Lyra: Elastic Scheduling for Deep Learning Clusters
Author:
Affiliation:
1. City University of Hong Kong, Hong Kong, Hong Kong
2. The Chinese University of Hong Kong, Hong Kong, Hong Kong
3. Google, Kirkland, United States of America
4. ByteDance Inc., Beijing, China
5. Non affiliated, Bellevue, United States of America
Funder
Research Grants Council of Hong Kong
Chinese University of Hong Kong
Publisher
ACM
Link
https://dl.acm.org/doi/pdf/10.1145/3552326.3587445
Reference60 articles.
1. Efficient Large Message Broadcast using NCCL and CUDA-Aware MPI for Deep Learning
2. Zhihao Bai , Zhen Zhang , Yibo Zhu , and Xin Jin . 2020 . PipeSwitch: Fast Pipelined Context Switching for Deep Learning Applications . In Proc. USENIX OSDI. Zhihao Bai, Zhen Zhang, Yibo Zhu, and Xin Jin. 2020. PipeSwitch: Fast Pipelined Context Switching for Deep Learning Applications. In Proc. USENIX OSDI.
3. Balancing efficiency and fairness in heterogeneous GPU clusters for deep learning
4. Semi-dynamic load balancing
5. Trishul Chilimbi , Yutaka Suzue , Johnson Apacible , and Karthik Kalyanaraman . 2014 . Project adam: Building an efficient and scalable deep learning training system . In Proc. USENIX OSDI. Trishul Chilimbi, Yutaka Suzue, Johnson Apacible, and Karthik Kalyanaraman. 2014. Project adam: Building an efficient and scalable deep learning training system. In Proc. USENIX OSDI.
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. Application-Oriented Cloud Workload Prediction: A Survey and New Perspectives;Tsinghua Science and Technology;2025-02
2. Latency-Guaranteed Co-Location of Inference and Training for Reducing Data Center Expenses;2024 IEEE 44th International Conference on Distributed Computing Systems (ICDCS);2024-07-23
3. Enhanced Scheduling of AI Applications in Multi-Tenant Cloud Using Genetic Optimizations;Applied Sciences;2024-05-29
4. Heet: Accelerating Elastic Training in Heterogeneous Deep Learning Clusters;Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2;2024-04-27
5. Deferred Continuous Batching in Resource-Efficient Large Language Model Serving;Proceedings of the 4th Workshop on Machine Learning and Systems;2024-04-22
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