Lei Shi, Jun Pang, Lei Yang, Tiejun Zhang, Donghui Wang
{"title":"Fair-Priority-Expression-Based burst scheduling to enhance performance and fairness of shared dram systems","authors":"Lei Shi, Jun Pang, Lei Yang, Tiejun Zhang, Donghui Wang","doi":"10.1109/CCECE.2009.5090118","DOIUrl":null,"url":null,"abstract":"Many algorithms ofmemory access scheduling have been studied to attack the well-known memory-wall. However, some of them introduce unfairness problem in Chip-Multi-Processor (CMP) systems. Therefore a new approach called Fair-Priority-Expression-Based (FairPEB) burst scheduling is proposed in this paper to address both performance and unfairness problems. Experiment results show that this method improves data bus utilization of memory system by 15.4% and 31.87% over conventional First-Ready First-Come-First-Service (FR-FCFS) and original burst scheduling. Furthermore it keeps memory accesses of different applications fair.","PeriodicalId":153464,"journal":{"name":"2009 Canadian Conference on Electrical and Computer Engineering","volume":"93 3 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2009-05-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2009 Canadian Conference on Electrical and Computer Engineering","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/CCECE.2009.5090118","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
Many algorithms ofmemory access scheduling have been studied to attack the well-known memory-wall. However, some of them introduce unfairness problem in Chip-Multi-Processor (CMP) systems. Therefore a new approach called Fair-Priority-Expression-Based (FairPEB) burst scheduling is proposed in this paper to address both performance and unfairness problems. Experiment results show that this method improves data bus utilization of memory system by 15.4% and 31.87% over conventional First-Ready First-Come-First-Service (FR-FCFS) and original burst scheduling. Furthermore it keeps memory accesses of different applications fair.