Add Swizzling to tile scheduling #5158
Open
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
Summary: X-link: facebookresearch/FBGEMM#2158 as title Reviewed By: sryap Differential Revision: D86386070
Summary: X-link: facebookresearch/FBGEMM#2073 For causal and local causal mask, we better use the order below for deterministic. ``` qi\ki ki=0 ki=1 ki=2 ki=3 ki=4 ki=5 ki=6 ki=7 ki=8 ----------------------------------------------------------- qi=0 1 0 - - - - - - - qi=1 2 1 0 - - - - - - qi=2 - 2 1 0 - - - - - qi=3 - - 2 1 0 - - - - qi=4 - - - 2 1 0 - - - qi=5 - - - - 2 1 0 - - qi=6 - - - - - 2 1 0 - qi=7 - - - - - - 2 1 0 ``` Reviewed By: v0i0 Differential Revision: D85308820
Summary: X-link: facebookresearch/FBGEMM#2159 This diff adds swizzling to the tile scheduling in the CUTLASS Blackwell(BWD) kernel. The changes include adding L2 cache swizzle parameters, calculating them on the host, and using them to swizzle the tile scheduling. The swizzling is designed to improve cache utilization and reduce memory accesses. Differential Revision: D87572384
Contributor
|
@Aya-ZIbra has exported this pull request. If you are a Meta employee, you can view the originating Diff in D87572384. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.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.
Summary:
X-link: https://github.com/facebookresearch/FBGEMM/pull/2159
This diff adds swizzling to the tile scheduling in the CUTLASS Blackwell(BWD) kernel. The changes include adding L2 cache swizzle parameters, calculating them on the host, and using them to swizzle the tile scheduling. The swizzling is designed to improve cache utilization and reduce memory accesses.
Differential Revision: D87572384