这是我的配置

npu-smi info

--- 1. PyTorch & Torch-NPU Version ---

PyTorch: 2.6.0+cpu

Torch-NPU: 2.6.0.post5

--- 2. CANN Toolkit Version ---

package_name=Ascend-cann-toolkit

version=8.5.0

innerversion=V100R001C25SPC001B232

compatible_version=[V100R001C15],[V100R001C18],[V100R001C19],[V100R001C20],[V100R001C21],[V100R001C23]

arch=aarch64

os=linux

path=/usr/local/Ascend/cann-8.5.0

package_name=Ascend-cann-toolkit

version=8.5.0

innerversion=V100R001C25SPC001B232

compatible_version=[V100R001C15],[V100R001C18],[V100R001C19],[V100R001C20],[V100R001C21],[V100R001C23]

arch=aarch64

os=linux

path=/usr/local/Ascend/cann-8.5.0

通过vllm启动时报错
ERROR 02-04 02:03:59 [engine.py:458] arange_out_op_api:build/CMakeFiles/torch_npu.dir/compiler_depend.ts:36 NPU function error: call aclnnArange failed, error code is 561103

ERROR 02-04 02:03:59 [engine.py:458] [ERROR] 2026-02-04-02:03:59 (PID:1068, Device:0, RankID:-1) ERR00100 PTA call acl api failed.

ERROR 02-04 02:03:59 [engine.py:458] EZ9999: Inner Error!

ERROR 02-04 02:03:59 [engine.py:458] EZ9999: [PID: 1068] 2026-02-04-02:03:59.414.243 Parse dynamic kernel config fail.

ERROR 02-04 02:03:59 [engine.py:458]         TraceBack (most recent call last):

ERROR 02-04 02:03:59 [engine.py:458]        AclOpKernelInit failed opType

ERROR 02-04 02:03:59 [engine.py:458]        ArangeAiCore ADD_TO_LAUNCHER_LIST_AICORE failed.

ERROR 02-04 02:03:59 [engine.py:458] Traceback (most recent call last):

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/engine/multiprocessing/engine.py", line 446, in run_mp_engine

ERROR 02-04 02:03:59 [engine.py:458]     engine = MQLLMEngine.from_vllm_config(

ERROR 02-04 02:03:59 [engine.py:458]              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/engine/multiprocessing/engine.py", line 133, in from_vllm_config

ERROR 02-04 02:03:59 [engine.py:458]     return cls(

ERROR 02-04 02:03:59 [engine.py:458]            ^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/engine/multiprocessing/engine.py", line 87, in __init__

ERROR 02-04 02:03:59 [engine.py:458]     self.engine = LLMEngine(*args, **kwargs)

ERROR 02-04 02:03:59 [engine.py:458]                   ^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/engine/llm_engine.py", line 265, in __init__

ERROR 02-04 02:03:59 [engine.py:458]     self.model_executor = executor_class(vllm_config=vllm_config)

ERROR 02-04 02:03:59 [engine.py:458]                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/executor/executor_base.py", line 53, in __init__

ERROR 02-04 02:03:59 [engine.py:458]     self._init_executor()

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/executor/uniproc_executor.py", line 48, in _init_executor

ERROR 02-04 02:03:59 [engine.py:458]     self.collective_rpc("load_model")

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/executor/uniproc_executor.py", line 57, in collective_rpc

ERROR 02-04 02:03:59 [engine.py:458]     answer = run_method(self.driver_worker, method, args, kwargs)

ERROR 02-04 02:03:59 [engine.py:458]              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/utils.py", line 2671, in run_method

ERROR 02-04 02:03:59 [engine.py:458]     return func(*args, **kwargs)

ERROR 02-04 02:03:59 [engine.py:458]            ^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm-ascend/vllm_ascend/worker/worker.py", line 240, in load_model

ERROR 02-04 02:03:59 [engine.py:458]     self.model_runner.load_model()

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm-ascend/vllm_ascend/worker/model_runner.py", line 997, in load_model

ERROR 02-04 02:03:59 [engine.py:458]     self.model = get_model(vllm_config=self.vllm_config)

ERROR 02-04 02:03:59 [engine.py:458]                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/model_executor/model_loader/__init__.py", line 59, in get_model

ERROR 02-04 02:03:59 [engine.py:458]     return loader.load_model(vllm_config=vllm_config,

ERROR 02-04 02:03:59 [engine.py:458]            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/model_executor/model_loader/base_loader.py", line 38, in load_model

ERROR 02-04 02:03:59 [engine.py:458]     model = initialize_model(vllm_config=vllm_config,

ERROR 02-04 02:03:59 [engine.py:458]             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/model_executor/model_loader/utils.py", line 62, in initialize_model

ERROR 02-04 02:03:59 [engine.py:458]     return model_class(vllm_config=vllm_config, prefix=prefix)

ERROR 02-04 02:03:59 [engine.py:458]            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm-ascend/vllm_ascend/models/qwen3.py", line 422, in __init__

ERROR 02-04 02:03:59 [engine.py:458]     self.model = CustomQwen3Model(vllm_config=vllm_config,

ERROR 02-04 02:03:59 [engine.py:458]                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/compilation/decorators.py", line 152, in __init__

ERROR 02-04 02:03:59 [engine.py:458]     old_init(self, vllm_config=vllm_config, prefix=prefix, **kwargs)

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm-ascend/vllm_ascend/models/qwen3.py", line 341, in __init__

ERROR 02-04 02:03:59 [engine.py:458]     super().__init__(vllm_config=vllm_config,

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/compilation/decorators.py", line 152, in __init__

ERROR 02-04 02:03:59 [engine.py:458]     old_init(self, vllm_config=vllm_config, prefix=prefix, **kwargs)

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/model_executor/models/qwen2.py", line 316, in __init__

ERROR 02-04 02:03:59 [engine.py:458]     self.start_layer, self.end_layer, self.layers = make_layers(

ERROR 02-04 02:03:59 [engine.py:458]                                                     ^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/model_executor/models/utils.py", line 626, in make_layers

ERROR 02-04 02:03:59 [engine.py:458]     [PPMissingLayer() for _ in range(start_layer)] + [

ERROR 02-04 02:03:59 [engine.py:458]                                                      ^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/model_executor/models/utils.py", line 627, in <listcomp>

ERROR 02-04 02:03:59 [engine.py:458]     maybe_offload_to_cpu(layer_fn(prefix=f"{prefix}.{idx}"))

ERROR 02-04 02:03:59 [engine.py:458]                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/model_executor/models/qwen2.py", line 318, in <lambda>

ERROR 02-04 02:03:59 [engine.py:458]     lambda prefix: decoder_layer_type(config=config,

ERROR 02-04 02:03:59 [engine.py:458]                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm-ascend/vllm_ascend/models/qwen3.py", line 219, in __init__

ERROR 02-04 02:03:59 [engine.py:458]     self.self_attn = CustomQwen3Attention(

ERROR 02-04 02:03:59 [engine.py:458]                      ^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm-ascend/vllm_ascend/models/qwen3.py", line 117, in __init__

ERROR 02-04 02:03:59 [engine.py:458]     super().__init__(hidden_size=hidden_size,

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/model_executor/models/qwen3.py", line 110, in __init__

ERROR 02-04 02:03:59 [engine.py:458]     self.rotary_emb = get_rope(

ERROR 02-04 02:03:59 [engine.py:458]                       ^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/model_executor/layers/rotary_embedding.py", line 1756, in get_rope

ERROR 02-04 02:03:59 [engine.py:458]     rotary_emb = RotaryEmbedding(head_size, rotary_dim, max_position, base,

ERROR 02-04 02:03:59 [engine.py:458]                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/model_executor/layers/rotary_embedding.py", line 112, in __init__

ERROR 02-04 02:03:59 [engine.py:458]     cache = self._compute_cos_sin_cache()

ERROR 02-04 02:03:59 [engine.py:458]             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/model_executor/layers/rotary_embedding.py", line 129, in _compute_cos_sin_cache

ERROR 02-04 02:03:59 [engine.py:458]     inv_freq = self._compute_inv_freq(self.base)

ERROR 02-04 02:03:59 [engine.py:458]                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/vllm-workspace/vllm/vllm/model_executor/layers/rotary_embedding.py", line 123, in _compute_inv_freq

ERROR 02-04 02:03:59 [engine.py:458]     inv_freq = 1.0 / (base**(torch.arange(

ERROR 02-04 02:03:59 [engine.py:458]                              ^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458]   File "/usr/local/python3.11.13/lib/python3.11/site-packages/torch/utils/_device.py", line 106, in __torch_function__

ERROR 02-04 02:03:59 [engine.py:458]     return func(*args, **kwargs)

ERROR 02-04 02:03:59 [engine.py:458]            ^^^^^^^^^^^^^^^^^^^^^

ERROR 02-04 02:03:59 [engine.py:458] RuntimeError: arange_out_op_api:build/CMakeFiles/torch_npu.dir/compiler_depend.ts:36 NPU function error: call aclnnArange failed, error code is 561103