GPU: Replace OpenMP parallization with TBB - WIP #13997
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Contributor
|
REQUEST FOR PRODUCTION RELEASES: This will add The following labels are available |
davidrohr
force-pushed
the
dev_pull_request
branch
from
f6b40be to
be91a01
Compare
Collaborator
|
Error while checking build/O2/fullCI for be91a01 at 2025-02-21 21:57: Full log here. |
davidrohr
force-pushed
the
dev_pull_request
branch
from
be91a01 to
184803f
Compare
Collaborator
|
Error while checking build/O2/fullCI_slc9 for 184803f at 2025-02-21 23:50: Full log here. |
davidrohr
force-pushed
the
dev_pull_request
branch
from
184803f to
158afe1
Compare
Collaborator
|
Error while checking build/O2/fullCI for 184803f at 2025-02-22 00:20: Full log here. |
Collaborator
Collaborator
|
Error while checking build/O2/fullCI_slc9 for 158afe1 at 2025-02-22 13:21: Full log here. |
davidrohr
force-pushed
the
dev_pull_request
branch
from
158afe1 to
357e76e
Compare
Collaborator
|
Error while checking build/O2/fullCI_slc9 for 357e76e at 2025-02-22 16:25: Full log here. |
davidrohr
force-pushed
the
dev_pull_request
branch
from
357e76e to
b21fbd8
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Work in progress, don't merge yet!
For now, want to test in CI.
Background for this PR is that OpenMP does not have a proper scheduling approach for nested loop, or for loops with reduced parallelism. Instead of using a thread pool, it just keeps spawning and termiinating threads.
For me that is order of 10k thread spawns on an EPN node for 1 TF.
I could fix that by hacking GCC libgomp, but GCC is not interested in taking that patch, since their behavior is compliant with the OpenMP spec, which is IMHO broken.
So I decided to just do away with OpenMP.