一种适应多核的流水并行编程模式  

Pipeline Paralleling Model Based on Multicorer

作　　者：郭惠芳[1] 赵荣彩[1] 

机构地区：[1]解放军信息工程大学,河南郑州450002

出　　处：《小型微型计算机系统》2010年第9期1918-1920,共3页Journal of Chinese Computer Systems

基　　金：国家"八六三"高技术研究发展计划基金项目(2006AA01Z408)资助

摘　　要：随着多核处理器的日益普及,存储墙问题的严重性被越来越多的人所认识.所以,相比对称多处理器,充分利用片上共享的高速缓存对于多核来说,显得尤为迫切.本文分析了传统迭代空间划分的并行模式的局限性,用线程级流水并行模式重构了红-黑算法,使消费者线程能够及时重用生产者线程之前调入片内的数据,降低了高速缓存的失效率,同时减少了对片外主存的访问量.我们的实验表明红-黑算法的线程级流水并行模式比串行模式在core2上最多可得到40%的性能提高.Today,with the increasing popularity of chip multiple processors（CMP）,the severity of memory wall problem are known by more and more people.So,to make better use of the shared cache on chip is more necessary on CMP than other multiple processors architecture.In this paper,we analyze the performance of traditional spatial decomposed parallel implement of red-black algorithm,and find that this parallel model does not exploit the temporary data locality of this application.Then,we restructure red-black algorithm to be a producer-consumer thread pipeline.Under this thread-level pipeline model,consumer threads can reuse the data which former producers have fetched into the shared cache.Then the number of cache miss reduces.Our experiment results show the application performance under the thread-level pipeline parallel model achieves an additional improvement of about 40% on core 2.

关 键 词：片上多处理器 对称处理器 流水并行模式 迭代空间划分 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]