-
Notifications
You must be signed in to change notification settings - Fork 54
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
Add CUDA 12.2 in nightlies #483
Conversation
✅ Deploy Preview for docs-rapids-ai ready!Built without sensitive environment variables
To edit notification comments on pull requests, go to your Netlify site configuration. |
pip users should be able to use any CTK from the same major version, I believe. We should drop the ".0" in places like that, and just say CUDA 12. |
Co-authored-by: Bradley Dice <bdice@bradleydice.com>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Suggesting one link change, which I'll commit. Otherwise LGTM!
I think we're waiting on rapidsai/docker#634 to have the Docker images for 24.04 nightlies. I'll add a |
Now that we merged PR ( rapidsai/docker#634 ), is it alright to remove "do not merge" here? Or is there something else still needed? |
Docker images are live, so I think this is ready to go. https://hub.docker.com/layers/rapidsai/base/24.04a-cuda12.2-py3.10/images/sha256-19a47f80236b6367f4f36d40c6d7f8aa35152b1537abec48582e51ad9c0a8607?context=explore I'll do a final review pass. I feel pretty comfortable with merging this if everything looks good after I review. |
I verified that Docker images for CUDA 12.2 (base & notebooks) are usable, and the conda installation command for RAPIDS 24.04 nightlies with CUDA 12.2 is usable. ("Usable" is defined as the ability to execute Merging. |
Thank you both! 🙏 |
Contributes to #482.
This is a small PR to update the selector for CUDA 12.2 nightlys on Conda, and when 24.04 rolls around it will be easy to remove the nightly restriction on 12.2.
I also made a couple unrelated changes for pip, removing the need to spell out every package we host and just saying that all RAPIDS libraries are hosted by NVIDIA. Hopefully less noise in the notes makes it easier to read the Python version support in pip.
Open question around CUDA 12.2 and pip:
In the install notes for WSL2 we tell users to install the 12.0 CTK; is it necessary that our pip users have the exact same CTK? My recollection is that any 12+ will work with our pip packages, but wanted to check before doing so.