05-17 11:00:33 [engine.py:454] RuntimeError: Not support device type: cpu
Process SpawnProcess-1:
Traceback (most recent call last):
  File "/usr/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap
    self.run()
  File "/usr/lib/python3.10/multiprocessing/process.py", line 108, in run
    self._target(*self._args, **self._kwargs)
  File "/workspace/vllm/vllm/engine/multiprocessing/engine.py", line 456, in run_mp_engine
    raise e from None
  File "/workspace/vllm/vllm/engine/multiprocessing/engine.py", line 442, in run_mp_engine
    engine = MQLLMEngine.from_vllm_config(
  File "/workspace/vllm/vllm/engine/multiprocessing/engine.py", line 129, in from_vllm_config
    return cls(
  File "/workspace/vllm/vllm/engine/multiprocessing/engine.py", line 83, in __init__
    self.engine = LLMEngine(*args, **kwargs)
  File "/workspace/vllm/vllm/engine/llm_engine.py", line 275, in __init__
    self.model_executor = executor_class(vllm_config=vllm_config)
  File "/workspace/vllm/vllm/executor/executor_base.py", line 52, in __init__
    self._init_executor()
  File "/workspace/vllm/vllm/executor/uniproc_executor.py", line 46, in _init_executor
    self.collective_rpc("init_device")
  File "/workspace/vllm/vllm/executor/uniproc_executor.py", line 56, in collective_rpc
    answer = run_method(self.driver_worker, method, args, kwargs)
  File "/workspace/vllm/vllm/utils.py", line 2598, in run_method
    return func(*args, **kwargs)
  File "/workspace/vllm/vllm/worker/worker_base.py", line 604, in init_device
    self.worker.init_device()  # type: ignore
  File "/workspace/vllm/vllm/worker/worker.py", line 187, in init_device
    raise RuntimeError(
RuntimeError: Not support device type: cpu
Traceback (most recent call last):
  File "/usr/local/bin/vllm", line 8, in <module>
    sys.exit(main())
  File "/workspace/vllm/vllm/entrypoints/cli/main.py", line 53, in main
    args.dispatch_function(args)
  File "/workspace/vllm/vllm/entrypoints/cli/serve.py", line 40, in cmd
    uvloop.run(run_server(args))
  File "/usr/local/lib/python3.10/dist-packages/uvloop/__init__.py", line 82, in run
    return loop.run_until_complete(wrapper())
  File "uvloop/loop.pyx", line 1518, in uvloop.loop.Loop.run_until_complete
  File "/usr/local/lib/python3.10/dist-packages/uvloop/__init__.py", line 61, in wrapper
    return await main
  File "/workspace/vllm/vllm/entrypoints/openai/api_server.py", line 1324, in run_server
    async with build_async_engine_client(args) as engine_client:
  File "/usr/lib/python3.10/contextlib.py", line 199, in __aenter__
    return await anext(self.gen)
  File "/workspace/vllm/vllm/entrypoints/openai/api_server.py", line 153, in build_async_engine_client
    async with build_async_engine_client_from_engine_args(
  File "/usr/lib/python3.10/contextlib.py", line 199, in __aenter__
    return await anext(self.gen)
  File "/workspace/vllm/vllm/entrypoints/openai/api_server.py", line 280, in build_async_engine_client_from_engine_args
    raise RuntimeError(
RuntimeError: Engine process failed to start. See stack trace for the root cause.

After I set the engine to V0 still not working, By the way, even if only the V0 version is available, it should automatically attempt to fall back the engine to V0 when detecting that the device is set to CPU.

[princepride]

@DarkLight1337 Can you assign this task to me, I can try to solve it, thanks.

[DarkLight1337]

Thanks for helping look into this

[kebe7jun]

This seems to be because current_platform is automatically set to CudaPlatform on the node with GPU, causing the initialized Worker to be not a CPUWorker.
Currently, current_platform is initialized when importing vllm, does not care about the --device parameter, and is used a lot.
If you want to temporarily solve this problem, you can install vllm in cpu mode, VLLM_TARGET_DEVICE=cpu pip install -e .
If you want to fix this problem, the easiest way may be to add an environment variable to force the Platform to be specified, but it doesn't seem very elegant. Do you have any ideas? @DarkLight1337

[DarkLight1337]

I think vLLM is only built against one platform at a time, so you do need to follow the installation instructions and set VLLM_TARGET_DEVICE to build the CPU backend. Correct me if I'm wrong @youkaichao @Isotr0py

[princepride]

@kebe7jun You are basically correct. The specific error occurred because of the code commit ad34c0d#diff-3a55a5c83a819b573e10f49bc524c33c1734aeab5166d99fc64033fce4464512 in vllm==0.7.0. This commit refactored the executor, removing the cpu_executor. After this refactoring, the worker initialization now happens within self.collective_rpc("init_device"). This caused the platform to be determined from an earlier initialization stage. The platform's initialization, in turn, depends on the vllm version name, ultimately leading to vllm always being set to the cuda platform.

[princepride]

I believe that making the --device setting effective should solve this problem. I noticed that in this code refactoring, it seems that no unit tests were written to cover support for different devices.

[princepride]

By the way, starting from vllm==0.6.5, setting device to cpu causes an error: #11986. The last version that supported setting --device cpu was vllm==0.6.4.post1.

[Isotr0py]

Hmmm, I tried vllm==0.6.4.post1 with --device cpu, but it can't exactly work because it will still try to use Xformers backend on CPU runner:

(vllm) /tmp# vllm serve facebook/opt-125m --device cpu
INFO 05-20 06:46:59 api_server.py:585] vLLM API server version 0.6.4.post1
INFO 05-20 06:46:59 api_server.py:586] args: Namespace(subparser='serve', model_tag='facebook/opt-125m', config='', host=None, port=8000, uvicorn_log_level='info', allow_credentials=False, allowed_origins=['*'], allowed_methods=['*'], allowed_headers=['*'], api_key=None, lora_modules=None, prompt_adapters=None, chat_template=None, response_role='assistant', ssl_keyfile=None, ssl_certfile=None, ssl_ca_certs=None, ssl_cert_reqs=0, root_path=None, middleware=[], return_tokens_as_token_ids=False, disable_frontend_multiprocessing=False, enable_auto_tool_choice=False, tool_call_parser=None, tool_parser_plugin='', model='facebook/opt-125m', task='auto', tokenizer=None, skip_tokenizer_init=False, revision=None, code_revision=None, tokenizer_revision=None, tokenizer_mode='auto', chat_template_text_format='string', trust_remote_code=False, allowed_local_media_path=None, download_dir=None, load_format='auto', config_format=<ConfigFormat.AUTO: 'auto'>, dtype='auto', kv_cache_dtype='auto', quantization_param_path=None, max_model_len=None, guided_decoding_backend='outlines', distributed_executor_backend=None, worker_use_ray=False, pipeline_parallel_size=1, tensor_parallel_size=1, max_parallel_loading_workers=None, ray_workers_use_nsight=False, block_size=16, enable_prefix_caching=False, disable_sliding_window=False, use_v2_block_manager=False, num_lookahead_slots=0, seed=0, swap_space=4, cpu_offload_gb=0, gpu_memory_utilization=0.9, num_gpu_blocks_override=None, max_num_batched_tokens=None, max_num_seqs=256, max_logprobs=20, disable_log_stats=False, quantization=None, rope_scaling=None, rope_theta=None, hf_overrides=None, enforce_eager=False, max_seq_len_to_capture=8192, disable_custom_all_reduce=False, tokenizer_pool_size=0, tokenizer_pool_type='ray', tokenizer_pool_extra_config=None, limit_mm_per_prompt=None, mm_processor_kwargs=None, enable_lora=False, enable_lora_bias=False, max_loras=1, max_lora_rank=16, lora_extra_vocab_size=256, lora_dtype='auto', long_lora_scaling_factors=None, max_cpu_loras=None, fully_sharded_loras=False, enable_prompt_adapter=False, max_prompt_adapters=1, max_prompt_adapter_token=0, device='cpu', num_scheduler_steps=1, multi_step_stream_outputs=True, scheduler_delay_factor=0.0, enable_chunked_prefill=None, speculative_model=None, speculative_model_quantization=None, num_speculative_tokens=None, speculative_disable_mqa_scorer=False, speculative_draft_tensor_parallel_size=None, speculative_max_model_len=None, speculative_disable_by_batch_size=None, ngram_prompt_lookup_max=None, ngram_prompt_lookup_min=None, spec_decoding_acceptance_method='rejection_sampler', typical_acceptance_sampler_posterior_threshold=None, typical_acceptance_sampler_posterior_alpha=None, disable_logprobs_during_spec_decoding=None, model_loader_extra_config=None, ignore_patterns=[], preemption_mode=None, served_model_name=None, qlora_adapter_name_or_path=None, otlp_traces_endpoint=None, collect_detailed_traces=None, disable_async_output_proc=False, scheduling_policy='fcfs', override_neuron_config=None, override_pooler_config=None, disable_log_requests=False, max_log_len=None, disable_fastapi_docs=False, enable_prompt_tokens_details=False, dispatch_function=<function serve at 0x784f4cf29bc0>)
INFO 05-20 06:46:59 api_server.py:175] Multiprocessing frontend to use ipc:///tmp/35e5f859-feea-4848-8cca-0ed31b255f71 for IPC Path.
INFO 05-20 06:46:59 api_server.py:194] Started engine process with PID 3333
WARNING 05-20 06:47:09 arg_utils.py:1075] [DEPRECATED] Block manager v1 has been removed, and setting --use-v2-block-manager to True or False has no effect on vLLM behavior. Please remove --use-v2-block-manager in your engine argument. If your use case is not supported by SelfAttnBlockSpaceManager (i.e. block manager v2), please file an issue with detailed information.
WARNING 05-20 06:47:09 config.py:487] Async output processing is only supported for CUDA, TPU, XPU and HPU.Disabling it for other platforms.
tokenizer_config.json: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 685/685 [00:00<00:00, 3.21MB/s]
vocab.json: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 899k/899k [00:00<00:00, 14.2MB/s]
merges.txt: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 456k/456k [00:00<00:00, 2.54MB/s]
special_tokens_map.json: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 441/441 [00:00<00:00, 2.30MB/s]
WARNING 05-20 06:47:14 arg_utils.py:1075] [DEPRECATED] Block manager v1 has been removed, and setting --use-v2-block-manager to True or False has no effect on vLLM behavior. Please remove --use-v2-block-manager in your engine argument. If your use case is not supported by SelfAttnBlockSpaceManager (i.e. block manager v2), please file an issue with detailed information.
WARNING 05-20 06:47:14 config.py:487] Async output processing is only supported for CUDA, TPU, XPU and HPU.Disabling it for other platforms.
INFO 05-20 06:47:14 llm_engine.py:249] Initializing an LLM engine (v0.6.4.post1) with config: model='facebook/opt-125m', speculative_config=None, tokenizer='facebook/opt-125m', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, override_neuron_config=None, tokenizer_revision=None, trust_remote_code=False, dtype=torch.float16, max_seq_len=2048, download_dir=None, load_format=LoadFormat.AUTO, tensor_parallel_size=1, pipeline_parallel_size=1, disable_custom_all_reduce=False, quantization=None, enforce_eager=False, kv_cache_dtype=auto, quantization_param_path=None, device_config=cpu, decoding_config=DecodingConfig(guided_decoding_backend='outlines'), observability_config=ObservabilityConfig(otlp_traces_endpoint=None, collect_model_forward_time=False, collect_model_execute_time=False), seed=0, served_model_name=facebook/opt-125m, num_scheduler_steps=1, chunked_prefill_enabled=False multi_step_stream_outputs=True, enable_prefix_caching=False, use_async_output_proc=False, use_cached_outputs=True, chat_template_text_format=string, mm_processor_kwargs=None, pooler_config=None)
generation_config.json: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 137/137 [00:00<00:00, 779kB/s]
WARNING 05-20 06:47:15 cpu_executor.py:320] CUDA graph is not supported on CPU, fallback to the eager mode.
WARNING 05-20 06:47:15 cpu_executor.py:350] Environment variable VLLM_CPU_KVCACHE_SPACE (GB) for CPU backend is not set, using 4 by default.
INFO 05-20 06:47:15 selector.py:261] Cannot use FlashAttention-2 backend for Volta and Turing GPUs.
INFO 05-20 06:47:15 selector.py:144] Using XFormers backend.

Besides, I don't think we build CPU custom ops in release wheel as well, because CPU ops will be only built when VLLM_TARGET_DEVICE=CPU:

vllm/CMakeLists.txt

Lines 85 to 93 in d981396

	if (NOT VLLM_TARGET_DEVICE STREQUAL "cuda" AND
	NOT VLLM_TARGET_DEVICE STREQUAL "rocm")
	if (VLLM_TARGET_DEVICE STREQUAL "cpu")
	include(${CMAKE_CURRENT_LIST_DIR}/cmake/cpu_extension.cmake)
	else()
	return()
	endif()
	return()
	endif()

[princepride]

@Isotr0py That's weird, I installed vllm==0.6.4.post1, my log is different with you

root@c8f90d3b27c5:/workspace# vllm serve facebook/opt-125m --device cpu
INFO 05-20 07:08:58 api_server.py:585] vLLM API server version 0.6.4.post1
INFO 05-20 07:08:58 api_server.py:586] args: Namespace(subparser='serve', model_tag='facebook/opt-125m', config='', host=None, port=8000, uvicorn_log_level='info', allow_credentials=False, allowed_origins=['*'], allowed_methods=['*'], allowed_headers=['*'], api_key=None, lora_modules=None, prompt_adapters=None, chat_template=None, response_role='assistant', ssl_keyfile=None, ssl_certfile=None, ssl_ca_certs=None, ssl_cert_reqs=0, root_path=None, middleware=[], return_tokens_as_token_ids=False, disable_frontend_multiprocessing=False, enable_auto_tool_choice=False, tool_call_parser=None, tool_parser_plugin='', model='facebook/opt-125m', task='auto', tokenizer=None, skip_tokenizer_init=False, revision=None, code_revision=None, tokenizer_revision=None, tokenizer_mode='auto', chat_template_text_format='string', trust_remote_code=False, allowed_local_media_path=None, download_dir=None, load_format='auto', config_format=<ConfigFormat.AUTO: 'auto'>, dtype='auto', kv_cache_dtype='auto', quantization_param_path=None, max_model_len=None, guided_decoding_backend='outlines', distributed_executor_backend=None, worker_use_ray=False, pipeline_parallel_size=1, tensor_parallel_size=1, max_parallel_loading_workers=None, ray_workers_use_nsight=False, block_size=16, enable_prefix_caching=False, disable_sliding_window=False, use_v2_block_manager=False, num_lookahead_slots=0, seed=0, swap_space=4, cpu_offload_gb=0, gpu_memory_utilization=0.9, num_gpu_blocks_override=None, max_num_batched_tokens=None, max_num_seqs=256, max_logprobs=20, disable_log_stats=False, quantization=None, rope_scaling=None, rope_theta=None, hf_overrides=None, enforce_eager=False, max_seq_len_to_capture=8192, disable_custom_all_reduce=False, tokenizer_pool_size=0, tokenizer_pool_type='ray', tokenizer_pool_extra_config=None, limit_mm_per_prompt=None, mm_processor_kwargs=None, enable_lora=False, enable_lora_bias=False, max_loras=1, max_lora_rank=16, lora_extra_vocab_size=256, lora_dtype='auto', long_lora_scaling_factors=None, max_cpu_loras=None, fully_sharded_loras=False, enable_prompt_adapter=False, max_prompt_adapters=1, max_prompt_adapter_token=0, device='cpu', num_scheduler_steps=1, multi_step_stream_outputs=True, scheduler_delay_factor=0.0, enable_chunked_prefill=None, speculative_model=None, speculative_model_quantization=None, num_speculative_tokens=None, speculative_disable_mqa_scorer=False, speculative_draft_tensor_parallel_size=None, speculative_max_model_len=None, speculative_disable_by_batch_size=None, ngram_prompt_lookup_max=None, ngram_prompt_lookup_min=None, spec_decoding_acceptance_method='rejection_sampler', typical_acceptance_sampler_posterior_threshold=None, typical_acceptance_sampler_posterior_alpha=None, disable_logprobs_during_spec_decoding=None, model_loader_extra_config=None, ignore_patterns=[], preemption_mode=None, served_model_name=None, qlora_adapter_name_or_path=None, otlp_traces_endpoint=None, collect_detailed_traces=None, disable_async_output_proc=False, scheduling_policy='fcfs', override_neuron_config=None, override_pooler_config=None, disable_log_requests=False, max_log_len=None, disable_fastapi_docs=False, enable_prompt_tokens_details=False, dispatch_function=<function serve at 0x7031a7b2fa30>)
INFO 05-20 07:08:58 api_server.py:175] Multiprocessing frontend to use ipc:///tmp/80b97a4d-8e16-4d8a-a9a0-099787338eb7 for IPC Path.
INFO 05-20 07:08:58 api_server.py:194] Started engine process with PID 1128
config.json: 100%|██████████████████████████████████| 651/651 [00:00<00:00, 5.99MB/s]
WARNING 05-20 07:09:04 arg_utils.py:1075] [DEPRECATED] Block manager v1 has been removed, and setting --use-v2-block-manager to True or False has no effect on vLLM behavior. Please remove --use-v2-block-manager in your engine argument. If your use case is not supported by SelfAttnBlockSpaceManager (i.e. block manager v2), please file an issue with detailed information.
WARNING 05-20 07:09:04 config.py:487] Async output processing is only supported for CUDA, TPU, XPU and HPU.Disabling it for other platforms.
tokenizer_config.json: 100%|████████████████████████| 685/685 [00:00<00:00, 7.29MB/s]
vocab.json: 100%|█████████████████████████████████| 899k/899k [00:00<00:00, 26.7MB/s]
merges.txt: 100%|█████████████████████████████████| 456k/456k [00:00<00:00, 6.94MB/s]
special_tokens_map.json: 100%|██████████████████████| 441/441 [00:00<00:00, 6.10MB/s]
WARNING 05-20 07:09:09 arg_utils.py:1075] [DEPRECATED] Block manager v1 has been removed, and setting --use-v2-block-manager to True or False has no effect on vLLM behavior. Please remove --use-v2-block-manager in your engine argument. If your use case is not supported by SelfAttnBlockSpaceManager (i.e. block manager v2), please file an issue with detailed information.
WARNING 05-20 07:09:09 config.py:487] Async output processing is only supported for CUDA, TPU, XPU and HPU.Disabling it for other platforms.
INFO 05-20 07:09:09 llm_engine.py:249] Initializing an LLM engine (v0.6.4.post1) with config: model='facebook/opt-125m', speculative_config=None, tokenizer='facebook/opt-125m', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, override_neuron_config=None, tokenizer_revision=None, trust_remote_code=False, dtype=torch.float16, max_seq_len=2048, download_dir=None, load_format=LoadFormat.AUTO, tensor_parallel_size=1, pipeline_parallel_size=1, disable_custom_all_reduce=False, quantization=None, enforce_eager=False, kv_cache_dtype=auto, quantization_param_path=None, device_config=cpu, decoding_config=DecodingConfig(guided_decoding_backend='outlines'), observability_config=ObservabilityConfig(otlp_traces_endpoint=None, collect_model_forward_time=False, collect_model_execute_time=False), seed=0, served_model_name=facebook/opt-125m, num_scheduler_steps=1, chunked_prefill_enabled=False multi_step_stream_outputs=True, enable_prefix_caching=False, use_async_output_proc=False, use_cached_outputs=True, chat_template_text_format=string, mm_processor_kwargs=None, pooler_config=None)
generation_config.json: 100%|███████████████████████| 137/137 [00:00<00:00, 1.13MB/s]
WARNING 05-20 07:09:11 cpu_executor.py:320] CUDA graph is not supported on CPU, fallback to the eager mode.
WARNING 05-20 07:09:11 cpu_executor.py:350] Environment variable VLLM_CPU_KVCACHE_SPACE (GB) for CPU backend is not set, using 4 by default.
INFO 05-20 07:09:11 selector.py:135] Using Flash Attention backend.
INFO 05-20 07:09:12 weight_utils.py:243] Using model weights format ['*.bin']
Xet Storage is enabled for this repo, but the 'hf_xet' package is not installed. Falling back to regular HTTP download. For better performance, install the package with: `pip install huggingface_hub[hf_xet]` or `pip install hf_xet`
pytorch_model.bin: 100%|███████████████████████████| 251M/251M [00:02<00:00, 115MB/s]
Loading pt checkpoint shards:   0% Completed | 0/1 [00:00<?, ?it/s]
/usr/local/lib/python3.10/dist-packages/vllm/model_executor/model_loader/weight_utils.py:425: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
  state = torch.load(bin_file, map_location="cpu")
Loading pt checkpoint shards: 100% Completed | 1/1 [00:00<00:00,  2.80it/s]
Loading pt checkpoint shards: 100% Completed | 1/1 [00:00<00:00,  2.80it/s]

INFO 05-20 07:09:15 cpu_executor.py:205] # CPU blocks: 7281
INFO 05-20 07:09:16 api_server.py:249] vLLM to use /tmp/tmpaizfkobw as PROMETHEUS_MULTIPROC_DIR
INFO 05-20 07:09:16 launcher.py:19] Available routes are:
INFO 05-20 07:09:16 launcher.py:27] Route: /openapi.json, Methods: GET, HEAD
INFO 05-20 07:09:16 launcher.py:27] Route: /docs, Methods: GET, HEAD
INFO 05-20 07:09:16 launcher.py:27] Route: /docs/oauth2-redirect, Methods: GET, HEAD
INFO 05-20 07:09:16 launcher.py:27] Route: /redoc, Methods: GET, HEAD
INFO 05-20 07:09:16 launcher.py:27] Route: /health, Methods: GET
INFO 05-20 07:09:16 launcher.py:27] Route: /tokenize, Methods: POST
INFO 05-20 07:09:16 launcher.py:27] Route: /detokenize, Methods: POST
INFO 05-20 07:09:16 launcher.py:27] Route: /v1/models, Methods: GET
INFO 05-20 07:09:16 launcher.py:27] Route: /version, Methods: GET
INFO 05-20 07:09:16 launcher.py:27] Route: /v1/chat/completions, Methods: POST
INFO 05-20 07:09:16 launcher.py:27] Route: /v1/completions, Methods: POST
INFO 05-20 07:09:16 launcher.py:27] Route: /v1/embeddings, Methods: POST
INFO:     Started server process [1042]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:8000 (Press CTRL+C to quit)

Here is my environment:

root@c8f90d3b27c5:/workspace# python collect_env.py
Collecting environment information...
/usr/local/lib/python3.10/dist-packages/_distutils_hack/__init__.py:33: UserWarning: Setuptools is replacing distutils.
  warnings.warn("Setuptools is replacing distutils.")
==============================
        System Info
==============================
OS                           : Ubuntu 22.04.3 LTS (x86_64)
GCC version                  : (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version                : Could not collect
CMake version                : Could not collect
Libc version                 : glibc-2.35

==============================
       PyTorch Info
==============================
PyTorch version              : 2.5.1+cu124
Is debug build               : False
CUDA used to build PyTorch   : 12.4
ROCM used to build PyTorch   : N/A

==============================
      Python Environment
==============================
Python version               : 3.10.12 (main, Jun 11 2023, 05:26:28) [GCC 11.4.0] (64-bit runtime)
Python platform              : Linux-6.5.0-35-generic-x86_64-with-glibc2.35

==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : 11.8.89
CUDA_MODULE_LOADING set to   : LAZY
GPU models and configuration : GPU 0: NVIDIA RTX A6000
Nvidia driver version        : 555.42.02
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:                      48 bits physical, 48 bits virtual
Byte Order:                         Little Endian
CPU(s):                             64
On-line CPU(s) list:                0-63
Vendor ID:                          AuthenticAMD
Model name:                         AMD EPYC 7343 16-Core Processor
CPU family:                         25
Model:                              1
Thread(s) per core:                 2
Core(s) per socket:                 16
Socket(s):                          2
Stepping:                           1
Frequency boost:                    enabled
CPU max MHz:                        3940.6250
CPU min MHz:                        1500.0000
BogoMIPS:                           6400.20
Flags:                              fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 invpcid_single hw_pstate ssbd mba ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 invpcid cqm rdt_a rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin brs arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold v_vmsave_vmload vgif v_spec_ctrl umip pku ospke vaes vpclmulqdq rdpid overflow_recov succor smca
Virtualization:                     AMD-V
L1d cache:                          1 MiB (32 instances)
L1i cache:                          1 MiB (32 instances)
L2 cache:                           16 MiB (32 instances)
L3 cache:                           256 MiB (8 instances)
NUMA node(s):                       8
NUMA node0 CPU(s):                  0-3,32-35
NUMA node1 CPU(s):                  4-7,36-39
NUMA node2 CPU(s):                  8-11,40-43
NUMA node3 CPU(s):                  12-15,44-47
NUMA node4 CPU(s):                  16-19,48-51
NUMA node5 CPU(s):                  20-23,52-55
NUMA node6 CPU(s):                  24-27,56-59
NUMA node7 CPU(s):                  28-31,60-63
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 Retbleed:             Not affected
Vulnerability Spec rstack overflow: Mitigation; Safe RET
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; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds:                Not affected
Vulnerability Tsx async abort:      Not affected

==============================
Versions of relevant libraries
==============================
[pip3] numpy==1.26.4
[pip3] nvidia-cublas-cu12==12.4.5.8
[pip3] nvidia-cuda-cupti-cu12==12.4.127
[pip3] nvidia-cuda-nvrtc-cu12==12.4.127
[pip3] nvidia-cuda-runtime-cu12==12.4.127
[pip3] nvidia-cudnn-cu12==9.1.0.70
[pip3] nvidia-cufft-cu12==11.2.1.3
[pip3] nvidia-curand-cu12==10.3.5.147
[pip3] nvidia-cusolver-cu12==11.6.1.9
[pip3] nvidia-cusparse-cu12==12.3.1.170
[pip3] nvidia-ml-py==12.575.51
[pip3] nvidia-nccl-cu12==2.21.5
[pip3] nvidia-nvjitlink-cu12==12.4.127
[pip3] nvidia-nvtx-cu12==12.4.127
[pip3] pyzmq==24.0.1
[pip3] torch==2.5.1
[pip3] torchaudio==2.1.0+cu118
[pip3] torchvision==0.20.1
[pip3] transformers==4.51.3
[pip3] triton==3.1.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : Could not collect
Neuron SDK Version           : N/A
vLLM Version                 : 0.6.4.post1
vLLM Build Flags:
  CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
        GPU0    CPU Affinity    NUMA Affinity   GPU NUMA ID
GPU0     X      20-23,52-55     5               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
==============================
NVIDIA_VISIBLE_DEVICES=GPU-84a5e858-678e-3047-411a-2236bafbb484
NVIDIA_REQUIRE_CUDA=cuda>=11.8 brand=tesla,driver>=450,driver<451 brand=tesla,driver>=470,driver<471 brand=unknown,driver>=470,driver<471 brand=nvidia,driver>=470,driver<471 brand=nvidiartx,driver>=470,driver<471 brand=geforce,driver>=470,driver<471 brand=geforcertx,driver>=470,driver<471 brand=quadro,driver>=470,driver<471 brand=quadrortx,driver>=470,driver<471 brand=titan,driver>=470,driver<471 brand=titanrtx,driver>=470,driver<471 brand=tesla,driver>=510,driver<511 brand=unknown,driver>=510,driver<511 brand=nvidia,driver>=510,driver<511 brand=nvidiartx,driver>=510,driver<511 brand=geforce,driver>=510,driver<511 brand=geforcertx,driver>=510,driver<511 brand=quadro,driver>=510,driver<511 brand=quadrortx,driver>=510,driver<511 brand=titan,driver>=510,driver<511 brand=titanrtx,driver>=510,driver<511 brand=tesla,driver>=515,driver<516 brand=unknown,driver>=515,driver<516 brand=nvidia,driver>=515,driver<516 brand=nvidiartx,driver>=515,driver<516 brand=geforce,driver>=515,driver<516 brand=geforcertx,driver>=515,driver<516 brand=quadro,driver>=515,driver<516 brand=quadrortx,driver>=515,driver<516 brand=titan,driver>=515,driver<516 brand=titanrtx,driver>=515,driver<516
NCCL_VERSION=2.15.5-1
NVIDIA_DRIVER_CAPABILITIES=compute,display,graphics,utility,video
NVIDIA_PRODUCT_NAME=CUDA
CUDA_VERSION=11.8.0
LD_LIBRARY_PATH=/usr/local/lib/python3.10/dist-packages/cv2/../../lib64:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
CUDA_MODULE_LOADING=LAZY

I am using the runpod RTX A6000 Pod Template: runpod/pytorch:2.1.0-py3.10-cuda11.8.0-devel-ubuntu22.04

[Isotr0py]

That's weird, I installed vllm==0.6.4.post1, my log is different with you

That's just because I'm using an old GPU to test this.

INFO 05-20 07:09:11 selector.py:135] Using Flash Attention backend.

However, Flash Attention backend won't work on CPU backend as well. If you send one request to the server, it will crash the engine due to failing attention maetadata creation.