TensorFlow:ImportError: libcudnn.so.*: cannot open shared object file: No such file or dictionary
阿新 • • 發佈:2019-01-27
環境:
- Ubuntu16.04
- Nvidia 384
- CUDA 8.0
- cuDNN 5
錯誤
TensorFlow、CUDA、cuDNN的版本關係小白我時常懵懵的,經常出現各種不支援的情況,比如今天安裝了TensorFlow1.4,報錯:
ImportError: libcudnn.so.6: cannot open shared object file: No such file or dictionary
錯誤緣由
google了一圈發現, 問題出在 TensorFlow 1.4-gpu 是基於cuDNN6,需要的也就是libcudnn.so.6了。
解決方案:
- Just download cuDNN v6 and follow the steps (Tested on Ubuntu 16.04, CUDA toolkit 8.0 )https://developer.nvidia.com/cudnn
$ tar xvzf cudnn-8.0-linux-x64-v5.1-ga.tgz
$ sudo cp -P cuda/include/cudnn.h /usr/local/cuda/include
$ sudo cp -P cuda/lib64/libcudnn* /usr/local/cuda/lib64
$ sudo chmod a+r /usr/local/cuda/include/cudnn.h /usr/local/cuda/lib64/libcudnn*
- Now set Path variables
$ vim ~/.bashrc
翻到最底部加上:
export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64"
export CUDA_HOME=/usr/local/cuda