摘要
随着多核处理器日渐普及,开发高效易用的并行编程模型成为新的挑战.MapReduce是Google开发的一种并行分布式计算模型,在其搜索业务中获得了巨大的成功.将MapReduce模型引入科学计算领域,并结合实例阐述了如何使用面向高性能计算的HPMR/HPMR-s系统在分布式或共享存储系统中采用统一的方式描述并实现并行科学计算.
With multi-core processors becoming more popular, developing high efficient and easy-to-use parallel programruing model poses brand new challenges. MapReduce, developed by Google, is a parallel and distributed computing model, which has been successfully applied to Google＇ s search engine application. This paper introduces MapReduce Model into scientific computing, illustrating some issues of using HPMR/HPMR-s to describe and implement scientific parallel computing in a uniform style under distributed or shared-memory systems.
出处
《微电子学与计算机》
CSCD
北大核心
2009年第8期13-17,共5页
Microelectronics & Computer
基金
国家自然科学基金项目(60533020)
安徽省自然科学基金项目(090412068)
