DSM中基于遗传算法的页迁移  

A Dynamic Page Migration Mechanism Based on Genetic Algorithm in Distributed Shared Memory

作　　者：陈家琪[1] 肖梁[1] 

机构地区：[1]上海理工大学计算机工程学院,上海200093

出　　处：《计算机科学》2007年第7期110-112,192,共4页Computer Science

摘　　要：在分布式共享存储系统(Distributed Shared Memory,DSM)中,远程数据不命中时间和同步时间是两个最主要的开销。本文在分析页迁移技术以及分布式系统性能开销的基础上,提出基于遗传算法的动态页迁移技术以提高数据的局部性。实验结果表明,该方法能在一定程度上提高数据的局部性,且适用于规模较大的程序。The most important cost of time on Distributed Shared Memory is non-hit in targe time and synchronization time of remote data. Based on the analysis of page migration technology and distributed system performance cost , we represented dynamic page migration in base of genetic algorithm to improve the location of data. The results of experiment indicate that this method can improve the location of data in a certain extent and is also applicable to large-scale programs.

关 键 词：分布式共享存储 页迁移 遗传算法 

分 类 号：TP316[自动化与计算机技术—计算机软件与理论]