分别使用Mindie2.0.T18和Mindie2.2.rc1镜像部署模型都会出现如下错误：
File "/usr/local/Ascend/atb-models/atb_llm/models/base/flash_causal_lm.py", line 117, in __init__
self.placeholder = torch.zeros(1, dtype=self.dtype, device="npu")
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: zero_:build/CMakeFiles/torch_npu.dir/compiler_depend.ts:26 NPU function error: call aclnnInplaceZero failed, error code is 561103
[ERROR] 2025-12-01-19:38:21 (PID:261, Device:0, RankID:-1) ERR00100 PTA call acl api failed.
EZ9999: Inner Error!
EZ9999[PID: 261] 2025-12-01-19:38:21.338.486 AclNN_Inner_Error(EZ9999): Parse dynamic kernel config fail.
TraceBack (most recent call last):
Failed to ParseDynamicKernels.
AclOpKernelInit failed opType
ZerosLike ADD_TO_LAUNCHER_LIST_AICORE failed.
环境如下：
cpu:鲲鹏920 64核*2
计算卡:910ProB*2
操作系统：kylinV10 server
点赞
收藏
回复
分享
举报
已解决
分别使用Mindie2.0.T18和Mindie2.2.rc1镜像部署模型都会出现如下错误：
File "/usr/local/Ascend/atb-models/atb_llm/models/base/flash_causal_lm.py", line 117, in __init__
self.placeholder = torch.zeros(1, dtype=self.dtype, device="npu")
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: zero_:build/CMakeFiles/torch_npu.dir/compiler_depend.ts:26 NPU function error: call aclnnInplaceZero failed, error code is 561103
[ERROR] 2025-12-01-19:38:21 (PID:261, Device:0, RankID:-1) ERR00100 PTA call acl api failed.
EZ9999: Inner Error!
EZ9999[PID: 261] 2025-12-01-19:38:21.338.486 AclNN_Inner_Error(EZ9999): Parse dynamic kernel config fail.
TraceBack (most recent call last):
Failed to ParseDynamicKernels.
AclOpKernelInit failed opType
ZerosLike ADD_TO_LAUNCHER_LIST_AICORE failed.
环境如下：
cpu:鲲鹏920 64核*2
计算卡:910ProB*2
操作系统：kylinV10 server