Error about DNN

RuntimeError: Unimplemented: DNN library is not found.

cuDNN is not properly installed. Download cuDNN from NVIDIA website.

Exojax uses cuDNN when using jnp.convolve.

Case 1

Check the beggining of the error message. If you have

2021-08-15 12:06:46.013903: W external/org_tensorflow/tensorflow/stream_executor/platform/default/dso_loader.cc:64] 
Could not load dynamic library 'libcudnn.so.7'; 
dlerror: libcudnn.so.7: cannot open shared object file: No such file or directory;

You should check libcudnn.so.7 not libcudnn.so.8.

Case 2

Check the beggining of the error message. If you have

2021-08-15 12:24:35.915836: E external/org_tensorflow/tensorflow/stream_executor/cuda/cuda_dnn.cc:358] 
Loaded runtime CuDNN library: 7.5.0 but source was compiled with: 7.6.5. 

Download for instance cudnn-10.1-linux-x64-v7.6.5.32.tgz and perform

tar xvfz cudnn-10.1-linux-x64-v7.6.5.32.tgz
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/targets/x86_64-linux/lib/
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*

This might solves the problem.