[https://nvbugs/6473397][test] Raise BF16 Laguna DFlash memory skip threshold to prevent OOM

[DomBrown]

Summary by CodeRabbit

• Bug Fixes
  • Improved test compatibility by requiring additional device memory for the BF16+DFLASH accuracy test.
  • Removed outdated platform-specific test waivers so supported environments can run the accuracy test normally.

Description

• TestLagunaXS_2_1::test_bf16_dflash OOMs on 80 GB devices: BF16 Laguna-XS-2.1 weights are ~64 GiB, leaving too little headroom for KV cache + CUDA graph capture
  at max_batch_size=128, max_seq_len=4096.
• Raise skip_less_device_memory from 80000 → 90000 on BF16 only. FP8 and NVFP4 variants fit on 80 GB and are unchanged.
• Drop the now-redundant A100/H100 waives (nvbugs/6473397) — the marker handles skipping structurally.

Test plan

• pytest test_bf16_dflash skips cleanly on H100 80 GB
• Verified passing on RTX PRO 6000 Blackwell (~96 GiB)

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.

[DomBrown]

/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: 2991cbfc-78d7-4537-87ae-5a32ad8a4edb

📥 Commits

Reviewing files that changed from the base of the PR and between 5d8a786 and d982fb0.

📒 Files selected for processing (2)

• tests/integration/defs/accuracy/test_llm_api_pytorch.py
• tests/integration/test_lists/waives.txt

💤 Files with no reviewable changes (1)

• tests/integration/test_lists/waives.txt

---

📝 Walkthrough

Walkthrough

The BF16 DFLASH accuracy test now skips on devices with less than 90000 device memory units. Its A100 and H100 waiver entries are removed.

Changes

BF16 DFLASH test eligibility

Layer / File(s)	Summary
Update test threshold and waivers tests/integration/defs/accuracy/test_llm_api_pytorch.py, tests/integration/test_lists/waives.txt	Raises the BF16 DFLASH memory threshold from 80000 to 90000 and removes the A100 and H100 waiver entries.

Estimated code review effort: 1 (Trivial) | ~2 minutes

Possibly related PRs

• NVIDIA/TensorRT-LLM#16607: Modifies the same BF16 DFLASH waiver entries in the opposite direction.

Suggested reviewers: bowenfu

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title is specific, matches the change, and follows the ticket/type format.
Description check	✅ Passed	The PR includes the required summary, explanation, test plan, and checklist, with only a minor heading mismatch.
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 #60371 [ run ] triggered by Bot. Commit: d982fb0 Link to invocation

[tensorrt-cicd]

PR_Github #60371 [ run ] completed with state SUCCESS. Commit: d982fb0
/LLM/main/L0_MergeRequest_PR pipeline #48710 completed with status: 'SUCCESS'

CI Report

Link to invocation