Add accelerator-awareness to most allreduce implementations #12611
Conversation
This file contains 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
Adding accelerator-awareness requires allocation of temporary memory on the a device selected based on the input buffer (possibly cached by the allocator) and reintroduces the use of 3buff reductions to combine copy and operator application. This change also improves performance on CPU for larger operations. Where possible Signed-off-by: Joseph Schuchart <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
This is the fifth patch based on #12318 and introduces accelerator-awareness for the most of the allreduce implementations in
coll/base. There were changes introduced toompi_coll_base_allreduce_intra_allgather_reduce, which led to conflicts. These implementations will still need to be adapted.This PR relies on the following PRs:
Adding accelerator-awareness requires allocation of temporary memory on the a device selected based on the input buffer (possibly cached by the allocator) and reintroduces the use of 3buff reductions to combine copy and operator application. This change also improves performance on CPU for larger operations. Where possible