Skip to content
This repository has been archived by the owner on Oct 31, 2023. It is now read-only.

Incompatibility with CUDA 11.0 #42

Closed
purvaten opened this issue Sep 15, 2021 · 2 comments
Closed

Incompatibility with CUDA 11.0 #42

purvaten opened this issue Sep 15, 2021 · 2 comments

Comments

@purvaten
Copy link

purvaten commented Sep 15, 2021

Hi, I am running into issues while setting this codebase for CUDA 11.0 since it is not compatible with pytorch 1.4.

Specifically I am installing pytorch as follows: conda install pytorch==1.7.0 torchvision==0.8.0 torchaudio==0.7.0 cudatoolkit=11.0 -c pytorch.

But then I run into issues while installing neural_renderer, specifically RuntimeError: Error compiling objects for extension . I checked several answers online and it seems that this error appears when using pytorch >= 1.5. I unfortunately cannot use that because of my CUDA driver restrictions.

Is there a solution to make it compatible for CUDA 11.0?

@purvaten
Copy link
Author

This seems to be working with cudatoolkit 10.0, i.e., conda install pytorch=1.4.0 torchvision=0.5.0 cudatoolkit=10.0 -c pytorch as mentioned in the README.

But please update here in case there is a way to get this to work for a higher version of pytorch. Thanks!

@jasonyzhang
Copy link
Contributor

Thanks, I've updated the README. If you wish to use a higher version of Pytorch, you can update all the AT_CHECKs to TORCH_CHECK in NMR. See #3.

Best,
JAson

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants