| 分布式系统中数据分解代码的自动产生 |
| |
| 引用本文： | 沈亚楠
,姚远
,龚雪容
,张平
,赵荣彩.分布式系统中数据分解代码的自动产生[J].微计算机信息,2005(20). |
| |
| 作者姓名： | 沈亚楠 姚远 龚雪容 张平 赵荣彩 |
| |
| 作者单位： | 河南郑州信息工程大学 450002 |
| |
| 基金项目： | 河士南省杰出人才创新基金,项目编号0521000200 |
| |
| 摘 要： | 为分布内存系统开发的并行编译器碰到的第一个问题就是如何分解一个应用程序中的数据。由于访问非本地节点上数据的代价是昂贵的,所以数据分解必须仔细考虑。尽管数据分解的定义已被提出,但是文献并没有给出相应的算法.本文介绍了在一个已被证明且功能强大的数学模型下如何产生数据分解代码的算法,并在SUIF(Stanforduniversityintermediateformat)系统中的Paraguin编译器上得到实现。
|
| 关 键 词： | 并行编译器 消息传递 不等式多面体系统 数据分解 |
Automatic Code Generation of Data Decomposition in Distributed Shared Memory System |
| |
| Shen,Yanan Yao,Yuan Gong,Xuerong Zhang,Pin Zhao,Rongcai.Automatic Code Generation of Data Decomposition in Distributed Shared Memory System[J].Control & Automation,2005(20). |
| |
| Authors: | Shen Yanan Yao Yuan Gong Xuerong Zhang Pin Zhao Rongcai |
| |
| Affiliation: | (University of Information Engineering,HeNan,Zhengzhou 45002)Shen,Yanan Yao,Yuan Gong,Xuerong Zhang,Pin Zhao,Rongcai |
| |
| Abstract: | The fist problem of developing parallelizing compilerin DSM(Distributed Shared Memory) system is how to decomposethe data of a program.. The non- local data accesses result incommunication that can greatly impact performance, thus themapping or decomposition of data must be carefully considered.Although a concept of data decomposition has been introducedin previous work,the algorithm was not shown. In this paper, analgorithm to decompose data based on a proven and powerfulmathematical model is introduced and is implemented inParaguin compiler of SUIF. |
| |
| Keywords: | Parallelizing compilers Message passing Datadecomposition Linear inequalities |
| 本文献已被 CNKI 等数据库收录！ |
|
