-
Notifications
You must be signed in to change notification settings - Fork 407
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
Potential data race in Cuda parallel reduce #6236
Potential data race in Cuda parallel reduce #6236
Conversation
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.
We probably need the same fix for HIP
Have you measured if you see any impact in performance for, say https://github.com/kokkos/kokkos-tutorials/tree/main/Exercises/04/Solution? |
Yes, the differences were negligible (<2% relative increase/decrease between this branch and develop commit it is based on, various problem sizes, launching p_for/reduce kernels). |
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.
Looks good to me 👍
We probably need the same fix for HIP
I propose we merge this as is and fix HIP with a separate issue / PR.
Retest this please |
The previous Jenkins failure was long enough ago that the logs have been discarded |
|
Retest this please. |
CUDA builds all passed. None of the pending Jenkins builds is a CUDA backend, so this should be safe to merge. |
See comment #6217 (comment). Suggested in #4855 (comment) as being a potential issue, I think this is an actual issue. At the very least we avoid doing computation that is unnecessary, and it clears a warning when running
compute-sanitizer --tool=racecheck
(see #6217).