MergeComp: A Compression Scheduler for Scalable Communication-Efficient Distributed Training

Autor:	Wang, Zhuang, Wu, Xinyu, Ng, T. S. Eugene
Rok vydání:	2021
Předmět:	Computer Science - Distributed Parallel and Cluster Computing Computer Science - Machine Learning
Druh dokumentu:	Working Paper
Popis:	Large-scale distributed training is increasingly becoming communication bound. Many gradient compression algorithms have been proposed to reduce the communication overhead and improve scalability. However, it has been observed that in some cases gradient compression may even harm the performance of distributed training. In this paper, we propose MergeComp, a compression scheduler to optimize the scalability of communication-efficient distributed training. It automatically schedules the compression operations to optimize the performance of compression algorithms without the knowledge of model architectures or system parameters. We have applied MergeComp to nine popular compression algorithms. Our evaluations show that MergeComp can improve the performance of compression algorithms by up to 3.83x without losing accuracy. It can even achieve a scaling factor of distributed training up to 99% over high-speed networks. Comment: 8 papes
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2103.15195 Zobrazit plný text záznamu View this record from Arxiv