Add g+g problem configs to tier1 model list #2023
Open
+26
−0
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
dhernandez0
reviewed
Oct 8, 2025
dhernandez0
approved these changes
Oct 8, 2025
|
@dhernandez0 the CIs say too many unauthenticated requests, I'm not sure if you guys are aware of that? |
umangyadav
approved these changes
Oct 10, 2025
There was a problem hiding this comment.
Pull Request Overview
This PR adds g+g (GEMM+GEMM) problem configurations to the tier 1 model list for performance testing. These configurations enable testing of fused GEMM operations that include additional parameters like -transC, -transO, and -gemmO for more complex matrix multiplication patterns.
- Adds g+g problem configs generated using migraphx-benchmark-utils script
- Adds configurations for various models including BERT, GPT-2, Whisper, and Stable Diffusion variants
- Includes both f32 and f16 precision variants across different model types
Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.
dhernandez0
reviewed
Oct 13, 2025
| -t f32 -out_datatype f32 -transA false -transB false -g 1 -m 1 -n 4096 -k 14336 | ||
| -t f32 -out_datatype f32 -transA false -transB false -g 1 -m 1 -n 4096 -k 4096 | ||
| -t f32 -out_datatype f32 -transA false -transB false -g 1 -m 1 -n 32000 -k 4096 | ||
| #unet_model |
There was a problem hiding this comment.
It's in this list: https://amd.atlassian.net/wiki/spaces/MLSE/pages/934586591/Tier+1+models
But it did not previously have configs in this file
bdevorem
force-pushed
the
bdevorem/geg_t1_problemconfigs
branch
from
5464398 to
73c9be5
Compare
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.
Motivation
This PR just adds the g+g problem configs for tier 1 models to the preexisting list.
Closes rocMLIR-internal #2027. The issue is on rocmlir-internal, so it may need to be closed manually.
Technical Details
Problem configs were generated with this script, plus the env var for g+g fusion and filtering for g+g problem configs, and those resulting problem configs were added to this list.
Test Plan
N/A
Test Result
N/A
Submission Checklist