[Bug]: Device selection broken in v0.9

[derfred]

Your current environment

The output of python collect_env.py

(venv) root@instance-....:~# python collect_env.py
INFO 06-03 08:36:06 [__init__.py:243] Automatically detected platform cuda.
Collecting environment information...
==============================
        System Info
==============================
OS                           : Ubuntu 24.04.2 LTS (x86_64)
GCC version                  : (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0
Clang version                : Could not collect
CMake version                : Could not collect
Libc version                 : glibc-2.39

==============================
       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               : 3.12.3 (main, Feb  4 2025, 14:48:35) [GCC 13.3.0] (64-bit runtime)
Python platform              : Linux-6.8.0-59-generic-x86_64-with-glibc2.39

==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : 12.6.85
CUDA_MODULE_LOADING set to   : LAZY
GPU models and configuration : 
GPU 0: NVIDIA A100-SXM4-80GB
GPU 1: NVIDIA A100-SXM4-80GB

Nvidia driver version        : 570.133.20
cuDNN version                : Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.9.9.0
/usr/lib/x86_64-linux-gnu/libcudnn_adv.so.9.9.0
/usr/lib/x86_64-linux-gnu/libcudnn_cnn.so.9.9.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_precompiled.so.9.9.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_runtime_compiled.so.9.9.0
/usr/lib/x86_64-linux-gnu/libcudnn_graph.so.9.9.0
/usr/lib/x86_64-linux-gnu/libcudnn_heuristic.so.9.9.0
/usr/lib/x86_64-linux-gnu/libcudnn_ops.so.9.9.0
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):                               44
On-line CPU(s) list:                  0-43
Vendor ID:                            AuthenticAMD
BIOS Vendor ID:                       QEMU
Model name:                           AMD EPYC 7642 48-Core Processor
BIOS Model name:                      pc-q35-6.2  CPU @ 2.0GHz
BIOS CPU family:                      1
CPU family:                           23
Model:                                49
Thread(s) per core:                   1
Core(s) per socket:                   1
Socket(s):                            44
Stepping:                             0
BogoMIPS:                             4599.99
Flags:                                fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm rep_good nopl cpuid extd_apicid tsc_known_freq pni pclmulqdq ssse3 fma cx16 sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy svm cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw perfctr_core ssbd ibrs ibpb stibp vmmcall fsgsbase tsc_adjust bmi1 avx2 smep bmi2 rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 clzero xsaveerptr wbnoinvd arat npt lbrv nrip_save tsc_scale vmcb_clean pausefilter pfthreshold v_vmsave_vmload vgif umip rdpid arch_capabilities
Virtualization:                       AMD-V
Hypervisor vendor:                    KVM
Virtualization type:                  full
L1d cache:                            2.8 MiB (44 instances)
L1i cache:                            2.8 MiB (44 instances)
L2 cache:                             22 MiB (44 instances)
L3 cache:                             704 MiB (44 instances)
NUMA node(s):                         1
NUMA node0 CPU(s):                    0-43
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:               Mitigation; untrained return thunk; SMT disabled
Vulnerability Spec rstack overflow:   Vulnerable: Safe RET, no microcode
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; STIBP disabled; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds:                  Not affected
Vulnerability Tsx async abort:        Not affected

==============================
Versions of relevant libraries
==============================
[pip3] flashinfer-python==0.2.5+cu126torch2.6
[pip3] mypy_extensions==1.1.0
[pip3] numpy==1.26.4
[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==4.52.4
[pip3] triton==3.3.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : Could not collect
Neuron SDK Version           : N/A
vLLM Version                 : 0.9.0.1
vLLM Build Flags:
  CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
  	GPU0	GPU1	CPU Affinity	NUMA Affinity	GPU NUMA ID
GPU0	 X 	NV2	0-43	0		N/A
GPU1	NV2	 X 	0-43	0		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_CUMEM_ENABLE=0
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
CUDA_MODULE_LOADING=LAZY

🐛 Describe the bug

Create a new instance of LLM with device specified in a multi-gpu machine

from vllm import LLM
llm = LLM(device="cuda:1", quantization='bitsandbytes', load_format='bitsandbytes', model='unsloth/Mistral-Small-3.1-24B-Instruct-2503-bnb-4bit')

in v0.9.0.1 the model is then placed on cuda instead of cuda:1 as can also be seen in the log output (look for device_config):

...
INFO 06-03 08:36:34 [core.py:65] Initializing a V1 LLM engine (v0.9.0.1) with config: model='unsloth/Mistral-Small-3.1-24B-Instruct-2503-bnb-4bit', speculative_config=None, tokenizer='unsloth/Mistral-Small-3.1-24B-Instruct-2503-bnb-4bit', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, override_neuron_config={}, tokenizer_revision=None, trust_remote_code=False, dtype=torch.bfloat16, max_seq_len=14500, download_dir=None, load_format=LoadFormat.BITSANDBYTES, tensor_parallel_size=1, pipeline_parallel_size=1, disable_custom_all_reduce=False, quantization=bitsandbytes, enforce_eager=False, kv_cache_dtype=auto,  device_config=cuda, decoding_config=DecodingConfig(backend='auto', 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=0, served_model_name=unsloth/Mistral-Small-3.1-24B-Instruct-2503-bnb-4bit, num_scheduler_steps=1, multi_step_stream_outputs=True, enable_prefix_caching=True, chunked_prefill_enabled=True, use_async_output_proc=True, pooler_config=None, compilation_config={"level": 3, "custom_ops": ["none"], "splitting_ops": ["vllm.unified_attention", "vllm.unified_attention_with_output"], "compile_sizes": [], "inductor_compile_config": {"enable_auto_functionalized_v2": false}, "use_cudagraph": true, "cudagraph_num_of_warmups": 1, "cudagraph_capture_sizes": [512, 504, 496, 488, 480, 472, 464, 456, 448, 440, 432, 424, 416, 408, 400, 392, 384, 376, 368, 360, 352, 344, 336, 328, 320, 312, 304, 296, 288, 280, 272, 264, 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": 512}
...

compare to v.8.5.post1

...
INFO 06-03 08:26:42 [core.py:58] Initializing a V1 LLM engine (v0.8.5.post1) with config: model='unsloth/Mistral-Small-3.1-24B-Instruct-2503-bnb-4bit', speculative_config=None, tokenizer='unsloth/Mistral-Small-3.1-24B-Instruct-2503-bnb-4bit', skip_tokenizer_init=False, tokenizer_mode=auto, revision=None, override_neuron_config=None, tokenizer_revision=None, trust_remote_code=False, dtype=torch.bfloat16, max_seq_len=14500, download_dir=None, load_format=LoadFormat.BITSANDBYTES, tensor_parallel_size=1, pipeline_parallel_size=1, disable_custom_all_reduce=False, quantization=bitsandbytes, enforce_eager=False, kv_cache_dtype=auto,  device_config=cuda:1, decoding_config=DecodingConfig(guided_decoding_backend='auto', reasoning_backend=None), observability_config=ObservabilityConfig(show_hidden_metrics=False, otlp_traces_endpoint=None, collect_model_forward_time=False, collect_model_execute_time=False), seed=None, served_model_name=unsloth/Mistral-Small-3.1-24B-Instruct-2503-bnb-4bit, num_scheduler_steps=1, multi_step_stream_outputs=True, enable_prefix_caching=True, chunked_prefill_enabled=True, use_async_output_proc=True, disable_mm_preprocessor_cache=False, mm_processor_kwargs=None, pooler_config=None, compilation_config={"level":3,"custom_ops":["none"],"splitting_ops":["vllm.unified_attention","vllm.unified_attention_with_output"],"use_inductor":true,"compile_sizes":[],"use_cudagraph":true,"cudagraph_num_of_warmups":1,"cudagraph_capture_sizes":[512,504,496,488,480,472,464,456,448,440,432,424,416,408,400,392,384,376,368,360,352,344,336,328,320,312,304,296,288,280,272,264,256,248,240,232,224,216,208,200,192,184,176,168,160,152,144,136,128,120,112,104,96,88,80,72,64,56,48,40,32,24,16,8,4,2,1],"max_capture_size":512}
...

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.

[DarkLight1337]

The --device argument is not intended to be used for selecting the specific device in vLLM. In fact, we have deprecated this argument in #18399.

To set the specific device, we recommend instead using environment variables such as CUDA_VISIBLE_DEVICES.

[derfred]

@DarkLight1337 Thanks for the info. That worked for me!