Tf not able to work with CUDA

TensorFlow General Discussion
, ,
1

Hi Folks,

I have been trying to make tf/keras work with with CUDA. I believe the CUDA and the Nvidia drivers are installed correctly because

  1. My Julia can use the GPUs with absolutely no problem. It would be great if tf/keras would have such a wonderful configuration for GPU as Julia.

  2. When I run nvidia-smi I get

NVIDIA-SMI 470.103.01 Driver Version: 470.103.01 CUDA Version: 11.4
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |

I installed the python tools for cuda, including

cudatoolkit, cudann

but when I run python and ask to tell me about the GPUs I get
tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudnn.so.8

Could anyone in this forum please help?

3

These are the tested version of the CUDA/cudnn dependencies: