Repository navigation
[AutoDeploy]: Qwen-3.5 400B perf optimization聽#11548
Description
馃殌 The feature, motivation and pitch
Follow up on #11440
BS=512, 1000/1000, B200, TP=8
Output TPS ~0.5x, TTFT is ~0.2x compared to vLLM.
Targeting >0.8x as initial target.
Dump traces and identify performance optimizations.
Alternatives
No response
Additional context
No response
Before submitting a new issue...
- Make sure you already searched for relevant issues, and checked the documentation and examples for answers to frequently asked questions.
Activity
taylor-yb-lee commentedon Feb 25, 2026
https://docs.google.com/document/d/1swpR_MhUknXsQ6d5waVQengFKpdGBeSpvkI1kKKYv94/edit?tab=t.0 Updating daily status & progress here
[#11548][feat] AutoDeploy: Optimize Qwen3.5 perf (#12265)