0

I was trying to install pytorch with cuda on my system. My gpu is a little old. Its NVIDIA GeForce GT 720m and my driver version is 391.35.

When I installed cudatoolkit in anaconda, it installed CUDA 11.3. Then, I went to https://pytorch.org/get-started/previous-versions/

and installed the version that suits cuda 11.3 and it was installed correctly. Now when I check cuda version by print(torch.version.cuda), it says cuda 11.3. However, when I check if cuda is available, it returns, False. I even downloaded cuda 11.3 from https://developer.nvidia.com/cuda-11.3.0-download-archive

and installed it on my system without any error. Still, I get False for availability of cuda. Can some one please help me and tell me what is my problem and how to resolve it?

P.S. I checked similar question here at https://stackoverflow.com/questions/60987997/why-torch-cuda-is-available-returns-false-even-after-installing-pytorch-with

but honestly, I did not get it because I am new to programming.

Ali.A
  • 159
  • 8

1 Answers1

1

If you search for your GPU's specs online you can find the version of CUDA you need.

At https://www.techpowerup.com/gpu-specs/geforce-gt-720m.c2297 it says that your processor needs CUDA version 2.1 whereas you're trying to install version 11.3. After that you need to check if your Graphics driver is compatible and your version of PyTorch.

The top answer in the stack overflow you linked explains it a lot better.

iancullinane06
  • 36
  • 2