[None][test] Reduce Nemotron V3 Super/Ultra pre-merge test cases

[nv-guomingz]

Consolidate Nemotron V3 Super/Ultra accuracy tests since both models share the same NemotronH architecture and code paths:

• Remove Super BF16 tests (test_auto_dtype_4gpus, test_bf16_trtllm_gen_moe_backend, test_bf16_4gpu_mtp_ar)
• Remove Super test_nvfp4_8gpus; 8-GPU MoE backend coverage is represented by the Ultra variants
• Remove Ultra test_nvfp4_4gpu_mtp_ar (duplicate of the Super case) and drop the B300 waive of the Super case (nvbugs/6336990)
• Use context manager for LLM in MTP acceptance-rate tests to ensure proper shutdown
• Rename Super block_reuse ids: TEP4_ADP -> DEP4_MTP_OFF, TEP4_ADP_MTP -> DEP4_MTP_ON
• Move to post-merge: Super fp8_4gpus (all combos), 8gpus_mtp_custom_op, parallelism[TP4_PP2/TP8_PP1], online_eplb[TRTLLM], block_reuse[TEP4/DEP4_MTP_OFF]; Ultra online_eplb[CUTEDSL], parallelism[ADP2_PP2]

Pre-merge L0 cases reduced from 22 to 7 while keeping one representative per functional dimension (MoE backends, MTP accuracy and acceptance rate, block reuse with TEP/DEP mappings, Hopper W4A16); the QA list retains broader coverage.

Summary by CodeRabbit

• Bug Fixes

  • Updated several integration test selections to better match current GPU coverage and merge-stage expectations.
  • Removed outdated waived test coverage so the affected scenario can run normally again.

• Tests

  • Refined Nemotron V3 Super/Ultra test lists across multiple platforms.
  • Moved some cases between pre-merge and post-merge suites and reduced redundant 4-GPU coverage 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)

• If PR introduces API changes, an appropriate PR label is added - either api-compatible or api-breaking. For api-breaking, include BREAKING in the PR title.

• 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.

[nv-guomingz]

/bot run

[coderabbitai]

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: cf2b642f-75bd-4324-8adc-edd1d470e200

📥 Commits

Reviewing files that changed from the base of the PR and between c92ac90 and e419641.

📒 Files selected for processing (6)

• tests/integration/defs/accuracy/test_llm_api_pytorch.py
• tests/integration/test_lists/qa/llm_function_core.txt
• tests/integration/test_lists/test-db/l0_dgx_b200.yml
• tests/integration/test_lists/test-db/l0_dgx_h100.yml
• tests/integration/test_lists/test-db/l0_gb200_multi_gpus.yml
• tests/integration/test_lists/waives.txt

💤 Files with no reviewable changes (1)

• tests/integration/test_lists/waives.txt

---

📝 Walkthrough

Walkthrough

This PR reorganizes Nemotron V3 Super/Ultra integration tests: several test methods are removed, NVFP4 parametrization ids are renamed, MTP acceptance-rate tests are refactored to use LLM as a context manager, and corresponding QA test lists, test-db matrices (B200/H100/GB200), and a waiver entry are updated to match.

Changes

Nemotron V3 test suite reorganization

Layer / File(s)	Summary
Test method removals and LLM context-manager refactor tests/integration/defs/accuracy/test_llm_api_pytorch.py	Removes test_auto_dtype_4gpus, test_bf16_trtllm_gen_moe_backend, an NVFP4 8-GPU placeholder, and TestNemotronV3Ultra.test_nvfp4_4gpu_mtp_ar; renames NVFP4 parametrization ids from TEP4_ADP* to DEP4_MTP_OFF/DEP4_MTP_ON; refactors test_nvfp4_4gpu_mtp_ar and test_nvfp4_4gpu_mtp_ar_custom_op to build LLM via a with context manager.
QA test list updates tests/integration/test_lists/qa/llm_function_core.txt	Removes entries for removed/renamed tests (test_nvfp4_4gpu_mtp_ar, test_bf16_4gpu_mtp_ar, test_nvfp4_8gpus[attention_dp_on-...], test_bf16_trtllm_gen_moe_backend) and adds test_nvfp4_4gpus_block_reuse[DEP4_MTP_ON].
DGX B200/H100 test-db matrix updates tests/integration/test_lists/test-db/l0_dgx_b200.yml, tests/integration/test_lists/test-db/l0_dgx_h100.yml	Moves NVFP4 8-GPU and test_auto_dtype_4gpus entries between pre/post-merge sections on B200, and moves test_fp8_4gpus mamba-cache entries from pre-merge to post-merge on H100.
GB200 multi-GPU matrix updates and waiver cleanup tests/integration/test_lists/test-db/l0_gb200_multi_gpus.yml, tests/integration/test_lists/waives.txt	Replaces/reinserts block_reuse, online_eplb, and parallelism test entries across GB200 sections; removes the obsolete waiver for test_nvfp4_4gpu_mtp_ar.

Estimated code review effort: 2 (Simple) | ~12 minutes

Possibly related PRs

• NVIDIA/TensorRT-LLM#16109: Also modifies tests/integration/test_lists/waives.txt waiver entries for Nemotron V3 Super nvfp4/test_mtp integration tests.

Suggested reviewers: mzweilz, jieli-matrix, dongxuy04

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	The PR has a clear summary and checklist, but the required Description and Test Coverage sections are empty.	Fill in the Description and Test Coverage sections with a brief issue/solution summary and the tests or CI coverage that validate the change.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title is concise and accurately summarizes the main change: reducing Nemotron V3 Super/Ultra pre-merge test cases.
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]

PR_Github #58245 [ run ] triggered by Bot. Commit: e419641 Link to invocation

[tensorrt-cicd]

PR_Github #58245 [ run ] completed with state FAILURE. Commit: e419641
/LLM/main/L0_MergeRequest_PR pipeline #46886 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

CI Agent Failure Analysis

Link to invocation

[nv-guomingz]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #58339 [ run ] triggered by Bot. Commit: e419641 Link to invocation

[QiJune]

LGTM

[tensorrt-cicd]

PR_Github #58339 [ run ] completed with state SUCCESS. Commit: e419641
/LLM/main/L0_MergeRequest_PR pipeline #46968 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

CI Agent Failure Analysis

Link to invocation

[nv-guomingz]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #58409 [ run ] triggered by Bot. Commit: e419641 Link to invocation

[tensorrt-cicd]

PR_Github #58409 [ run ] completed with state SUCCESS. Commit: e419641
/LLM/main/L0_MergeRequest_PR pipeline #47028 completed with status: 'SUCCESS'

CI Report

Link to invocation