求助paddle預測時GPU運行出錯
w_蓦然回首_cn 发布于2020-10 浏览:1302 回复:2
0
收藏

具體內容如下:

W1011 09:45:38.090179 10321 device_context.cc:252] Please NOTE: device: 0, CUDA Capability: 75, Driver API Version: 11.0, Runtime API Version: 10.0
W1011 09:45:38.091177 10321 device_context.cc:260] device: 0, cuDNN Version: 8.0.
W1011 09:45:39.529865 10321 init.cc:226] Warning: PaddlePaddle catches a failure signal, it may not work properly
W1011 09:45:39.529884 10321 init.cc:228] You could check whether you killed PaddlePaddle thread/process accidentally or report the case to PaddlePaddle
W1011 09:45:39.529889 10321 init.cc:231] The detail failure signal is:

W1011 09:45:39.529893 10321 init.cc:234] *** Aborted at 1602380739 (unix time) try "date -d @1602380739" if you are using GNU date ***
W1011 09:45:39.530689 10321 init.cc:234] PC: @ 0x0 (unknown)
W1011 09:45:39.530755 10321 init.cc:234] *** SIGSEGV (@0x0) received by PID 10321 (TID 0x7f9ad6579740) from PID 0; stack trace: ***
W1011 09:45:39.531333 10321 init.cc:234] @ 0x7f9ad61788a0 (unknown)
W1011 09:45:39.531908 10321 init.cc:234] @ 0x0 (unknown)
Segmentation fault (core dumped)

收藏
点赞
0
个赞
共2条回复 最后由w_蓦然回首_cn回复于2020-10
#3w_蓦然回首_cn回复于2020-10

代碼如上,用CPU運行沒問題,GPU運行就出錯。另外,訓練的時候用GPU沒問題

0
#2w_蓦然回首_cn回复于2020-10

0
TOP
切换版块