数据并行计算:概念、模型与系统  

Data Parallel Computing: Concept ,Models, and Systems

作　　者：李晓明[1] 

机构地区：[1]北京大学计算机科学技术系

出　　处：《计算机科学》2000年第6期1-5,共5页Computer Science

基　　金：国家自然科学基金69873004;863高技术计划863-306-ZT01-05-1资助

摘　　要：一、引言并行计算,或者并行处理,指的是这样一种努力和相关的研究:利用多个具有计算能力的部件来共同完成一个计算工作,以获得比用一个部件来完成要快的效果。这显然是一个很自然的想法。历史地看,几乎是自从有了计算机,就有了并行处理的想法和实践。在80年代后期到90年代初期。From the perspective of application developers, comparing a few programming styles o example, this article describes the concept and programming models of data parallel computing, a as ideas and implications of runtime support. Data parallel computing is referred to such a class of a] cation of computing process: the same or similar operations are performed on each elements of the a cation data set. The main and the most natural programming model is SPMD (Single Program Multiple Data) for this kind of application on distributed memory systems. And SPMD programming model needs runtime support in order to obtain a good trade-off between execution efficiency and ease of programming.

关 键 词：数据并行计算 算法 并行计算机 数学模型 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]