Mercury: A Simple Transport Layer Scheduler to Accelerate Distributed DNN Training
Author:
Affiliation:
1. School of Information Science and Technology, Fudan University,Shanghai,China
2. Huawei Technologies Co. Ltd.,Shenzhen,China
3. School of Computer Science and Technology, Fudan University,Shanghai,China
Funder
Research and Development
Publisher
IEEE
Link
http://xplorestaging.ieee.org/ielx7/9796607/9796652/09796820.pdf?arnumber=9796820
Reference36 articles.
1. Motivating in-network aggregation for distributed deep neural network training;luo;Workshop on Approximate Computing Across the System Stack,2017
2. Hogwild!: A lock-free approach to parallelizing stochastic gradient descent;niu;arXiv preprint arXiv 1106 5730,2011
3. Round-Robin Synchronization: Mitigating Communication Bottlenecks in Parameter Servers
4. More effective distributed ML via a stale synchronous parallel parameter server;ho;Advances in neural information processing systems,2013
5. ATP: In-network aggregation for multi-tenant learning;lao;USENIX NSDI,2021
Cited by 8 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. Host-driven In-Network Aggregation on RDMA;IEEE INFOCOM 2024 - IEEE Conference on Computer Communications;2024-05-20
2. An efficient bandwidth-adaptive gradient compression algorithm for distributed training of deep neural networks;Journal of Systems Architecture;2024-05
3. US-Byte: An Efficient Communication Framework for Scheduling Unequal-Sized Tensor Blocks in Distributed Deep Learning;IEEE Transactions on Parallel and Distributed Systems;2024-01
4. OF-WFBP: A near-optimal communication mechanism for tensor fusion in distributed deep learning;Parallel Computing;2023-11
5. Prophet: Fine-grained Load Balancing for Parallel Training of Large-scale MoE Models;2023 IEEE International Conference on Cluster Computing (CLUSTER);2023-10-31
1.学者识别学者识别
2.学术分析学术分析
3.人才评估人才评估
"同舟云学术"是以全球学者为主线，采集、加工和组织学术论文而形成的新型学术文献查询和分析系统，可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容，当前同舟云学术共收录了国内外主流学术期刊6万余种，收集的期刊论文及会议论文总量共计约1.5亿篇，并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询！咨询电话：010-8811{复制后删除}0370
www.globalauthorid.com
TOP
Copyright © 2019-2024 北京同舟云网络信息技术有限公司 京公网安备11010802033243号 京ICP备18003416号-3