Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Which Cuda versions are not supported by Cog? #1330

Closed
geoxpert0001 opened this issue Oct 5, 2023 · 2 comments
Closed

Which Cuda versions are not supported by Cog? #1330

geoxpert0001 opened this issue Oct 5, 2023 · 2 comments

Comments

@geoxpert0001
Copy link

Does Cog support all CUDA versions or not?
I want to know if older CUDA versions are compatible and functional.

@catsby
Copy link

catsby commented Oct 5, 2023

Hi 👋 there's a generated file pkg/config/cuda_base_images.json which appears to be a list of CUDA base images and versions included. I can't say if it's definitive or up to date, but reading it implies to me the following versions are supported:

12.1.1
12.0.1
12.0.0
11.8.0
11.7.1
11.6.2
11.5.2
11.4.3
11.3.1
11.2.2
11.1.1
11.0.3

I could be wrong though... it looks like this file is used for validating cog.yaml files(?). Hopefully this helps, but I realize it's not definitive.

@mattt
Copy link
Contributor

mattt commented Oct 5, 2023

Thanks, @catsby! That's exactly right. Cog uses this generated list of CUDA base images to resolve PyTorch and TensorFlow packages to compatible versions.

For some historical context, Nvidia recently removed some older CUDA image tags from Docker Hub, which resulted in Cog resolving to images that no longer existed (#1162). So if the list of supported CUDA images seems sparse, that may be why.

@mattt mattt closed this as completed Oct 5, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants