实在不知道该如何处理了

部分报错日志：

deepseek-server  | [ERROR] the socversion Ascend910B1 of bin package does not match the current device socverison Ascend310P3. Please modify default socversion in run.sh or execute run.sh with socversion parameter.

deepseek-server  | [ERROR] the socversion Ascend910B1 of bin package does not match the current device socverison Ascend310P3. Please modify default socversion in run.sh or execute run.sh with socversion parameter.

deepseek-server  | [ERROR] the socversion Ascend910B1 of bin package does not match the current device socverison Ascend310P3. Please modify default socversion in run.sh or execute run.sh with socversion parameter.

deepseek-server  | ..[1;36m(EngineCore_DP0 pid=283)[0;0m INFO 04-01 03:28:41 [worker_v1.py:234] Available memory: 36437120000, total memory: 46429265920

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m INFO 04-01 03:28:41 [kv_cache_utils.py:1087] GPU KV cache size: 1,270,784 tokens

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m INFO 04-01 03:28:41 [kv_cache_utils.py:1091] Maximum concurrency for 4,096 tokens per request: 310.25x

deepseek-server  | mki_log mkdir /root/ascend/log/atb

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708] EngineCore failed to start.

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708] Traceback (most recent call last):

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]   File "/vllm-workspace/vllm/vllm/v1/engine/core.py", line 699, in run_engine_core

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]     engine_core = EngineCoreProc(*args, **kwargs)

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]   File "/vllm-workspace/vllm/vllm/v1/engine/core.py", line 498, in __init__

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]     super().__init__(vllm_config, executor_class, log_stats,

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]   File "/vllm-workspace/vllm/vllm/v1/engine/core.py", line 92, in __init__

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]     self._initialize_kv_caches(vllm_config)

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]   File "/vllm-workspace/vllm/vllm/v1/engine/core.py", line 207, in _initialize_kv_caches

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]     self.model_executor.initialize_from_config(kv_cache_configs)

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]   File "/vllm-workspace/vllm/vllm/v1/executor/abstract.py", line 75, in initialize_from_config

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]     self.collective_rpc("compile_or_warm_up_model")

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]   File "/vllm-workspace/vllm/vllm/executor/uniproc_executor.py", line 83, in collective_rpc

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]     return [run_method(self.driver_worker, method, args, kwargs)]

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]   File "/vllm-workspace/vllm/vllm/utils/__init__.py", line 3122, in run_method

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]     return func(*args, **kwargs)

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]   File "/vllm-workspace/vllm-ascend/vllm_ascend/worker/worker_v1.py", line 310, in compile_or_warm_up_model

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]     self._warm_up_atb()

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]   File "/vllm-workspace/vllm-ascend/vllm_ascend/worker/worker_v1.py", line 319, in _warm_up_atb

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]     torch_npu._npu_matmul_add_fp32(x, weight, c)

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]   File "/usr/local/python3.10.17/lib/python3.10/site-packages/torch/_ops.py", line 1158, in __call__

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708]     return self._op(*args, **(kwargs or {}))

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m ERROR 04-01 03:28:42 [core.py:708] RuntimeError: LinearOperation CreateOperation failed!

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m Process EngineCore_DP0:

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m Traceback (most recent call last):

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/usr/local/python3.10.17/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     self.run()

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/usr/local/python3.10.17/lib/python3.10/multiprocessing/process.py", line 108, in run

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     self._target(*self._args, **self._kwargs)

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/vllm-workspace/vllm/vllm/v1/engine/core.py", line 712, in run_engine_core

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     raise e

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/vllm-workspace/vllm/vllm/v1/engine/core.py", line 699, in run_engine_core

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     engine_core = EngineCoreProc(*args, **kwargs)

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/vllm-workspace/vllm/vllm/v1/engine/core.py", line 498, in __init__

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     super().__init__(vllm_config, executor_class, log_stats,

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/vllm-workspace/vllm/vllm/v1/engine/core.py", line 92, in __init__

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     self._initialize_kv_caches(vllm_config)

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/vllm-workspace/vllm/vllm/v1/engine/core.py", line 207, in _initialize_kv_caches

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     self.model_executor.initialize_from_config(kv_cache_configs)

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/vllm-workspace/vllm/vllm/v1/executor/abstract.py", line 75, in initialize_from_config

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     self.collective_rpc("compile_or_warm_up_model")

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/vllm-workspace/vllm/vllm/executor/uniproc_executor.py", line 83, in collective_rpc

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     return [run_method(self.driver_worker, method, args, kwargs)]

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/vllm-workspace/vllm/vllm/utils/__init__.py", line 3122, in run_method

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     return func(*args, **kwargs)

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/vllm-workspace/vllm-ascend/vllm_ascend/worker/worker_v1.py", line 310, in compile_or_warm_up_model

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     self._warm_up_atb()

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/vllm-workspace/vllm-ascend/vllm_ascend/worker/worker_v1.py", line 319, in _warm_up_atb

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     torch_npu._npu_matmul_add_fp32(x, weight, c)

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m   File "/usr/local/python3.10.17/lib/python3.10/site-packages/torch/_ops.py", line 1158, in __call__

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m     return self._op(*args, **(kwargs or {}))

deepseek-server  | [1;36m(EngineCore_DP0 pid=283)[0;0m RuntimeError: LinearOperation CreateOperation failed!

deepseek-server  | [1;36m(APIServer pid=1)[0;0m Traceback (most recent call last):

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/usr/local/python3.10.17/lib/python3.10/runpy.py", line 196, in _run_module_as_main

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     return _run_code(code, main_globals, None,

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/usr/local/python3.10.17/lib/python3.10/runpy.py", line 86, in _run_code

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     exec(code, run_globals)

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/entrypoints/openai/api_server.py", line 1953, in <module>

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     uvloop.run(run_server(args))

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/usr/local/python3.10.17/lib/python3.10/site-packages/uvloop/__init__.py", line 69, in run

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     return loop.run_until_complete(wrapper())

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "uvloop/loop.pyx", line 1518, in uvloop.loop.Loop.run_until_complete

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/usr/local/python3.10.17/lib/python3.10/site-packages/uvloop/__init__.py", line 48, in wrapper

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     return await main

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/entrypoints/openai/api_server.py", line 1884, in run_server

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     await run_server_worker(listen_address, sock, args, **uvicorn_kwargs)

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/entrypoints/openai/api_server.py", line 1902, in run_server_worker

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     async with build_async_engine_client(

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/usr/local/python3.10.17/lib/python3.10/contextlib.py", line 199, in __aenter__

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     return await anext(self.gen)

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/entrypoints/openai/api_server.py", line 180, in build_async_engine_client

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     async with build_async_engine_client_from_engine_args(

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/usr/local/python3.10.17/lib/python3.10/contextlib.py", line 199, in __aenter__

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     return await anext(self.gen)

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/entrypoints/openai/api_server.py", line 225, in build_async_engine_client_from_engine_args

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     async_llm = AsyncLLM.from_vllm_config(

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/utils/__init__.py", line 1572, in inner

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     return fn(*args, **kwargs)

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/v1/engine/async_llm.py", line 207, in from_vllm_config

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     return cls(

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/v1/engine/async_llm.py", line 134, in __init__

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     self.engine_core = EngineCoreClient.make_async_mp_client(

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/v1/engine/core_client.py", line 102, in make_async_mp_client

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     return AsyncMPClient(*client_args)

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/v1/engine/core_client.py", line 769, in __init__

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     super().__init__(

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/v1/engine/core_client.py", line 448, in __init__

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     with launch_core_engines(vllm_config, executor_class,

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/usr/local/python3.10.17/lib/python3.10/contextlib.py", line 142, in __exit__

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     next(self.gen)

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/v1/engine/utils.py", line 732, in launch_core_engines

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     wait_for_engine_startup(

deepseek-server  | [1;36m(APIServer pid=1)[0;0m   File "/vllm-workspace/vllm/vllm/v1/engine/utils.py", line 785, in wait_for_engine_startup

deepseek-server  | [1;36m(APIServer pid=1)[0;0m     raise RuntimeError("Engine core initialization failed. "

deepseek-server  | [1;36m(APIServer pid=1)[0;0m RuntimeError: Engine core initialization failed. See root cause above. Failed core proc(s): {}

deepseek-server  | [1;36m(APIServer pid=1)[0;0m [ERROR] 2026-04-01-03:28:54 (PID:1, Device:-1, RankID:-1) ERR99999 UNKNOWN applicaiton exception