A localized dynamic load balancing strategy for highly parallel systems

[1990 Proceedings] The Third Symposium on the Frontiers of Massively Parallel Computation Pub Date : 1990-10-08 DOI:10.1109/FMPC.1990.89487

M. Willebeek-LeMair, A. Reeves

{"title":"A localized dynamic load balancing strategy for highly parallel systems","authors":"M. Willebeek-LeMair, A. Reeves","doi":"10.1109/FMPC.1990.89487","DOIUrl":null,"url":null,"abstract":"Two dynamic load-balancing strategies, a local diffusion (RID) and a global exchange (DEM) strategy, designed to support massively parallel systems are presented and compared. The effects of system size and task granularity are studied. Both strategies are implemented on a 32-processor iPSC/2 and a 256-processor IBM Victor. Even for low degrees of parallelism the performance of the DEM and RID strategies is very similar. The efficiency of the DEM strategy, however, depends heavily on the system interconnection topology. Furthermore, the system sizes tested were small in the context of massively parallel systems. The overhead costs of synchronization (scale as O(N)) for the DEM approach may cause a serious deterioration of performance. The RID strategy is easily embedded into simpler topologies, and can scale gracefully for larger systems. Finally, the RID scheme is able to maintain task locality, supporting a wider variety of applications that exhibit local communication dependencies between tasks. Therefore, the RID strategy may offer a superior performance when locality is important.<<ETX>>","PeriodicalId":193332,"journal":{"name":"[1990 Proceedings] The Third Symposium on the Frontiers of Massively Parallel Computation","volume":"9 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1990-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"5","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"[1990 Proceedings] The Third Symposium on the Frontiers of Massively Parallel Computation","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/FMPC.1990.89487","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}

引用次数: 5

Abstract

Two dynamic load-balancing strategies, a local diffusion (RID) and a global exchange (DEM) strategy, designed to support massively parallel systems are presented and compared. The effects of system size and task granularity are studied. Both strategies are implemented on a 32-processor iPSC/2 and a 256-processor IBM Victor. Even for low degrees of parallelism the performance of the DEM and RID strategies is very similar. The efficiency of the DEM strategy, however, depends heavily on the system interconnection topology. Furthermore, the system sizes tested were small in the context of massively parallel systems. The overhead costs of synchronization (scale as O(N)) for the DEM approach may cause a serious deterioration of performance. The RID strategy is easily embedded into simpler topologies, and can scale gracefully for larger systems. Finally, the RID scheme is able to maintain task locality, supporting a wider variety of applications that exhibit local communication dependencies between tasks. Therefore, the RID strategy may offer a superior performance when locality is important.<>

查看原文

微信好友朋友圈 QQ好友复制链接

本刊更多论文

高度并行系统的局部动态负载平衡策略

提出并比较了两种支持大规模并行系统的动态负载平衡策略:局部扩散(RID)和全局交换(DEM)策略。研究了系统大小和任务粒度的影响。这两种策略都在32个处理器的iPSC/2和256个处理器的IBM Victor上实现。即使在低并行度的情况下，DEM和RID策略的性能也非常相似。然而，DEM策略的效率很大程度上取决于系统互连拓扑结构。此外，在大规模并行系统的背景下，测试的系统尺寸很小。DEM方法的同步开销(规模为O(N))可能会导致性能严重恶化。RID策略很容易嵌入到更简单的拓扑中，并且可以优雅地扩展到更大的系统。最后，RID方案能够维护任务局部性，支持在任务之间显示本地通信依赖关系的更广泛的应用程序。因此，当局部性很重要时，RID策略可以提供更好的性能

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

[1990 Proceedings] The Third Symposium on the Frontiers of Massively Parallel Computation

自引率

0.00%

发文量

期刊最新文献

Topological properties of banyan-hypercube networks Designing the 3-LAP (three layers associative processor) for arithmetic and symbolic applications Partitioning on the banyan-hypercube networks Multiple channel architecture Performance prediction-How good is good?