Skip to content

Instantly share code, notes, and snippets.

@nerdalert

nerdalert/rocm-run-err.md

Last active April 24, 2025 18:58
Show Gist options
  • Save nerdalert/5ddf78dc570947a72e1a396e00c06cbc to your computer and use it in GitHub Desktop.
Save nerdalert/5ddf78dc570947a72e1a396e00c06cbc to your computer and use it in GitHub Desktop.
Download ZIP
Raw
rocm-run-err.md 
cloned and made sure the nightly image was up to date.

docker run -it --ipc=host --network=host --group-add render \
    --privileged --security-opt seccomp=unconfined \
    --cap-add=CAP_SYS_ADMIN --cap-add=SYS_PTRACE \
    --device=/dev/kfd --device=/dev/dri --device=/dev/mem \
    -e HF_TOKEN=$HF_TOKEN -e HF_HOME=/data/model_cache \
    -e MODEL=$MODEL \
    -v $HF_HOME:/data/model_cache \
    -v $PWD/vllm:/app/vllm-upstream \
    rocm/vllm-dev:nightly /bin/bash

cd /app/vllm-upstream
pip install -e . --no-build-isolation



root@ENC1-CLS01-SVR08:/app# HIP_VISIBLE_DEVICES=2  MODEL=mistralai/Mistral-Small-24B-Instruct-2501 vllm serve $MODEL --disable-log-requests

INFO 04-24 18:55:06 [__init__.py:239] Automatically detected platform rocm.
INFO 04-24 18:55:17 [api_server.py:1043] vLLM API server version 0.8.5.dev209+g6d0df0ebe
INFO 04-24 18:55:17 [api_server.py:1044] args: Namespace(subparser='serve', model_tag='mistralai/Mistral-Small-24B-Instruct-2501', config='', host=None, port=8000, uvicorn_log_level='info', disable_uvicorn_access_log=False, allow_credentials=False, allowed_origins=['*'], allowed_methods=['*'], allowed_headers=['*'], api_key=None, lora_modules=None, prompt_adapters=None, chat_template=None, chat_template_content_format='auto', response_role='assistant', ssl_keyfile=None, ssl_certfile=None, ssl_ca_certs=None, enable_ssl_refresh=False, ssl_cert_reqs=0, root_path=None, middleware=[], return_tokens_as_token_ids=False, disable_frontend_multiprocessing=False, enable_request_id_headers=False, enable_auto_tool_choice=False, tool_call_parser=None, tool_parser_plugin='', model='mistralai/Mistral-Small-24B-Instruct-2501', task='auto', tokenizer=None, hf_config_path=None, skip_tokenizer_init=False, revision=None, code_revision=None, tokenizer_revision=None, tokenizer_mode='auto', trust_remote_code=False, allowed_local_media_path=None, load_format='auto', download_dir=None, model_loader_extra_config={}, use_tqdm_on_load=True, config_format=<ConfigFormat.AUTO: 'auto'>, dtype='auto', max_model_len=None, guided_decoding_backend='auto', reasoning_parser=None, logits_processor_pattern=None, model_impl='auto', distributed_executor_backend=None, pipeline_parallel_size=1, tensor_parallel_size=1, data_parallel_size=1, enable_expert_parallel=False, max_parallel_loading_workers=None, ray_workers_use_nsight=False, disable_custom_all_reduce=False, block_size=None, gpu_memory_utilization=0.9, swap_space=4, kv_cache_dtype='auto', num_gpu_blocks_override=None, enable_prefix_caching=None, prefix_caching_hash_algo='builtin', cpu_offload_gb=0, calculate_kv_scales=False, disable_sliding_window=False, use_v2_block_manager=True, seed=None, max_logprobs=20, disable_log_stats=False, quantization=None, rope_scaling=None, rope_theta=None, hf_token=None, hf_overrides=None, enforce_eager=False, max_seq_len_to_capture=8192, tokenizer_pool_size=0, tokenizer_pool_type='ray', tokenizer_pool_extra_config={}, limit_mm_per_prompt={}, mm_processor_kwargs=None, disable_mm_preprocessor_cache=False, enable_lora=None, 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=None, max_prompt_adapters=1, max_prompt_adapter_token=0, device='auto', num_scheduler_steps=1, speculative_config=None, ignore_patterns=[], preemption_mode=None, served_model_name=None, qlora_adapter_name_or_path=None, show_hidden_metrics_for_version=None, otlp_traces_endpoint=None, collect_detailed_traces=None, disable_async_output_proc=False, max_num_batched_tokens=None, max_num_seqs=None, max_num_partial_prefills=1, max_long_partial_prefills=1, long_prefill_token_threshold=0, num_lookahead_slots=0, scheduler_delay_factor=0.0, enable_chunked_prefill=None, multi_step_stream_outputs=True, scheduling_policy='fcfs', disable_chunked_mm_input=False, scheduler_cls='vllm.core.scheduler.Scheduler', override_neuron_config=None, override_pooler_config=None, compilation_config=None, kv_transfer_config=None, worker_cls='auto', worker_extension_cls='', generation_config='auto', override_generation_config=None, enable_sleep_mode=False, additional_config=None, enable_reasoning=False, disable_cascade_attn=False, disable_log_requests=True, max_log_len=None, disable_fastapi_docs=False, enable_prompt_tokens_details=False, enable_server_load_tracking=False, dispatch_function=<function ServeSubcommand.cmd at 0x7fe7181dfa60>)
INFO 04-24 18:55:32 [config.py:716] This model supports multiple tasks: {'embed', 'generate', 'reward', 'classify', 'score'}. Defaulting to 'generate'.
INFO 04-24 18:55:32 [arg_utils.py:1691] rocm is experimental on VLLM_USE_V1=1. Falling back to V0 Engine.
INFO 04-24 18:55:32 [api_server.py:246] Started engine process with PID 3997
/app/vllm-upstream/vllm/transformers_utils/tokenizer_group.py:23: FutureWarning: It is strongly recommended to run mistral models with `--tokenizer-mode "mistral"` to ensure correct encoding and decoding.
  self.tokenizer = get_tokenizer(self.tokenizer_id, **tokenizer_config)
INFO 04-24 18:55:35 [__init__.py:239] Automatically detected platform rocm.
INFO 04-24 18:55:45 [llm_engine.py:242] Initializing a V0 LLM engine (v0.8.5.dev209+g6d0df0ebe) with config: model='mistralai/Mistral-Small-24B-Instruct-2501', speculative_config=None, tokenizer='mistralai/Mistral-Small-24B-Instruct-2501', 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=32768, 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,  device_config=cuda, 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=mistralai/Mistral-Small-24B-Instruct-2501, num_scheduler_steps=1, multi_step_stream_outputs=True, enable_prefix_caching=None, chunked_prefill_enabled=False, use_async_output_proc=True, disable_mm_preprocessor_cache=False, mm_processor_kwargs=None, pooler_config=None, compilation_config={"splitting_ops":[],"compile_sizes":[],"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,
/app/vllm-upstream/vllm/transformers_utils/tokenizer_group.py:23: FutureWarning: It is strongly recommended to run mistral models with `--tokenizer-mode "mistral"` to ensure correct encoding and decoding.
  self.tokenizer = get_tokenizer(self.tokenizer_id, **tokenizer_config)
INFO 04-24 18:55:46 [rocm.py:186] None is not supported in AMD GPUs.
INFO 04-24 18:55:46 [rocm.py:187] Using ROCmFlashAttention backend.
INFO 04-24 18:55:46 [parallel_state.py:946] rank 0 in world size 1 is assigned as DP rank 0, PP rank 0, TP rank 0
INFO 04-24 18:55:46 [model_runner.py:1120] Starting to load model mistralai/Mistral-Small-24B-Instruct-2501...
WARNING 04-24 18:55:46 [rocm.py:288] Model architecture 'MistralForCausalLM' is partially supported by ROCm: Sliding window attention (SWA) is not yet supported in Triton flash attention. For half-precision SWA support, please use CK flash attention by setting `VLLM_USE_TRITON_FLASH_ATTN=0`
INFO 04-24 18:55:46 [weight_utils.py:265] Using model weights format ['*.safetensors']
Loading safetensors checkpoint shards:   0% Completed | 0/10 [00:00<?, ?it/s]
Loading safetensors checkpoint shards:  10% Completed | 1/10 [00:02<00:18,  2.09s/it]
Loading safetensors checkpoint shards:  20% Completed | 2/10 [00:04<00:17,  2.14s/it]
Loading safetensors checkpoint shards:  30% Completed | 3/10 [00:06<00:15,  2.21s/it]
Loading safetensors checkpoint shards:  40% Completed | 4/10 [00:08<00:13,  2.21s/it]
Loading safetensors checkpoint shards:  50% Completed | 5/10 [00:10<00:11,  2.20s/it]
Loading safetensors checkpoint shards:  60% Completed | 6/10 [00:13<00:08,  2.18s/it]
Loading safetensors checkpoint shards:  70% Completed | 7/10 [00:15<00:06,  2.19s/it]
Loading safetensors checkpoint shards:  80% Completed | 8/10 [00:17<00:04,  2.18s/it]
Loading safetensors checkpoint shards:  90% Completed | 9/10 [00:19<00:02,  2.20s/it]
Loading safetensors checkpoint shards: 100% Completed | 10/10 [00:21<00:00,  2.07s/it]
Loading safetensors checkpoint shards: 100% Completed | 10/10 [00:21<00:00,  2.15s/it]

INFO 04-24 18:56:08 [loader.py:458] Loading weights took 21.70 seconds
INFO 04-24 18:56:09 [model_runner.py:1156] Model loading took 44.1250 GiB and 22.298778 seconds
ERROR 04-24 18:56:09 [engine.py:448] expected np.ndarray (got numpy.ndarray)
ERROR 04-24 18:56:09 [engine.py:448] Traceback (most recent call last):
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/engine/multiprocessing/engine.py", line 436, in run_mp_engine
ERROR 04-24 18:56:09 [engine.py:448]     engine = MQLLMEngine.from_vllm_config(
ERROR 04-24 18:56:09 [engine.py:448]              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/engine/multiprocessing/engine.py", line 128, in from_vllm_config
ERROR 04-24 18:56:09 [engine.py:448]     return cls(
ERROR 04-24 18:56:09 [engine.py:448]            ^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/engine/multiprocessing/engine.py", line 82, in __init__
ERROR 04-24 18:56:09 [engine.py:448]     self.engine = LLMEngine(*args, **kwargs)
ERROR 04-24 18:56:09 [engine.py:448]                   ^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/engine/llm_engine.py", line 284, in __init__
ERROR 04-24 18:56:09 [engine.py:448]     self._initialize_kv_caches()
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/engine/llm_engine.py", line 428, in _initialize_kv_caches
ERROR 04-24 18:56:09 [engine.py:448]     self.model_executor.determine_num_available_blocks())
ERROR 04-24 18:56:09 [engine.py:448]     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/executor/executor_base.py", line 103, in determine_num_available_blocks
ERROR 04-24 18:56:09 [engine.py:448]     results = self.collective_rpc("determine_num_available_blocks")
ERROR 04-24 18:56:09 [engine.py:448]               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/executor/uniproc_executor.py", line 56, in collective_rpc
ERROR 04-24 18:56:09 [engine.py:448]     answer = run_method(self.driver_worker, method, args, kwargs)
ERROR 04-24 18:56:09 [engine.py:448]              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/utils.py", line 2446, in run_method
ERROR 04-24 18:56:09 [engine.py:448]     return func(*args, **kwargs)
ERROR 04-24 18:56:09 [engine.py:448]            ^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/usr/local/lib/python3.12/dist-packages/torch/utils/_contextlib.py", line 116, in decorate_context
ERROR 04-24 18:56:09 [engine.py:448]     return func(*args, **kwargs)
ERROR 04-24 18:56:09 [engine.py:448]            ^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/worker/worker.py", line 249, in determine_num_available_blocks
ERROR 04-24 18:56:09 [engine.py:448]     self.model_runner.profile_run()
ERROR 04-24 18:56:09 [engine.py:448]   File "/usr/local/lib/python3.12/dist-packages/torch/utils/_contextlib.py", line 116, in decorate_context
ERROR 04-24 18:56:09 [engine.py:448]     return func(*args, **kwargs)
ERROR 04-24 18:56:09 [engine.py:448]            ^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/worker/model_runner.py", line 1253, in profile_run
ERROR 04-24 18:56:09 [engine.py:448]     self._dummy_run(max_num_batched_tokens, max_num_seqs)
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/worker/model_runner.py", line 1365, in _dummy_run
ERROR 04-24 18:56:09 [engine.py:448]     model_input = self.prepare_model_input(
ERROR 04-24 18:56:09 [engine.py:448]                   ^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/worker/model_runner.py", line 1674, in prepare_model_input
ERROR 04-24 18:56:09 [engine.py:448]     model_input = self._prepare_model_input_tensors(
ERROR 04-24 18:56:09 [engine.py:448]                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/worker/model_runner.py", line 1238, in _prepare_model_input_tensors
ERROR 04-24 18:56:09 [engine.py:448]     return self.builder.build()  # type: ignore
ERROR 04-24 18:56:09 [engine.py:448]            ^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/worker/model_runner.py", line 934, in build
ERROR 04-24 18:56:09 [engine.py:448]     attn_metadata = self.attn_metadata_builder.build(
ERROR 04-24 18:56:09 [engine.py:448]                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/attention/backends/utils.py", line 245, in build
ERROR 04-24 18:56:09 [engine.py:448]     block_tables = make_tensor_with_pad(
ERROR 04-24 18:56:09 [engine.py:448]                    ^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448]   File "/app/vllm-upstream/vllm/utils.py", line 927, in make_tensor_with_pad
ERROR 04-24 18:56:09 [engine.py:448]     tensor = torch.from_numpy(padded_x).to(device)
ERROR 04-24 18:56:09 [engine.py:448]              ^^^^^^^^^^^^^^^^^^^^^^^^^^
ERROR 04-24 18:56:09 [engine.py:448] TypeError: expected np.ndarray (got numpy.ndarray)
Process SpawnProcess-1:
Traceback (most recent call last):
  File "/usr/lib/python3.12/multiprocessing/process.py", line 314, in _bootstrap
    self.run()
  File "/usr/lib/python3.12/multiprocessing/process.py", line 108, in run
    self._target(*self._args, **self._kwargs)
  File "/app/vllm-upstream/vllm/engine/multiprocessing/engine.py", line 450, in run_mp_engine
    raise e
  File "/app/vllm-upstream/vllm/engine/multiprocessing/engine.py", line 436, in run_mp_engine
    engine = MQLLMEngine.from_vllm_config(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/engine/multiprocessing/engine.py", line 128, in from_vllm_config
    return cls(
           ^^^^
  File "/app/vllm-upstream/vllm/engine/multiprocessing/engine.py", line 82, in __init__
    self.engine = LLMEngine(*args, **kwargs)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/engine/llm_engine.py", line 284, in __init__
    self._initialize_kv_caches()
  File "/app/vllm-upstream/vllm/engine/llm_engine.py", line 428, in _initialize_kv_caches
    self.model_executor.determine_num_available_blocks())
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/executor/executor_base.py", line 103, in determine_num_available_blocks
    results = self.collective_rpc("determine_num_available_blocks")
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/executor/uniproc_executor.py", line 56, in collective_rpc
    answer = run_method(self.driver_worker, method, args, kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/utils.py", line 2446, in run_method
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/torch/utils/_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/worker/worker.py", line 249, in determine_num_available_blocks
    self.model_runner.profile_run()
  File "/usr/local/lib/python3.12/dist-packages/torch/utils/_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/worker/model_runner.py", line 1253, in profile_run
    self._dummy_run(max_num_batched_tokens, max_num_seqs)
  File "/app/vllm-upstream/vllm/worker/model_runner.py", line 1365, in _dummy_run
    model_input = self.prepare_model_input(
                  ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/worker/model_runner.py", line 1674, in prepare_model_input
    model_input = self._prepare_model_input_tensors(
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/worker/model_runner.py", line 1238, in _prepare_model_input_tensors
    return self.builder.build()  # type: ignore
           ^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/worker/model_runner.py", line 934, in build
    attn_metadata = self.attn_metadata_builder.build(
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/attention/backends/utils.py", line 245, in build
    block_tables = make_tensor_with_pad(
                   ^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/utils.py", line 927, in make_tensor_with_pad
    tensor = torch.from_numpy(padded_x).to(device)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: expected np.ndarray (got numpy.ndarray)
[rank0]:[W424 18:56:09.851832590 ProcessGroupNCCL.cpp:1476] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
Traceback (most recent call last):
  File "/usr/local/bin/vllm", line 8, in <module>
    sys.exit(main())
             ^^^^^^
  File "/app/vllm-upstream/vllm/entrypoints/cli/main.py", line 53, in main
    args.dispatch_function(args)
  File "/app/vllm-upstream/vllm/entrypoints/cli/serve.py", line 27, in cmd
    uvloop.run(run_server(args))
  File "/usr/local/lib/python3.12/dist-packages/uvloop/__init__.py", line 109, in run
    return __asyncio.run(
           ^^^^^^^^^^^^^^
  File "/usr/lib/python3.12/asyncio/runners.py", line 195, in run
    return runner.run(main)
           ^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.12/asyncio/runners.py", line 118, in run
    return self._loop.run_until_complete(task)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "uvloop/loop.pyx", line 1518, in uvloop.loop.Loop.run_until_complete
  File "/usr/local/lib/python3.12/dist-packages/uvloop/__init__.py", line 61, in wrapper
    return await main
           ^^^^^^^^^^
  File "/app/vllm-upstream/vllm/entrypoints/openai/api_server.py", line 1078, in run_server
    async with build_async_engine_client(args) as engine_client:
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.12/contextlib.py", line 210, in __aenter__
    return await anext(self.gen)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/entrypoints/openai/api_server.py", line 146, in build_async_engine_client
    async with build_async_engine_client_from_engine_args(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.12/contextlib.py", line 210, in __aenter__
    return await anext(self.gen)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/app/vllm-upstream/vllm/entrypoints/openai/api_server.py", line 269, in build_async_engine_client_from_engine_args
    raise RuntimeError(
RuntimeError: Engine process failed to start. See stack trace for the root cause.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment