[Hopper TMA] Add intrinsic to create barriers for synchronization#15684
Merged
csullivan merged 4 commits intoapache:mainfrom Sep 11, 2023
Merged
[Hopper TMA] Add intrinsic to create barriers for synchronization#15684csullivan merged 4 commits intoapache:mainfrom
csullivan merged 4 commits intoapache:mainfrom
Conversation
JosephTheOctonaut
approved these changes
Sep 7, 2023
csullivan
approved these changes
Sep 7, 2023
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR adds an intrinsic to create barriers that can be used with existing barrier intrinsics for synchronization. The prior method of barrier allocation was to use
alloc_buffere.g. as followsbarrier = T.alloc_buffer([1], "uint64", scope="shared")and then pass the pointer and offset to that barrier allocation for use in the barrier intrinsics. This was a functional interface, but also caused problems with alignment of other non-barrier shared memory allocations. See removed workarounds marked withTODOin the tests in this PR. At the expense of the additionalcreate_barriersintrinsic we get a simplified interface using barrier ID rather than pointer / offset passed to the barrier intrinsics and more low-level codegen control which is used in this PR to solve the alignment issue.