[GPU] modify GroupQueryAttention decompose logic to fit gpu unsqueeze_broadcast_reshape_sdpa_fusion pattern #31507
+18
−7
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
github-actions
bot
added
the
category: transformations
…a_fusion pattern for MQA model performance
bopeng1234
force-pushed
the
gqa_multi_query_atten_fix_gpu
branch
from
ecbe728 to
be4e135
Compare
rkazants
requested changes
Jul 31, 2025
|
added JIRA, https://jira.devtools.intel.com/browse/CVS-171437 |
|
Hi @rkazants , I think these is existing GQA test with query head 2, kv head 1 This PR only changes the implementation of the kv broadcast to q, the existing testcase can cover and verify this changes, so I think we don't need to add another duplicate testcase. |
|
build_jenkins |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
For MultiQueryAttention model, for example, Qwen2.5-1.5B, KV head num = 2, Q head num = 12.
KV need to broadcast to 12 then do SDPA computation.
GPU has pattern to fuse KV broadcast into SDPA kernel to raise performance, in unsqueeze_broadcast_reshape_sdpa_fusion.cpp
This PR adjust GQA decompose logic to match the pattern.