torch not recognizing CUDA is installed

I have a laptop with a cuda-enabled NVIDIA GPU. I can successfully build and run the deviceQuery sample project as described here and here I think this means that the GPU and CUDA are working as expected. However when I install and load the torch package (after restarting my computer) it does not recognize that CUDA is available.

I also installed pytorch and checked that cuda is available.

What am I missing?

Unfortunately we don't support cuda 10.0.
Updating to CUDA 10.1 or 10.2 should fix this issue.

1 Like

Apparently I had multiple CUDA versions installed and pointing the CUDA_PATH environment variable to version 10.2 did the trick. Thanks!

1 Like

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.