Skip to content

Conversation

@victor-eds
Copy link
Contributor

Add pass splitting reductions and performing layout conversions to avoid using sub-group operations.

Add pass splitting reductions and performing layout conversions to avoid using sub-group operations.

Signed-off-by: victor-eds <[email protected]>
@victor-eds victor-eds requested review from a team, etiotto and whitneywhtsang October 15, 2024 12:06
@victor-eds victor-eds self-assigned this Oct 15, 2024
@victor-eds
Copy link
Contributor Author

Part of #2266.

@victor-eds victor-eds changed the title [OptRed] Define -triton-intelgpu-optimize-reduction pass [OptRed] Define -tritonintelgpu-optimize-reduction-locality pass Oct 15, 2024
@victor-eds victor-eds marked this pull request as draft October 15, 2024 13:09
@victor-eds victor-eds marked this pull request as ready for review October 15, 2024 16:28
@vlad-penkin vlad-penkin linked an issue Oct 16, 2024 that may be closed by this pull request
@etiotto
Copy link
Contributor

etiotto commented Oct 22, 2024

Merging it in, if other reviewers have comments add them as post review comments.

@etiotto etiotto merged commit a9ca5f0 into main Oct 22, 2024
4 checks passed
@etiotto etiotto deleted the fast-sub-group-transpose branch October 22, 2024 13:31
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Port "sub-group transpose reduction" to default path

4 participants