chore(ai_dynamo): set aiperf as default benchmark in vllm and sglang …

[saivishal1999]

Summary

• Sets workloads = "aiperf.sh" as the default in both vllm.toml and sglang.toml
• genai-perf remains available as opt-in
• Validated on EOS (H100) — single-node and multi-node runs both PASSED

Test Plan

• Ran on EOS H100 with aiperf.sh — both test.disagg.single-node and test.disagg.multinode PASSED
• aiperf_report.csv generated correctly

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Enterprise

Run ID: 9004b041-9621-406d-b395-62b352677f93

📥 Commits

Reviewing files that changed from the base of the PR and between 0738c4b and 7dd7f87.

📒 Files selected for processing (2)

• conf/experimental/ai_dynamo/test/sglang.toml
• conf/experimental/ai_dynamo/test/vllm.toml

---

📝 Walkthrough

Walkthrough

Two test configuration files update their workload script references from genai_perf.sh to aiperf.sh. The SGLang and vLLM test templates now point to the new aiperf workload script, while all other configuration values remain unchanged.

Changes

Workload Script Migration

Layer / File(s)	Summary
Update workload script references in test configurations conf/experimental/ai_dynamo/test/sglang.toml, conf/experimental/ai_dynamo/test/vllm.toml	Both test configuration files update the workloads setting from genai_perf.sh to aiperf.sh to use the new performance measurement script.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Possibly related PRs

• NVIDIA/cloudai#898: Adds AIPerf/aiperf.sh workload support, which aligns with these configuration changes that switch test templates to use aiperf.sh.

Suggested reviewers

• srivatsankrishnan
• jeffnvidia
• podkidyshev

Poem

🐰 A workload script swap,
From genai to aiperf's map,
Two configs aligned,
Performance refined,
Testing now follows the new track! ✨

🚥 Pre-merge checks | ✅ 4

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and specifically describes the main change: setting aiperf as the default benchmark in vllm and sglang configuration files.
Description check	✅ Passed	The description is directly related to the changeset, detailing what was changed (workloads setting), why (setting aiperf as default), and validation evidence from testing.
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

⚔️ Resolve merge conflicts

• Resolve merge conflict in branch spothula/aiperf-default

---

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

[coderabbitai]

Actionable comments posted: 0