I created a virtualenv and installed TensorFlow 2.4.0. The CUDA was already installed so I tried to install Cudnn according to these instructions
$ tar -xzvf cudnn-11.0-linux-x64-v8.0.2.39.tgz
$ sudo cp cuda/include/cudnn*.h /usr/local/cuda/include
$ sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64
$ sudo chmod a r /usr/local/cuda/include/cudnn*.h /usr/local/cuda/lib64/libcudnn*
Indeed after the installation files were installed to these paths /usr/local/cuda/include and /usr/local/cuda/lib64 .This was the result ( a fracture of the list)
Although the file 'libcudnn.so.8' is present ,it doesnt recognize it
Should I copy the files to another folder due to the created virtual environment? What does light green file indicate??
CodePudding user response:
I just simply added this in .bashrc
export PATH=/usr/local/cuda-11.0/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-11.0/lib64:$LD_LIBRARY_PATH
according to this Installing the NVIDIA driver, CUDA and cuDNN on Linux (Ubuntu 20.04)