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

Fix multi-GPU data movement #33

Merged
merged 3 commits into from
Dec 20, 2023
Merged

Fix multi-GPU data movement #33

merged 3 commits into from
Dec 20, 2023

Conversation

jpsamaroo
Copy link
Member

@jpsamaroo jpsamaroo commented Dec 20, 2023

As found by @Rabab53 when testing with multiple GPUs, we don't correctly switch devices when making copies of GPU arrays, particularly when moving a raw GPU array from CPU processors to a GPU processor (and back).

Todo:

  • Fixes for ROCm

@jpsamaroo jpsamaroo added bug Something isn't working cuda rocm metal labels Dec 20, 2023
@jpsamaroo jpsamaroo force-pushed the jps/fix-multigpu-move branch from 6cdbb30 to 10f3c85 Compare December 20, 2023 20:21
@jpsamaroo jpsamaroo removed the metal label Dec 20, 2023
@jpsamaroo jpsamaroo marked this pull request as ready for review December 20, 2023 20:22
@jpsamaroo jpsamaroo merged commit 8f9d530 into master Dec 20, 2023
@jpsamaroo jpsamaroo deleted the jps/fix-multigpu-move branch December 20, 2023 20:24
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working cuda rocm
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant