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

Documentation fixes #1585

Merged
merged 3 commits into from Jul 13, 2023
Merged

Documentation fixes #1585

merged 3 commits into from Jul 13, 2023

Conversation

joaander
Copy link
Member

Description

  • Recommend installing CUDA 12.0 builds from conda-forge.
  • Fix typo in net_virial documentation.

Motivation and context

  • CUDA 12 builds are now available on conda-forge thanks to the efforts of @bdice. The installation documentation should recommend installing the latest version available.
  • The net_virial documentation incorrectly identified the shape of the array.

Change log

Fixed:

* Improved documentation.

Checklist:

@joaander joaander requested review from a team as code owners July 13, 2023 12:14
@joaander joaander requested review from tommy-waltmann and vyasr and removed request for a team July 13, 2023 12:14
@joaander joaander added the validate Execute long running validation tests on pull requests label Jul 13, 2023
@joaander joaander requested review from b-butler and removed request for tommy-waltmann July 13, 2023 12:22
Copy link
Member

@bdice bdice left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good.

Also, users can control which CUDA Toolkit version they get by installing
hoomd=4.0.1=*gpu* cuda-version=12.*
or
hoomd=4.0.1=*gpu* cuda-version=11.*
or more specifically cuda-version=11.5, etc. (Not tested but this should work, let me know if not.) The CONDA_OVERRIDE_CUDA environment variable will override the __cuda virtual package constraint. You could satisfy cudatoolkit cuda-version=11.8 with a CUDA 12 driver that shows in __cuda. In other words, it’s fine to install CUDA 11 environments on CUDA 12 drivers if needed.

@joaander
Copy link
Member Author

@bdice Yes, requesting "hoomd=4.0.1=*gpu*" "cuda-version=12.*" and "hoomd=4.0.1=*gpu*" "cuda-version=12.*" works as you describe. CONDA_OVERRIDE_CUDA is necessary for many HOOMD-blue users that install on cluster login nodes that lack GPUs with the intention to use it on GPU compute nodes.

You are correct to note the forward compatibility with newer drivers. I think that distinction is too much information to include in installation documentation aimed at first time users.

@bdice
Copy link
Member

bdice commented Jul 13, 2023

Awesome. I agree that might be too much for the HOOMD docs, just wanted to be sure you were aware of that option. Thanks again!

@joaander joaander merged commit 5a1ac56 into trunk-patch Jul 13, 2023
40 checks passed
@joaander joaander deleted the misc-documentation-fixes branch July 13, 2023 16:12
@vyasr
Copy link
Contributor

vyasr commented Jul 13, 2023

This got merged real quick so I didn't review, but it looks good! Excited to see uptake of our new conda-forge packages.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
validate Execute long running validation tests on pull requests
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants