A Multidimensional Communication Scheduling Method for Hybrid Parallel DNN Training

IF 5.6 2区 计算机科学 Q1 COMPUTER SCIENCE, THEORY & METHODS IEEE Transactions on Parallel and Distributed Systems Pub Date : 2024-03-28 DOI:10.1109/TPDS.2024.3406420
Shengwei Li;Kai Lu;Zhiquan Lai;Weijie Liu;Keshi Ge;Dongsheng Li
{"title":"A Multidimensional Communication Scheduling Method for Hybrid Parallel DNN Training","authors":"Shengwei Li;Kai Lu;Zhiquan Lai;Weijie Liu;Keshi Ge;Dongsheng Li","doi":"10.1109/TPDS.2024.3406420","DOIUrl":null,"url":null,"abstract":"The transformer-based deep neural network (DNN) models have shown considerable success across diverse tasks, prompting widespread adoption of distributed training methods such as data parallelism and pipeline parallelism. With the increasing parameter number, hybrid parallel training becomes imperative to scale training. The primary bottleneck in scaling remains the communication overhead. The communication scheduling technique, emphasizing the overlap of communication with computation, has demonstrated its benefits in scaling. However, most existing works focus on data parallelism, overlooking the nuances of hybrid parallel training. In this paper, we propose \n<monospace>TriRace</monospace>\n, an efficient communication scheduling framework for accelerating communications in hybrid parallel training of asynchronous pipeline parallelism and data parallelism. To achieve effective computation-communication overlap, \n<monospace>TriRace</monospace>\n introduces \n<italic>3D communication scheduling</i>\n, which adeptly leverages data dependencies between communication and computations, efficiently scheduling AllReduce communication, sparse communication, and peer-to-peer communication in hybrid parallel training. To avoid possible communication contentions, \n<monospace>TriRace</monospace>\n also incorporates a \n<italic>topology-aware runtime</i>\n which optimizes the execution of communication operations by considering ongoing communication operations and real-time network status. We have implemented a prototype of \n<monospace>TriRace</monospace>\n based on PyTorch and Pipedream-2BW, and conducted comprehensive evaluations with three representative baselines. Experimental results show that \n<monospace>TriRace</monospace>\n achieves up to 1.07–1.45× speedup compared to the state-of-the-art pipeline parallelism training baseline Pipedream-2BW, and 1.24–1.81× speedup compared to the Megatron.","PeriodicalId":13257,"journal":{"name":"IEEE Transactions on Parallel and Distributed Systems","volume":null,"pages":null},"PeriodicalIF":5.6000,"publicationDate":"2024-03-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE Transactions on Parallel and Distributed Systems","FirstCategoryId":"94","ListUrlMain":"https://ieeexplore.ieee.org/document/10540313/","RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, THEORY & METHODS","Score":null,"Total":0}
引用次数: 0

Abstract

The transformer-based deep neural network (DNN) models have shown considerable success across diverse tasks, prompting widespread adoption of distributed training methods such as data parallelism and pipeline parallelism. With the increasing parameter number, hybrid parallel training becomes imperative to scale training. The primary bottleneck in scaling remains the communication overhead. The communication scheduling technique, emphasizing the overlap of communication with computation, has demonstrated its benefits in scaling. However, most existing works focus on data parallelism, overlooking the nuances of hybrid parallel training. In this paper, we propose TriRace , an efficient communication scheduling framework for accelerating communications in hybrid parallel training of asynchronous pipeline parallelism and data parallelism. To achieve effective computation-communication overlap, TriRace introduces 3D communication scheduling , which adeptly leverages data dependencies between communication and computations, efficiently scheduling AllReduce communication, sparse communication, and peer-to-peer communication in hybrid parallel training. To avoid possible communication contentions, TriRace also incorporates a topology-aware runtime which optimizes the execution of communication operations by considering ongoing communication operations and real-time network status. We have implemented a prototype of TriRace based on PyTorch and Pipedream-2BW, and conducted comprehensive evaluations with three representative baselines. Experimental results show that TriRace achieves up to 1.07–1.45× speedup compared to the state-of-the-art pipeline parallelism training baseline Pipedream-2BW, and 1.24–1.81× speedup compared to the Megatron.
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
混合并行 DNN 训练的多维通信调度方法
基于变压器的深度神经网络(DNN)模型在各种任务中取得了相当大的成功,促使分布式训练方法(如数据并行和管道并行)得到广泛采用。随着参数数量的增加,混合并行训练成为扩展训练的当务之急。扩展的主要瓶颈仍然是通信开销。通信调度技术强调通信与计算的重叠,已经证明了其在扩展方面的优势。然而,现有的大多数工作都侧重于数据并行性,忽略了混合并行训练的细微差别。在本文中,我们提出了一个高效的通信调度框架 TriRace,用于在异步流水线并行和数据并行的混合并行训练中加速通信。为了实现有效的计算-通信重叠,TriRace 引入了三维通信调度,善于利用通信与计算之间的数据依赖关系,在混合并行训练中高效调度 AllReduce 通信、稀疏通信和点对点通信。为了避免可能出现的通信争议,TriRace 还采用了拓扑感知运行时,通过考虑正在进行的通信操作和实时网络状态来优化通信操作的执行。我们基于 PyTorch 和 Pipedream-2BW 实现了 TriRace 的原型,并用三个具有代表性的基线进行了综合评估。实验结果表明,与最先进的流水线并行性训练基线 Pipedream-2BW 相比,TriRace 实现了高达 1.07-1.45 倍的速度提升,与 Megatron 相比,实现了 1.24-1.81 倍的速度提升。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 去求助
来源期刊
IEEE Transactions on Parallel and Distributed Systems
IEEE Transactions on Parallel and Distributed Systems 工程技术-工程:电子与电气
CiteScore
11.00
自引率
9.40%
发文量
281
审稿时长
5.6 months
期刊介绍: IEEE Transactions on Parallel and Distributed Systems (TPDS) is published monthly. It publishes a range of papers, comments on previously published papers, and survey articles that deal with the parallel and distributed systems research areas of current importance to our readers. Particular areas of interest include, but are not limited to: a) Parallel and distributed algorithms, focusing on topics such as: models of computation; numerical, combinatorial, and data-intensive parallel algorithms, scalability of algorithms and data structures for parallel and distributed systems, communication and synchronization protocols, network algorithms, scheduling, and load balancing. b) Applications of parallel and distributed computing, including computational and data-enabled science and engineering, big data applications, parallel crowd sourcing, large-scale social network analysis, management of big data, cloud and grid computing, scientific and biomedical applications, mobile computing, and cyber-physical systems. c) Parallel and distributed architectures, including architectures for instruction-level and thread-level parallelism; design, analysis, implementation, fault resilience and performance measurements of multiple-processor systems; multicore processors, heterogeneous many-core systems; petascale and exascale systems designs; novel big data architectures; special purpose architectures, including graphics processors, signal processors, network processors, media accelerators, and other special purpose processors and accelerators; impact of technology on architecture; network and interconnect architectures; parallel I/O and storage systems; architecture of the memory hierarchy; power-efficient and green computing architectures; dependable architectures; and performance modeling and evaluation. d) Parallel and distributed software, including parallel and multicore programming languages and compilers, runtime systems, operating systems, Internet computing and web services, resource management including green computing, middleware for grids, clouds, and data centers, libraries, performance modeling and evaluation, parallel programming paradigms, and programming environments and tools.
期刊最新文献
DyLaClass: Dynamic Labeling Based Classification for Optimal Sparse Matrix Format Selection in Accelerating SpMV Design and Performance Evaluation of Linearly Extensible Cube-Triangle Network for Multicore Systems PeakFS: An Ultra-High Performance Parallel File System via Computing-Network-Storage Co-Optimization for HPC Applications Mitosis: A Scalable Sharding System Featuring Multiple Dynamic Relay Chains Breaking the Memory Wall for Heterogeneous Federated Learning via Model Splitting
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1