使用官方minde镜像推理Qwen2 Instruct模型报错
收藏回复举报
使用官方minde镜像推理Qwen2 Instruct模型报错
t('forum.solved') 已解决
发表于2024-12-20 08:40:53
0 查看
 我的服务器是:KunLun G5680,客服说对应的是800-9000,对应的卡是910B 32GB版本 
 

我用的驱动和cann版本如下: 

```shell 

Ascend-cann-kernels-910_8.0.RC3_linux-aarch64.run  

Ascend-cann-nnal_8.0.RC3_linux-aarch64.run  

Ascend-hdk-910-npu-driver_24.1.rc2_linux-aarch64.run  

Ascend-hdk-910-npu-firmware_7.3.0.1.231.run 

 

以下是报错日志: 

```shell

2024-12-20 08:24:07.458865] [error] [2162107] [rt_backend_module.cpp:162] Rt Register AllKernel fail, error:507028  

[2024-12-20 08:24:07.462149] [error] [2162107] [bin_tactic.cpp:176] Ast RtRegister AllFunction fail, error:507028  

[2024-12-20 08:24:07.463452] [error] [2162107] [rt_backend_module.cpp:162] Rt Register AllKernel fail, error:507028  

[2024-12-20 08:24:07.465983] [error] [2162107] [bin_tactic.cpp:176] Ast RtRegister AllFunction fail, error:507028  

[2024-12-20 08:24:07.467140] [error] [2162107] [rt_backend_module.cpp:150] RtFunction Register fail, error:507000  

[2024-12-20 08:24:07.467250] [error] [2162107] [bin_tactic.cpp:165] Asd RtModuleGetFunction fail, errCode:507000, errName:unknownerrDesc:unknown  

[2024-12-20 08:24:07.467345] [error] [2162107] [rt_backend_module.cpp:150] RtFunction Register fail, error:507000  

 

[2024-12-20 08:24:07.467439] [error] [2162107] [bin_tactic.cpp:165] Asd RtModuleGetFunction fail, errCode:507000, errName:unknownerrDesc:unknown  

[2024-12-20 08:24:07.513896] [error] [2162107] [auto_tiling_rt2.cc:194] [ParseAutoTilingRun] OpName:[TransData] "Fail to get core num from GE, core num get is 0."  

[2024-12-20 08:24:07.513934] [error] [2162107] [trans_data.cc:1018] [TilingPrepare4Transdata] OpName:[TransData] "compile_info->other_compile_info is nullptr!"  

[2024-12-20 08:24:07.514076] [error] [2162107] [tbe_tiling_runner.cpp:156] failed to run tiling parse  

[2024-12-20 08:24:07.514087] [error] [2162107] [tbe_tiling_runner.cpp:613] failed to run tiling runner  

[2024-12-20 08:24:07.514135] [error] [2162107] [tactic_base.cpp:54] failed to init run info code:1, msg:  

[2024-12-20 08:24:07.514149] [error] [2162107] [asdops_node_implement.cpp:171] TransdataOpsRunner_0:0  node[0] InitRunInfo failed!  

[2024-12-20 08:24:07.514155] [error] [2162107] [ops_runner.cpp:293] TransdataOpsRunner_0:0  node[0] InitRunInfo failed!  

[2024-12-20 08:24:07.514168] [error] [2162107] [ops_runner.cpp:269] TransdataOpsRunner_0:0  node[0] fill tiling buffer fail, error code:3  

[2024-12-20 08:24:07.514175] [error] [2162107] [operation_base.cpp:871] TransdataOperation_0 fill host tiling buffer fail  

[2024-12-20 08:24:07.514183] [error] [2162107] [kernel_info.cpp:256] kernelInfo is not inited, get scratch size fail  

[2024-12-20 08:24:07.514186] [error] [2162107] [ops_runner.cpp:321] TransdataOpsRunner_0:0  node[0] get workspace size failed  

[2024-12-20 08:24:07.514306] [error] [2164386] [bin_tactic.cpp:53] args size invalid  

[2024-12-20 08:24:07.514333] [error] [2164386] [asdops_node_implement.cpp:201] TransdataOpsRunner_0:0  node[0] TransdataNdToNzTactic run fail, launchParam  

:opdesc:opdesc.opName:TransdataOperation opdesc.specificParam: {"outCrops":[0,0],"transdataType":2}  

intensors[0]: dtype:float16, format:nd, dims:[1, 128, 128], dataSize:32768, data:1  

outtensors[0]: dtype:float16, format:fractal_nz, dims:[1, 8, 128, 16], dataSize:32768, data:1  

[2024-12-20 08:24:07.514385] [error] [2164386] [ops_runner.cpp:596] RunKernel failed! ret:3  

[2024-12-20 08:24:07.514390] [error] [2164386] [ops_runner.cpp:393] RunAllKernel failed! ret:3  

[2024-12-20 08:24:07.514396] [error] [2164386] [runner.cpp:115] TransdataOpsRunner_0:1 Execute Failed. st: 3  

[2024-12-20 08:24:07.514405] [error] [2164386] [operation_base.cpp:714] TransdataOperation_0 execute TransdataOpsRunner fail  

[2024-12-20 08:24:07.514410] [error] [2164386] [operation_base.cpp:752] TransdataOperation_0 execute fail, error code: 3  

[2024-12-20 08:24:07.518682] [error] [2162107] [auto_tiling_rt2.cc:194] [ParseAutoTilingRun] OpName:[DefaultImpl] "Fail to get core num from GE, core num get is 0."  

[2024-12-20 08:24:07.518976] [error] [2162107] [tbe_tiling_runner.cpp:156] failed to run tiling parse  

[2024-12-20 08:24:07.518989] [error] [2162107] [tbe_tiling_runner.cpp:613] failed to run tiling runner  

[2024-12-20 08:24:07.519030] [error] [2162107] [tactic_base.cpp:54] failed to init run info code:1, msg:  

[2024-12-20 08:24:07.519041] [error] [2162107] [asdops_node_implement.cpp:171] GatherOpsRunner_1_0:0  node[0] InitRunInfo failed!  

[2024-12-20 08:24:07.519045] [error] [2162107] [ops_runner.cpp:293] GatherOpsRunner_1_0:0  node[0] InitRunInfo failed!  

[2024-12-20 08:24:07.519068] [error] [2162107] [ops_runner.cpp:269] GatherOpsRunner_1_0:0  node[0] fill tiling buffer fail, error code:3  

[2024-12-20 08:24:07.519077] [error] [2162107] [graph_runner.cpp:324] WordEmbeddingRunner_1:0 GraphRunner::FillHostTilingBufferImpl failed! error code:3  

[2024-12-20 08:24:07.519083] [error] [2162107] [operation_base.cpp:871] WordEmbedding_1 fill host tiling buffer fail  

```

我要发帖子