[Bug] Fix DeepSeek-V2.5-1210-FP8 issue

[yewentao256]

Purpose

Fixes #27254

Test

Origin:

vllm serve RedHatAI/DeepSeek-V2.5-1210-FP8 --enable-expert-parallel -tp 4 --enforce_eager
......

(EngineCore_DP0 pid=89543) Traceback (most recent call last):
(EngineCore_DP0 pid=89543)   File "/usr/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap
(EngineCore_DP0 pid=89543)     self.run()
(EngineCore_DP0 pid=89543)   File "/usr/lib/python3.10/multiprocessing/process.py", line 108, in run
(EngineCore_DP0 pid=89543)     self._target(*self._args, **self._kwargs)
(EngineCore_DP0 pid=89543)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 712, in run_engine_core
(EngineCore_DP0 pid=89543)     raise e
(EngineCore_DP0 pid=89543)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 699, in run_engine_core
(EngineCore_DP0 pid=89543)     engine_core = EngineCoreProc(*args, **kwargs)
(EngineCore_DP0 pid=89543)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 498, in __init__
(EngineCore_DP0 pid=89543)     super().__init__(vllm_config, executor_class, log_stats,
(EngineCore_DP0 pid=89543)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 92, in __init__
(EngineCore_DP0 pid=89543)     self._initialize_kv_caches(vllm_config)
(EngineCore_DP0 pid=89543)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/engine/core.py", line 190, in _initialize_kv_caches
(EngineCore_DP0 pid=89543)     self.model_executor.determine_available_memory())
(EngineCore_DP0 pid=89543)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/executor/abstract.py", line 85, in determine_available_memory
(EngineCore_DP0 pid=89543)     return self.collective_rpc("determine_available_memory")
(EngineCore_DP0 pid=89543)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/executor/multiproc_executor.py", line 264, in collective_rpc
(EngineCore_DP0 pid=89543)     result = get_response(w, dequeue_timeout,
(EngineCore_DP0 pid=89543)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/executor/multiproc_executor.py", line 248, in get_response
(EngineCore_DP0 pid=89543)     raise RuntimeError(
(EngineCore_DP0 pid=89543) RuntimeError: Worker failed with error '', please check the stack trace above for the root cause
(APIServer pid=88560) Traceback (most recent call last):
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/bin/vllm", line 10, in <module>
(APIServer pid=88560)     sys.exit(main())
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/entrypoints/cli/main.py", line 54, in main
(APIServer pid=88560)     args.dispatch_function(args)
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/entrypoints/cli/serve.py", line 57, in cmd
(APIServer pid=88560)     uvloop.run(run_server(args))
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/uvloop/__init__.py", line 69, in run
(APIServer pid=88560)     return loop.run_until_complete(wrapper())
(APIServer pid=88560)   File "uvloop/loop.pyx", line 1518, in uvloop.loop.Loop.run_until_complete
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/uvloop/__init__.py", line 48, in wrapper
(APIServer pid=88560)     return await main
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/entrypoints/openai/api_server.py", line 1884, in run_server
(APIServer pid=88560)     await run_server_worker(listen_address, sock, args, **uvicorn_kwargs)
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/entrypoints/openai/api_server.py", line 1902, in run_server_worker
(APIServer pid=88560)     async with build_async_engine_client(
(APIServer pid=88560)   File "/usr/lib/python3.10/contextlib.py", line 199, in __aenter__
(APIServer pid=88560)     return await anext(self.gen)
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/entrypoints/openai/api_server.py", line 180, in build_async_engine_client
(APIServer pid=88560)     async with build_async_engine_client_from_engine_args(
(APIServer pid=88560)   File "/usr/lib/python3.10/contextlib.py", line 199, in __aenter__
(APIServer pid=88560)     return await anext(self.gen)
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/entrypoints/openai/api_server.py", line 225, in build_async_engine_client_from_engine_args
(APIServer pid=88560)     async_llm = AsyncLLM.from_vllm_config(
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/utils/__init__.py", line 1572, in inner
(APIServer pid=88560)     return fn(*args, **kwargs)
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/engine/async_llm.py", line 207, in from_vllm_config
(APIServer pid=88560)     return cls(
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/engine/async_llm.py", line 134, in __init__
(APIServer pid=88560)     self.engine_core = EngineCoreClient.make_async_mp_client(
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/engine/core_client.py", line 102, in make_async_mp_client
(APIServer pid=88560)     return AsyncMPClient(*client_args)
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/engine/core_client.py", line 769, in __init__
(APIServer pid=88560)     super().__init__(
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/engine/core_client.py", line 448, in __init__
(APIServer pid=88560)     with launch_core_engines(vllm_config, executor_class,
(APIServer pid=88560)   File "/usr/lib/python3.10/contextlib.py", line 142, in __exit__
(APIServer pid=88560)     next(self.gen)
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/engine/utils.py", line 732, in launch_core_engines
(APIServer pid=88560)     wait_for_engine_startup(
(APIServer pid=88560)   File "/proj-tango-pvc/users/zhipeng.wang/workspace/vllm/.venv/lib/python3.10/site-packages/vllm/v1/engine/utils.py", line 785, in wait_for_engine_startup
(APIServer pid=88560)     raise RuntimeError("Engine core initialization failed. "
(APIServer pid=88560) RuntimeError: Engine core initialization failed. See root cause above. Failed core proc(s): {}

Now:

(APIServer pid=655331) INFO:     Started server process [655331]
(APIServer pid=655331) INFO:     Waiting for application startup.
(APIServer pid=655331) INFO:     Application startup complete.

[gemini-code-assist]

Code Review

This pull request addresses an issue with the DeepSeek-V2.5-1210-FP8 model in vllm, specifically related to the configuration of quantization parameters in the cutlass_moe_fp8 function. The changes involve adding a check to ensure that the dimensions of w1_scale are compatible with w1_q when per-out-channel quantization is enabled, and adding another check when it is disabled.