Skip to content

optimize 3264x3072x1536xNxN#2558

Closed
jfactory07 wants to merge 3 commits into
users/common/cvs_devfrom
users/jzhou/3264x3072x1536xNxN-2
Closed

optimize 3264x3072x1536xNxN#2558
jfactory07 wants to merge 3 commits into
users/common/cvs_devfrom
users/jzhou/3264x3072x1536xNxN-2

Conversation

@jfactory07
Copy link
Copy Markdown
Contributor

@jfactory07 jfactory07 commented Nov 10, 2025 •
edited
Loading

Motivation

optimize 3264x3072x1536xNxN

Technical Details

change MT to 192x256x64, so it can leverage current CMS

Test Result

test for 3264x3072x1536xNxN
got 15% uplift

Submission Checklist

@jfactory07 jfactory07 added the gfx950 run CI on gfx950 label Nov 10, 2025
AlexBrownAMD pushed a commit that referenced this pull request Nov 10, 2025
@sjain-stanford @AaronStGeorge
[FusilliPlugin] Cleanup draft RFC (#2558)
87668e3 
Official RFC is here:
https://github.com/ROCm/TheRock/blob/main/docs/rfcs/RFC0004-Fusilli-IREE-Kernel-Provider-hipDNN.md
@jfactory07 jfactory07 marked this pull request as ready for review November 11, 2025 01:16
@jfactory07 jfactory07 requested a review from a team as a code owner November 11, 2025 01:16
@jfactory07 jfactory07 requested a review from b-shi November 11, 2025 01:16
@jfactory07 jfactory07 mentioned this pull request Nov 11, 2025
Closed
1 task
@jfactory07 jfactory07 closed this Nov 11, 2025
@jfactory07 jfactory07 deleted the users/jzhou/3264x3072x1536xNxN-2 branch November 11, 2025 02:49
ammallya pushed a commit that referenced this pull request Feb 3, 2026
@jefyang1
Add gemm universal f8 f8 bf16 mk nk instances on gfx950 (#2558)
a111f65 
[ROCm/composable_kernel commit: c64a0c6]
ammallya pushed a commit that referenced this pull request Feb 3, 2026
@illsilin
Revert "Add gemm universal f8 f8 bf16 mk nk instances on gfx950 (#2558)…
6c83ba0 
…" (#2584)

This reverts commit a111f65.

[ROCm/composable_kernel commit: b80099c]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@b-shi b-shi Awaiting requested review from b-shi

Assignees

No one assigned

Labels

gfx950 run CI on gfx950 organization: ROCm project: hipblaslt

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@jfactory07