Skip to content

[https://nvbugs/6468821][infra] Split slow B300 attention unit tests - #16709

Merged
yuxianq merged 2 commits into
NVIDIA:mainfrom
yuxianq:refactor/split-b300-attention-tests
Jul 22, 2026
Merged

[https://nvbugs/6468821][infra] Split slow B300 attention unit tests#16709
yuxianq merged 2 commits into
NVIDIA:mainfrom
yuxianq:refactor/split-b300-attention-tests

Conversation

@yuxianq

@yuxianq yuxianq commented Jul 22, 2026

Copy link
Copy Markdown
Collaborator

Dev Engineer Review

  • Updated tests/integration/test_lists/test-db/l0_b300.yml to restructure the l0_b300 pre_merge PyTorch attention selection: unittest/_torch/attention now excludes three slow CuTe DSL sparse attention tests via --ignore, and the excluded sparse tests are then added back as standalone list entries.
  • Kept the change scoped to the intended l0_b300 pre-merge PyTorch test list to preserve coverage while reducing aggregate attention runtime and enabling independent Jenkins sharding of the slow cases.
  • Config remains consistent with the expected test-list schema; YAML/test-list validation passed via pre-commit. No API, dependency, or public entity changes were introduced.

QA Engineer Review

  • Modified: tests/integration/test_lists/test-db/l0_b300.yml
    • Removed the three following sparse CuTe DSL attention tests from the aggregate unittest/_torch/attention run by adding --ignore exclusions:
      • unittest/_torch/attention/sparse/test_cute_dsl_fp8_paged_mqa_logits.py
      • unittest/_torch/attention/sparse/test_cute_dsl_fp4_paged_mqa_logits.py
      • unittest/_torch/attention/sparse/test_cute_dsl_gvr_topk_decode.py
    • Added the same three tests back as separate standalone entries in the list.
  • Verdict: needs follow-up (CBTS coverage data is unavailable).

Description

Split the three slow CuTe DSL attention test files out of the aggregate B300 attention entry. The aggregate entry ignores those files, and each file remains covered through a standalone CI entry so Jenkins can shard them independently.

The measured serial file profile attributed 2,721 of 4,257 seconds to these three files. Separating them reduces the aggregate attention invocation while preserving test coverage.

Test Coverage

  • Parsed l0_b300.yml and confirmed four distinct attention CI entries.
  • Confirmed all three standalone test paths exist.
  • Ran pre-commit against l0_b300.yml; all applicable hooks passed, including YAML and test-list validation.

PR Checklist

  • PR description clearly explains what and why.
  • PR follows the TRT-LLM coding guidelines to the best of my knowledge.
  • Existing tests remain covered through standalone CI entries.
  • No API changes are introduced.
  • No new dependencies are introduced.
  • CODEOWNERS changes are not required.
  • Documentation and architecture updates are not required.
  • Please check this after reviewing the above items as appropriate for this PR.

@yuxianq
yuxianq requested review from a team as code owners July 22, 2026 04:46
@yuxianq

yuxianq commented Jul 22, 2026

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast

Signed-off-by: Yuxian Qiu <142763828+yuxianq@users.noreply.github.com>
@yuxianq
yuxianq requested a review from EmmaQiaoCh July 22, 2026 04:47
@coderabbitai

coderabbitai Bot commented Jul 22, 2026

Copy link
Copy Markdown
Contributor

Review Change Stack

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info
⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: c35f64cd-33b0-41ac-b356-db2b855994a2

📥 Commits

Reviewing files that changed from the base of the PR and between 17ec48d and 54905a3.

📒 Files selected for processing (1)
  • tests/integration/test_lists/test-db/l0_b300.yml
🚧 Files skipped from review as they are similar to previous changes (1)
  • tests/integration/test_lists/test-db/l0_b300.yml

Walkthrough

The PyTorch integration test list excludes three sparse attention tests from the broader attention test invocation and runs them as separate entries.

Changes

Attention test selection

Layer / File(s) Summary
Split attention test execution
tests/integration/test_lists/test-db/l0_b300.yml
The broad unittest/_torch/attention entry adds three ignore flags, and the excluded sparse tests are registered as independent test entries.

Estimated code review effort: 2 (Simple) | ~10 minutes

Suggested reviewers: mzweilz

🚥 Pre-merge checks | ✅ 5
✅ Passed checks (5 passed)
Check name Status Explanation
Title check ✅ Passed The title matches the PR and follows the required ticket/type/summary format.
Description check ✅ Passed The description includes the required Description, Test Coverage, and PR Checklist sections and is sufficiently specific.
Docstring Coverage ✅ Passed No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check ✅ Passed Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check ✅ Passed Check skipped because no linked issues were found for this pull request.
✨ Finishing Touches
🧪 Generate unit tests (beta)
  • Create PR with unit tests

Comment @coderabbitai help to get the list of available commands.

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #60888 [ run ] triggered by Bot. Commit: 17ec48d Link to invocation

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #60888 [ run ] completed with state SUCCESS. Commit: 17ec48d
/LLM/main/L0_MergeRequest_PR pipeline #49156 completed with status: 'SUCCESS'

CI Report

Link to invocation

@yuxianq

yuxianq commented Jul 22, 2026

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast

…CBTS

Signed-off-by: Yuxian Qiu <142763828+yuxianq@users.noreply.github.com>
@yuxianq

yuxianq commented Jul 22, 2026

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #60924 [ run ] triggered by Bot. Commit: 54905a3 Link to invocation

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #60925 [ run ] triggered by Bot. Commit: 54905a3 Link to invocation

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github/16709-54905a3 #60924 was force-killed by a newer pipeline run.
L0 job information not available (job may not have been triggered yet).

Link to superseding invocation

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #60925 [ run ] completed with state SUCCESS. Commit: 54905a3
/LLM/main/L0_MergeRequest_PR pipeline #49190 completed with status: 'SUCCESS'

CI Report

Link to invocation

@yuxianq
yuxianq merged commit 7965292 into NVIDIA:main Jul 22, 2026
10 checks passed
yuanjingx87 pushed a commit to yuanjingx87/TensorRT-LLM that referenced this pull request Jul 26, 2026
…VIDIA#16709)

Signed-off-by: Yuxian Qiu <142763828+yuxianq@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants