Skip to content

minor: add dataset dump and questions shuffle - #2093

Merged
merrymercy merged 1 commit into
mainfrom
zhyncs/bench
Nov 19, 2024
Merged

minor: add dataset dump and questions shuffle#2093
merrymercy merged 1 commit into
mainfrom
zhyncs/bench

Conversation

@zhyncs

@zhyncs zhyncs commented Nov 19, 2024

Copy link
Copy Markdown
Contributor

Motivation

Modifications

Checklist

  • Format your code according to the Contributor Guide.
  • Add unit tests as outlined in the Contributor Guide.
  • Update documentation as needed, including docstrings or example tutorials.

@zhyncs
zhyncs requested a review from ByronHsu November 19, 2024 16:12
@merrymercy
merrymercy merged commit 55bd97f into main Nov 19, 2024
@merrymercy
merrymercy deleted the zhyncs/bench branch November 19, 2024 22:07
timethink pushed a commit to timethink/sglang that referenced this pull request Mar 9, 2025
@zhyncs
minor: add dataset dump and questions shuffle (sgl-project#2093)
f9410b5
ChangLiu0709 added a commit to ChangLiu0709/sglang that referenced this pull request Jul 9, 2026
@ChangLiu0709
cookbook(deepseek-v4): MI355X updates from InferenceX PRs sgl-project…
01c5c03 
…#2093/sgl-project#2108

Incorporates the following changes from InferenceX benchmark recipes:

- Bump MI355X docker image to v0.5.14-rocm720-mi35x-20260708
- Update swa-full-tokens-ratio from 0.1 to 0.15 across all MI355X cells
- Add DP-attention env vars for MI355X balanced/high-throughput cells:
  SGLANG_SHARED_EXPERT_TP1=1, SGLANG_DP_SHARED_EXPERT_LOCAL=1,
  SGLANG_DP_USE_REDUCE_SCATTER=1
- Replace --enable-prefill-delayer + --prefill-delayer-max-delay-ms 5000
  with --enable-two-batch-overlap for all MI355X DP-attention cells

Refs: SemiAnalysisAI/InferenceX#2093, SemiAnalysisAI/InferenceX#2108
@ChangLiu0709 ChangLiu0709 mentioned this pull request Jul 9, 2026
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@ByronHsu ByronHsu Awaiting requested review from ByronHsu

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@zhyncs @merrymercy