科研速览 · Science Skim继续刷下去 · Keep skimming →
◇ arXiv2026-09-17· cs.AR

A Multi-Engine Dataflow for MoE Decoding on Scratchpad-Based Tensor Accelerators

Bin Ma, Wenjie Fan, Dong Li

原始摘要(英文原文)· Original abstract
Mixture-of-Experts (MoE) decoding on scratchpad-based tensor accelerators (STA) is dominated by moving expert weights while the compute engines sit idle. This traffic is hard to hide, because the experts are known only after routing, and hard to shrink without losing quality or adding critical-path work. We present CARDAN, which represents each expert-weight matrix as a vector-quantized component plus a shared-basis low-rank component and co-designs this representation with a multi-engine decoding dataflow. The representation separates expert-common from expert-private work, so the dataflow overlaps DMA with computation on several engines. Across five MoE families on AWS Trainium3, CARDAN matches or improves BF16-teacher perplexity across all five models and speeds up batch-one decoding by 1.15-1.31x over AWS dense MoE megakernels, rising to 1.7x at batch size 16.
读原文 · Read the paper ↗

AI 追问PRO

登录后使用 AI 追问

讨论区

登录后参与讨论

相关论文 · Related

A Multi-Engine Dataflow for MoE Decoding on Scratchpad-Based Tensor Accelerators — 科研速览 Science Skim