Skip to content

[qwen next] Fuse QSA KV preparation and sparse block expansion - #40972

Merged
Qiaolin-Yu merged 4 commits into
mainfrom
qsa-fused-kv-prepare
Sep 30, 2026
Merged

[qwen next] Fuse QSA KV preparation and sparse block expansion#40972
Qiaolin-Yu merged 4 commits into
mainfrom
qsa-fused-kv-prepare

Conversation

@Qiaolin-Yu

@Qiaolin-Yu Qiaolin-Yu commented Sep 23, 2026 •
edited by github-actions Bot
Loading

Copy link
Copy Markdown
Collaborator

Motivation

before,
image
after,
image

Modifications

Accuracy Tests

Speed Tests and Profiling

Checklist

Review and Merge Process

  1. Ping Merge Oncalls to start the process. See the PR Merge Process.
  2. Get approvals from CODEOWNERS and other reviewers.
  3. Trigger CI tests with comments or contact authorized users to do so.
    • Common commands include /tag-and-rerun-ci, /tag-run-ci-label, /rerun-failed-ci
  4. After green CI and required approvals, ask Merge Oncalls or people with Write permission to merge the PR.

CI States

Latest PR Test (Base): ❌ Run #36622133756
Latest PR Test (Extra): ❌ Run #36622133257
Latest PR Test (AMD ROCm 10): ❌ Run #36622133710

This was referenced Sep 24, 2026
Merged
Open
Open
Merged
Open
Open
Open
Merged
Open
Merged
Open
YAMY1234
YAMY1234 reviewed Sep 26, 2026
View reviewed changes
Comment thread test/registered/kernels/ops/attention/qsa/test_qsa_fused_kv_prepare.py
Comment thread python/sglang/srt/layers/attention/qwen_sparse_attn_backend.py Outdated
Comment thread python/sglang/srt/layers/attention/qwen_sparse_attn_backend.py
Comment thread python/sglang/srt/layers/attention/qsa/qsa_indexer.py Outdated
@Qiaolin-Yu
Qiaolin-Yu requested a review from YAMY1234 September 28, 2026 20:03
@Qiaolin-Yu Qiaolin-Yu added run-ci CI: run the baseline test suite on this PR bypass-fail-fast CI: a failing job no longer aborts its siblings (lint still gates) labels Sep 28, 2026
@Qiaolin-Yu

Qiaolin-Yu commented Sep 28, 2026

Copy link
Copy Markdown
Collaborator Author

/tag-and-rerun-ci

@Qiaolin-Yu
Qiaolin-Yu merged commit 88f95f4 into main Sep 30, 2026
344 of 374 checks passed
@Qiaolin-Yu
Qiaolin-Yu deleted the qsa-fused-kv-prepare branch September 30, 2026 04:54
arathi-hlab added a commit to arathi-hlab/sglang that referenced this pull request Sep 30, 2026
@arathi-hlab
[XPU] accept defer_expansion in test_qsa_indexer_xpu's fake indexer
c97d0be 
sgl-project#40972 passes defer_expansion to select_decode_tokens and updated the
CUDA test's _DispatchIndexer but not this XPU copy, so it raised
TypeError in stage-a.
@arathi-hlab arathi-hlab mentioned this pull request Sep 30, 2026
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@YAMY1234 YAMY1234 YAMY1234 approved these changes
@merrymercy merrymercy Awaiting requested review from merrymercy merrymercy is a code owner
@Fridge003 Fridge003 Awaiting requested review from Fridge003 Fridge003 is a code owner
@ispobock ispobock Awaiting requested review from ispobock ispobock is a code owner
@hebiao064 hebiao064 Awaiting requested review from hebiao064 hebiao064 is a code owner
@HaiShaw HaiShaw Awaiting requested review from HaiShaw HaiShaw is a code owner

Assignees

@YAMY1234 YAMY1234

Labels

bypass-fail-fast CI: a failing job no longer aborts its siblings (lint still gates) run-ci CI: run the baseline test suite on this PR

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@Qiaolin-Yu @YAMY1234