AWS shares MoE kernel optimizations on SageMaker AI
AWS engineers published kernel optimizations for fused Mixture-of-Experts models on Amazon SageMaker AI, including Swap-AB for WGMMA utilization, Triton backend routing for FP8 quantization, and eliminating Triton JIT recompilation. They report up to 10% higher throughput and up to 9% faster Time to First Token on…
First seen 13 Oct, 15:17 UTC on AWS Machine Learning blog1 sourceLast update 6h ago
AmazonResearch · 13 Oct
Research
What the sources say
Linked, never rewritten. Official means the lab itself.
Official1
AWS Machine Learning blog· 13 Oct, 15:17Mixture-of-Experts kernel engineering on Amazon SageMaker AIPress0
No independent coverage yet.
How it unfolded
Every source in the order it appeared. Times in UTC.
Keep up with Amazon
Amazon on the ScoreIts timelineFollow to mark Amazon's stories across the site. No account needed.