运行环境:
cuda 10.2
cudnn 8.5
paddlepaddle-gpu 2.3.2
paddleocr CPU 的时候没事,GPU报错如下,
报错信息:
--------------------------------------
C++ Traceback (most recent call last):
--------------------------------------
No stack trace in paddle, may be caused by external reasons.----------------------
Error Message Summary:
----------------------
FatalError: `Segmentation fault` is detected by the operating system.
[TimeInfo: *** Aborted at 1686904070 (unix time) try "date -d @1686904070" if you are using GNU date ***]
[SignalInfo: *** SIGSEGV (@0x0) received by PID 5060 (TID 0x7f2f44b89700) from PID 0 ***]段错误 (核心已转储)
原因:cudnn 版本太高了,和cuda不匹配;官网推荐cudnn7.6
安装CUDNN
官网下载 cudnn:https://developer.nvidia.com/rdp/cudnn-archive
这三个都要下载:
libcudnn7-dev_7.6.5.32-1+cuda10.2_amd64.deb、libcudnn7_7.6.5.32-1+cuda10.2_amd64.deb、libcudnn7-doc_7.6.3.30-1+cuda10.1_amd64.deb
安装CUDNN
$ sudo dpkg -i libcudnn7_7.6.5.32-1+cuda10.1_amd64.deb
$ sudo dpkg -i libcudnn7-dev_7.6.5.32-1+cuda10.1_amd64.deb
$ sudo dpkg -i libcudnn7-doc_7.6.5.32-1+cuda10.1_amd64.deb
测试安装是否成功
$ cp -r /usr/src/cudnn_samples_v7/ $HOME #请注意,这个cudnn_samples_v7一定要根据你自己的来,你直接去/usr/src中查看即可
$ cd $HOME/cudnn_samples_v7/mnistCUDNN
$ make clean && make
$ ./mnistCUDNN
可以啦!