Skip to content
This repository was archived by the owner on Apr 20, 2026. It is now read-only.

Grho/feb2 b add >5 TTFT load points#132

Merged
gracehonv merged 4 commits intomainfrom
grho/feb2_b
Feb 3, 2026
Merged

Grho/feb2 b add >5 TTFT load points#132
gracehonv merged 4 commits intomainfrom
grho/feb2_b

Conversation

@gracehonv
Copy link
Copy Markdown
Collaborator

@gracehonv gracehonv commented Feb 3, 2026

Summary by CodeRabbit

  • Chores
    • Removed Lustre mount configurations from several deployment recipes.
    • Added explicit nginx frontend container references to specific configurations.
    • Adjusted benchmark concurrency parameters across deployment profiles to optimize testing scenarios.

@coderabbitai
Copy link
Copy Markdown
Contributor

coderabbitai Bot commented Feb 3, 2026

Caution

Review failed

The pull request is closed.

📝 Walkthrough

Walkthrough

This PR modifies YAML recipe configuration files for GB300-FP8 model variants across different input/output dimensions (1k1k, 1k8k, 8k1k). Changes include removing Lustre mount configurations from several variants, adding explicit nginx container frontend references to 1k8k variants, and adjusting benchmark concurrency parameters in select files.

Changes

Cohort / File(s) Summary
1k1k Lustre removal
recipes/gb300-fp8/1k1k/stp/low-latency.yaml, recipes/gb300-fp8/1k1k/stp/max.yaml
Removed extra\_mount blocks that mounted /lustre to container.
1k8k frontend additions
recipes/gb300-fp8/1k8k/stp/low-latency.yaml, recipes/gb300-fp8/1k8k/stp/mid.yaml
Added frontend section with explicit nginx\_container: nginx configuration.
1k8k max configuration
recipes/gb300-fp8/1k8k/stp/max.yaml
Added frontend nginx\_container entry and expanded benchmark concurrencies from [8192] to [8192,10240].
8k1k Lustre removal
recipes/gb300-fp8/8k1k/stp/low-latency.yaml, recipes/gb300-fp8/8k1k/stp/max.yaml, recipes/gb300-fp8/8k1k/stp/mid.yaml
Removed extra\_mount Lustre mount blocks and reduced benchmark concurrency values in max.yaml and mid.yaml.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Possibly related PRs

Suggested reviewers

  • trevor-m
  • kyleliang-nv

Poem

🐰 Recipe tweaks, a hop and a bound,
Lustre mounts removed, left the ground,
Nginx containers now clearly named,
Concurrency tuned—performance claimed! ✨

✨ Finishing touches
🧪 Generate unit tests (beta)
  • Create PR with unit tests
  • Post copyable unit tests in a comment
  • Commit unit tests in branch grho/feb2_b

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

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

@gracehonv gracehonv merged commit 7b0a38a into main Feb 3, 2026
4 of 5 checks passed
gracehonv added a commit that referenced this pull request Feb 3, 2026
* add nginx frontend containers

* lustre

* lustre

* modify points to put back TTFT>5 points

---------

Co-authored-by: Grace Ho <grho@login-lyris02.lyris.clusters.nvidia.com>
Co-authored-by: ishandhanani <ishandhanani@gmail.com>
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants