Skip to content

CMS implementation for 256x192x64NN#2758

Merged
jfactory07 merged 16 commits into
hipblaslt_common_cms_devfrom
users/jzhou/cms-256x192x64NN
Nov 28, 2025
Merged

CMS implementation for 256x192x64NN#2758
jfactory07 merged 16 commits into
hipblaslt_common_cms_devfrom
users/jzhou/cms-256x192x64NN

Conversation

@jfactory07
Copy link
Copy Markdown
Contributor

@jfactory07 jfactory07 commented Nov 19, 2025

Motivation

CMS implementation for 256x192x64NN

Test Result

test for 4096x3072x8192NN
got 4.7% uplift (enable CMS vs disable CMS)

I have run the test 11 times without failures
image

Submission Checklist

@jfactory07 jfactory07 changed the title Users/jzhou/cms 256x192x64 nn CMS implementation for 256x192x64NN Nov 19, 2025
@jfactory07 jfactory07 marked this pull request as ready for review November 19, 2025 10:52
@jfactory07 jfactory07 requested a review from a team as a code owner November 19, 2025 10:52
@sebvince sebvince added the gfx950 run CI on gfx950 label Nov 24, 2025
@talumbau talumbau self-requested a review November 24, 2025 20:21
Comment thread projects/hipblaslt/tensilelite/Tensile/Components/CustomSchedule.py Outdated
Copy link
Copy Markdown
Contributor

@sebvince sebvince left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM. Just need to shift be one the first GRB

Comment thread projects/hipblaslt/tensilelite/Tensile/Components/CustomSchedule.py
Comment thread projects/hipblaslt/tensilelite/Tensile/Components/CustomSchedule.py Outdated
@jfactory07 jfactory07 merged commit 593f72d into hipblaslt_common_cms_dev Nov 28, 2025
27 of 30 checks passed
@jfactory07 jfactory07 deleted the users/jzhou/cms-256x192x64NN branch November 28, 2025 05:03
b-shi pushed a commit that referenced this pull request Dec 12, 2025
CMS implementation for 256x192x64NN
ammallya pushed a commit that referenced this pull request Feb 3, 2026
[ROCm/composable_kernel commit: fcff004]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants