[None][test] Add more models into Pre-merge Perf Test

[chenfeiz0326]

Summary by CodeRabbit

• Tests
  • Added new pre-merge performance sanity test stage for GB200-4 GPU PyTorch configurations (split group 2).
  • Updated test entries for DeepSeek R1 FP4 v2 model variants across pre-merge and post-merge stages.
  • Adjusted test configurations for qwen model variants.
  • Modified test skip list entries.

Description

Test Coverage

PR Checklist

Please review the following before submitting your PR:

• PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.

• PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.

• Test cases are provided for new code paths (see test instructions)

• Any new dependencies have been scanned for license and vulnerabilities

• CODEOWNERS updated if ownership changes

• Documentation updated as needed

• Update tava architecture diagram if there is a significant design change in PR.

• The reviewers assigned automatically/manually are appropriate for the PR.

• Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

To see a list of available CI bot commands, please comment /bot help.

[chenfeiz0326]

/bot run --disable-fail-fast --stage-list "GB200-4_GPUs-PyTorch-PerfSanity-1,GB200-4_GPUs-PyTorch-PerfSanity-2"

[coderabbitai]

📝 Walkthrough

Walkthrough

This PR adds a new pre-merge GB200 performance sanity test stage (split 2/2) to the CI pipeline, rebalances test cases between pre_merge and post_merge phases in the test list, and removes one previously skipped test entry.

Changes

GB200 Performance Sanity Test Configuration

Layer / File(s)	Summary
CI Pipeline Stage Configuration jenkins/L0_Test.groovy	A new GB200-4_GPUs-PyTorch-PerfSanity-2 stage is added to the SBSA test dispatch configuration, configured as split group 2/2 and using the l0_gb200_multi_gpus_perf_sanity test list.
Test List Definition Updates tests/integration/test_lists/test-db/l0_gb200_multi_gpus_perf_sanity.yml	Pre_merge phase removes multiple DeepSeek R1 FP4 v2 variants and adds ctx-only entries; post_merge phase relocates the DeepSeek 1k1k aggregated test, consolidates k25-thinking-fp4 entries to a single line, and reduces qwen3-235b-fp4 ctx-only coverage from multiple variants to a single case.
Test Waives Update tests/integration/test_lists/waives.txt	Removes the skip entry for perf/test_perf_sanity.py::test_e2e[aggr_upload-deepseek_r1_fp4_v2_grace_blackwell-...-mtp1_1k1k], allowing the test to run.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

🚥 Pre-merge checks | ✅ 3 | ❌ 2

❌ Failed checks (1 warning, 1 inconclusive)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	The PR description is incomplete—it contains only the template structure with placeholders and no actual content filling the 'Description' and 'Test Coverage' sections.	Complete the Description section explaining what models were added and why; fill the Test Coverage section listing which test lists and stages verify these changes.
Title check	❓ Inconclusive	The title '[None][test] Add more models into Pre-merge Perf Test' is vague and generic, using non-descriptive terms like 'more models' without specifying which models or test configurations are affected.	Replace with a more specific title that identifies the key changes, e.g., '[None][test] Add GB200-4 GPU PyTorch PerfSanity split stage and update model test coverage'.

✅ Passed checks (3 passed)

Check name	Status	Explanation
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.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🧹 Nitpick comments (1)

tests/integration/test_lists/test-db/l0_gb200_multi_gpus_perf_sanity.yml (1)

23-31: Call out whether scheduled QA should pick these new GB200 perf cases up too.

This test-db update is enough for CI execution, but scheduled QA will only see the added ctx-only / k25 / qwen coverage if the corresponding tests/integration/test_lists/qa/llm_perf_*.yml entry is updated as well. If this PR is intentionally pre-merge-only, please say so in the PR; otherwise mirror the additions there too.

As per coding guidelines, "If the change adds or materially alters an integration test under tests/integration/defs/ (or otherwise affects what QA should run on a schedule), call out whether an entry is needed under tests/integration/test_lists/qa/."

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tests/integration/test_lists/test-db/l0_gb200_multi_gpus_perf_sanity.yml`
around lines 23 - 31, Summary: New GB200 perf cases (ctx-only, k25, qwen) were
added to the integration test list but the scheduled QA lists were not updated.
Update the QA scheduled test list named llm_perf_*.yml to mirror the new entries
added in test-db (the ctx-only GB200 cases, k25-thinking FP4 variants, and
qwen3-235b-fp4 variants) so scheduled QA will run them, or explicitly mark this
PR as pre-merge-only in the PR description if you do not want QA to pick them up
now; ensure you reference the exact test ids from the diff (e.g.,
aggr_upload-ctx_only-gb200_..., aggr_upload-k25_thinking_fp4_...,
aggr_upload-ctx_only-gb200_qwen3-235b-fp4_...) when adding or skipping them.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@jenkins/L0_Test.groovy`:
- Around line 3841-3842: The GB200 PerfSanity shard entries
("GB200-4_GPUs-PyTorch-PerfSanity-1" and "GB200-4_GPUs-PyTorch-PerfSanity-2")
are not being marked perf-mode because the code relies on
key.contains("-Perf-"); update the stage creation for these keys to explicitly
set perfMode=true by invoking the existing perf-mode helper used elsewhere (the
same helper you apply when perf-ness is inferred from the name) instead of
relying on key.contains("-Perf-"), so the perf-only pytest flags and perf
regression/report checks run for these shards.

---

Nitpick comments:
In `@tests/integration/test_lists/test-db/l0_gb200_multi_gpus_perf_sanity.yml`:
- Around line 23-31: Summary: New GB200 perf cases (ctx-only, k25, qwen) were
added to the integration test list but the scheduled QA lists were not updated.
Update the QA scheduled test list named llm_perf_*.yml to mirror the new entries
added in test-db (the ctx-only GB200 cases, k25-thinking FP4 variants, and
qwen3-235b-fp4 variants) so scheduled QA will run them, or explicitly mark this
PR as pre-merge-only in the PR description if you do not want QA to pick them up
now; ensure you reference the exact test ids from the diff (e.g.,
aggr_upload-ctx_only-gb200_..., aggr_upload-k25_thinking_fp4_...,
aggr_upload-ctx_only-gb200_qwen3-235b-fp4_...) when adding or skipping them.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: fa96727d-267d-4994-bb9c-90a0fbcb4d3c

📥 Commits

Reviewing files that changed from the base of the PR and between c18cf53 and 003c818.

📒 Files selected for processing (3)

• jenkins/L0_Test.groovy
• tests/integration/test_lists/test-db/l0_gb200_multi_gpus_perf_sanity.yml
• tests/integration/test_lists/waives.txt

💤 Files with no reviewable changes (1)

• tests/integration/test_lists/waives.txt

[tensorrt-cicd]

PR_Github #47329 [ run ] triggered by Bot. Commit: 003c818 Link to invocation

[tensorrt-cicd]

PR_Github #47329 [ run ] completed with state SUCCESS. Commit: 003c818
/LLM/main/L0_MergeRequest_PR pipeline #37268 (Partly Tested) completed with status: 'SUCCESS'

CI Report

Link to invocation

[chenfeiz0326]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #47603 [ run ] triggered by Bot. Commit: a353116 Link to invocation

[tensorrt-cicd]

PR_Github #47603 [ run ] completed with state SUCCESS. Commit: a353116
/LLM/main/L0_MergeRequest_PR pipeline #37509 completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

• Please check the failed tests and fix your PR
• If you cannot view the failures, ask the CI triggerer to share details
• Once fixed, request an NVIDIA team member to trigger CI again

Link to invocation

[chenfeiz0326]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #47629 [ run ] triggered by Bot. Commit: a353116 Link to invocation

[chenfeiz0326]

/bot skip --comment "Only add new pre-merge perf tests, no need to run the whole CI pipeline"

[tensorrt-cicd]

PR_Github #47646 [ skip ] triggered by Bot. Commit: a353116 Link to invocation

[tensorrt-cicd]

PR_Github #47646 [ skip ] completed with state SUCCESS. Commit: a353116
Skipping testing for commit a353116

Link to invocation