Accelerating Distributed Machine Learning in Disaggregated Architectures with Flexible Optically Interconnected Computing Resources

Shijia Yan,Ziyi Zhu,Madeleine S. Glick,Zhenguo Wu,Keren Bergman

2022 OPTICAL FIBER COMMUNICATIONS CONFERENCE AND EXHIBITION (OFC)（2022）

引用 1|浏览14

暂无评分

摘要

We introduce an optically interconnected disaggregated architecture for GPU resources and demonstrate a 3 x increase in GPU utilization and up to 73.2% acceleration of application runtime for distributed machine learning workloads. (C) 2022 The Author(s)

查看译文

关键词

GPU resources,GPU utilization,distributed machine learning workloads,optically interconnected disaggregated architecture,optically interconnected computing resources

AI 理解论文

溯源树

样例

生成溯源树，研究论文发展脉络

Chat Paper

正在生成论文摘要