{"title":"Phoenix rebirth: Scalable MapReduce on a large-scale shared-memory system","authors":"Richard M. Yoo, Anthony Romano, C. Kozyrakis","doi":"10.1109/IISWC.2009.5306783","DOIUrl":null,"url":null,"abstract":"Dynamic runtimes can simplify parallel programming by automatically managing concurrency and locality without further burdening the programmer. Nevertheless, implementing such runtime systems for large-scale, shared-memory systems can be challenging. This work optimizes Phoenix, a MapReduce runtime for shared-memory multi-cores and multiprocessors, on a quad-chip, 32-core, 256-thread UltraSPARC T2+ system with NUMA characteristics. We show how a multi-layered approach that comprises optimizations on the algorithm, implementation, and OS interaction leads to significant speedup improvements with 256 threads (average of 2.5× higher speedup, maximum of 19×). We also identify the roadblocks that limit the scalability of parallel runtimes on shared-memory systems, which are inherently tied to the OS scalability on large-scale systems.","PeriodicalId":387816,"journal":{"name":"2009 IEEE International Symposium on Workload Characterization (IISWC)","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2009-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"266","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2009 IEEE International Symposium on Workload Characterization (IISWC)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/IISWC.2009.5306783","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 266
Abstract
Dynamic runtimes can simplify parallel programming by automatically managing concurrency and locality without further burdening the programmer. Nevertheless, implementing such runtime systems for large-scale, shared-memory systems can be challenging. This work optimizes Phoenix, a MapReduce runtime for shared-memory multi-cores and multiprocessors, on a quad-chip, 32-core, 256-thread UltraSPARC T2+ system with NUMA characteristics. We show how a multi-layered approach that comprises optimizations on the algorithm, implementation, and OS interaction leads to significant speedup improvements with 256 threads (average of 2.5× higher speedup, maximum of 19×). We also identify the roadblocks that limit the scalability of parallel runtimes on shared-memory systems, which are inherently tied to the OS scalability on large-scale systems.