Lotus
-
Published:2021-01-06
Issue:1
Volume:17
Page:1-21
-
ISSN:1550-4832
-
Container-title:ACM Journal on Emerging Technologies in Computing Systems
-
language:en
-
Short-container-title:J. Emerg. Technol. Comput. Syst.
Author:
Lu Yunfeng1,
Gu Huaxi1,
Yu Xiaoshan1,
Chakrabarty Krishnendu2
Affiliation:
1. The State Key Laboratory of Integrated Service Networks, Xidian University, Xi’an, Shaanxi
2. The Department of Electrical and Computer Engineering, Duke University, Durham, NC
Abstract
Machine learning is at the heart of many services provided by data centers. To improve the performance of machine learning, several parameter (gradient) synchronization methods have been proposed in the literature. These synchronization algorithms have different communication characteristics and accordingly place different demands on the network architecture. However, traditional data-center networks cannot easily meet these demands. Therefore, we analyze the communication profiles associated with several common synchronization algorithms and propose a machine learning--oriented network architecture to match their characteristics. The proposed design, named Lotus, because it looks like a lotus flower, is a hybrid optical/electrical architecture based on arrayed waveguide grating routers (AWGRs). In Lotus, a complete bipartite graph is used within the group to improve bisection bandwidth and scalability. Each pair of groups is connected by an optical link, and AWGRs between adjacent groups enhance path diversity and network reliability. We also present an efficient routing algorithm to make full use of the path diversity of Lotus, which leads to a further increase in network performance. Simulation results show that the network performance of Lotus is better than Dragonfly and 3D-Torus under realistic traffic patterns for different synchronization algorithms.
Funder
National Natural Science Foundation of China
Natural Science Foundation of Shaanxi Province for Distinguished Young Scholars
National Key R8D Program of China
The Youth Innovation Team of Shaanxi Universities
Open Project Program of the State Key Laboratory of Mathematical Engineering and Advanced Computing
Fundamental Research Funds for the Central Universities
Publisher
Association for Computing Machinery (ACM)
Subject
Electrical and Electronic Engineering,Hardware and Architecture,Software
Reference42 articles.
1. [Online]. Retrieved from https://images.nvidia.com/content/pdf/dgx1-v100-system-architecture-whitepaper.pdf. [Online]. Retrieved from https://images.nvidia.com/content/pdf/dgx1-v100-system-architecture-whitepaper.pdf.
2. [Online]. Retrieved from https://item.jd.com/10448410875.html. [Online]. Retrieved from https://item.jd.com/10448410875.html.
3. [Online]. Retrieved from http://www.lusterinc.com/. [Online]. Retrieved from http://www.lusterinc.com/.
4. [Online]. Retrieved from https://www.finisar.com/. [Online]. Retrieved from https://www.finisar.com/.
Cited by
3 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献
1. Modoru: Clos nanosecond optical switching for distributed deep training [Invited];Journal of Optical Communications and Networking;2023-12-13
2. OSDL: Dedicated optical slice provisioning in support of distributed deep learning;Computer Networks;2022-09
3. CEFS;Proceedings of the 16th International Conference on emerging Networking EXperiments and Technologies;2020-11-23