Paper#llm🔬 ResearchAnalyzed: Jan 3, 2026 16:33

FUSCO: Faster Data Shuffling for MoE Models

Published:Dec 26, 2025 14:16
1 min read
ArXiv

Analysis

This paper addresses a critical bottleneck in training and inference of large Mixture-of-Experts (MoE) models: inefficient data shuffling. Existing communication libraries struggle with the expert-major data layout inherent in MoE, leading to significant overhead. FUSCO offers a novel solution by fusing data transformation and communication, creating a pipelined engine that efficiently shuffles data along the communication path. This is significant because it directly tackles a performance limitation in a rapidly growing area of AI research (MoE models). The performance improvements demonstrated over existing solutions are substantial, making FUSCO a potentially important contribution to the field.

Reference

FUSCO achieves up to 3.84x and 2.01x speedups over NCCL and DeepEP (the state-of-the-art MoE communication library), respectively.