Explicitely create new stream for Barrier concurrent #453
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
hlinsen
added
non-breaking
hlinsen
changed the title
rg20
reviewed
Oct 7, 2025
| // Otherwise, CUDA API calls to the problem stream may occur in both threads and throw graph | ||
| // capture off | ||
| auto barrier_handle = raft::handle_t(*op_problem.get_handle_ptr()); | ||
| detail::problem_t<i_t, f_t> d_barrier_problem(problem); |
There was a problem hiding this comment.
Suggested change
| detail::problem_t<i_t, f_t> d_barrier_problem(problem); | |
| auto barrier_straem = rmm::cuda_stream_per_thread; | |
| raft::resource::set_cuda_stream(barrier_handle, barrier_stream); | |
| detail::problem_t<i_t, f_t> d_barrier_problem(problem, barrier_handle); |
rgsl888prabhu
approved these changes
Oct 7, 2025
rg20
approved these changes
Oct 7, 2025
|
/merge |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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Outdated suggestions cannot be applied.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.
In C++ when spawning a new thread the cuda driver by default creates a legacy default stream for each thread with unique ids.
When calling concurrent mode from Python, PDLP and Barrier were running on the same stream.
Now we explicitly create a new handle for Barrier with a new stream.