Skip to content

Update Qwen3.5 B200 NVFP4 MTP config - #34357

Merged
ishandhanani merged 1 commit into
sgl-project:mainfrom
faradawn:qwen35-b200-fp4-mtp-tep2
Aug 11, 2026
Merged

ishandhanani merged 1 commit into
sgl-project:mainfrom
faradawn:qwen35-b200-fp4-mtp-tep2

Conversation

@faradawn

@faradawn faradawn commented Aug 11, 2026

Copy link
Copy Markdown
Contributor

Switch the Qwen3.5-397B-A17B NVFP4 B200 recipe to tp=2 with expert parallelism 2 when MTP is enabled, which outperforms the current tp=4 setting across the concurrency sweep. Best configs based on SemiAnalysisAI/InferenceX#2550.


CI States

Latest PR Test (Base): ✅ Run #31458614193
Latest PR Test (Extra): ❌ Run #31458614104

@github-actions github-actions Bot added the documentation Improvements or additions to documentation label Aug 11, 2026
@faradawn

Copy link
Copy Markdown
Contributor Author

Hi @JustinTong0323 @ishandhanani can you help comment/merge this? It has been tested as in the description

@ishandhanani
ishandhanani merged commit 857910b into sgl-project:main Aug 11, 2026
93 of 97 checks passed
vstone-w pushed a commit to ClownBin/sglang that referenced this pull request Aug 12, 2026
saturn-acc pushed a commit to saturn-acc/sglang that referenced this pull request Aug 16, 2026
Atituiset pushed a commit to Atituiset/sglang that referenced this pull request Sep 10, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

documentation Improvements or additions to documentation

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants