A method for programming matrix algorithms in array processors

[Proceedings] COMPEURO 88 - System Design: Concepts, Methods and Tools Pub Date : 1988-04-11 DOI:10.1109/CMPEUR.1988.4958

G. M. Bisio, F. Arduini, F. Curatelli, E. Zitti, G. Parodi, A. Zini

引用次数: 0

Abstract

A method for mapping uniformly distributed (wavefront like) algorithms into processor square meshes of fixed size, which is based on the concept of folding, is applied to matrix computation (matrix multiplication and LU decomposition). Array processors programmed in Occam are considered. The programming methodology, which can be automated, is analyzed in detail.<>

查看原文本刊更多论文

阵列处理器中矩阵算法的编程方法

将基于折叠概念的均匀分布(类波前)算法映射到固定大小的处理器方形网格的方法应用于矩阵计算(矩阵乘法和LU分解)。考虑用Occam编程的阵列处理器。详细分析了可实现自动化的编程方法。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

[Proceedings] COMPEURO 88 - System Design: Concepts, Methods and Tools

自引率

0.00%

发文量