[https://nvbugs/6194812][test] Update llm_perf_core.yml to require a minimum of 4 GPUs and add new performance tests

[yufeiwu-nv]

• Increased the minimum GPU count from 2 to 4 for llm_perf_core.
• Added new performance test cases for qwen3 models with specific configurations.

Signed-off-by: yufeiwu-nv 230315618+yufeiwu-nv@users.noreply.github.com

Summary by CodeRabbit

• Tests
  • Updated performance testing configuration with increased GPU resource requirements for improved system validation.
  • Added new test scenarios for comprehensive model configuration and scheduler optimization validation.

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.

[yufeiwu-nv]

/bot skip --comment "only test modify"

[yufeiwu-nv]

/bot skip --comment "only test modify"

[coderabbitai]

📝 Walkthrough

Walkthrough

This PR updates the FP8 performance test configuration file to increase the minimum GPU count requirement from 2 to 4 and adds two new test cases for Qwen models configured with parallelism and scheduler overlap parameters.

Changes

FP8 Performance Test Configuration

Layer / File(s)	Summary
FP8 test condition and scheduler test cases tests/integration/test_lists/qa/llm_perf_core.yml	System GPU count threshold for FP8 tests raised from 2 to 4, and two Qwen model test cases added with overlap scheduler configuration (con:1, pp:4-gpus:4).

🎯 1 (Trivial) | ⏱️ ~3 minutes

Possibly Related PRs

• NVIDIA/TensorRT-LLM#14749: Contains overlapping updates to llm_perf_core.yml with similar system_gpu_count minimum and FP8 test configuration changes.
• NVIDIA/TensorRT-LLM#14952: Modifies the same performance test matrix file with FP8/common scheduler-related parameterized test case adjustments.

Suggested Reviewers

• ruodil
• StanleySun639
• litaotju

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 inconclusive)

Check name	Status	Explanation	Resolution
Description check	❓ Inconclusive	The PR description provides the essential information about what changed (GPU count increase and new test cases) but lacks detailed explanations for 'why' and omits the Description and Test Coverage sections from the template.	Add a Description section explaining why the GPU count was increased and why new test cases are needed; add a Test Coverage section listing the test files affected.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and specifically describes the main changes: updating llm_perf_core.yml to require 4 GPUs minimum and adding new performance tests, directly matching the changeset modifications.
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

🤖 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 `@tests/integration/test_lists/qa/llm_perf_core.yml`:
- Line 25: The condition supports_fp8 in
tests/integration/test_lists/qa/llm_perf_core.yml currently sets
system_gpu_count: gte: 4 at the condition level, which unintentionally forces
every test under that condition to require 4+ GPUs; modify the YAML so
system_gpu_count: gte: 4 is removed from the top-level supports_fp8 condition
and instead add system_gpu_count: gte: 4 only to the individual test entries
that actually require 4 GPUs (or create a new sub-condition for
“requires_4_gpus” and attach it only to those tests); additionally verify the
two overlap scheduler entries that set maxnt: 2048 and input_output_len:
8000,1000 are deliberate by checking how maxnt is used in perf/test_perf.py and,
if needed, parameterize or document them so they match the perf runner’s
interpretation (adjust keys maxnt and input_output_len in the specific overlap
scheduler entries rather than changing the entire supports_fp8 condition).

🪄 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: dc784088-1d4b-4384-aee0-7adf7c7d03bc

📥 Commits

Reviewing files that changed from the base of the PR and between 2febb37 and e4ed8da.

📒 Files selected for processing (1)

• tests/integration/test_lists/qa/llm_perf_core.yml

[tensorrt-cicd]

PR_Github #52731 [ skip ] triggered by Bot. Commit: 9d695d8 Link to invocation

[tensorrt-cicd]

PR_Github #52731 [ skip ] completed with state SUCCESS. Commit: 9d695d8
Skipping testing for commit 9d695d8

Link to invocation

[yufeiwu-nv]

/bot skip --comment "only test modify"

[tensorrt-cicd]

PR_Github #52981 [ skip ] triggered by Bot. Commit: 5ae1eca Link to invocation

[tensorrt-cicd]

PR_Github #52981 [ skip ] completed with state SUCCESS. Commit: 5ae1eca
Skipping testing for commit 5ae1eca

Link to invocation

[yufeiwu-nv]

/bot skip --comment "only test modify"

[tensorrt-cicd]

PR_Github #52997 [ skip ] triggered by Bot. Commit: 8c648b4 Link to invocation

[tensorrt-cicd]

PR_Github #52997 [ skip ] completed with state SUCCESS. Commit: 8c648b4
Skipping testing for commit 8c648b4

Link to invocation