DocumentCode :
998260
Title :
Loop transformation using nonunimodular matrices
Author :
Fernández, Agustín ; Llabería, José M. ; Valero-Garcia, Miguel
Author_Institution :
Dept. d´´Arquitectura de Computadors, Univ. Politecnica de Catalunya, Barcelona, Spain
Volume :
6
Issue :
8
fYear :
1995
fDate :
8/1/1995 12:00:00 AM
Firstpage :
832
Lastpage :
840
Abstract :
Linear transformations are widely used to vectorize and parallelize loops. A subset of these transformations are unimodular transformations. When a unimodular transformation is used, the exact bounds of the transformed loop nest are easily computed and the steps of the loops are equal to 1. Unimodular loop transformations have been widely used since they permit the implementation of many useful loop transformations. Recently, nonunimodular transformations have been proposed to reduce communication requirements or to use the memory hierarchy efficiently. The methods used for unimodular transformations do not work in the case of nonunimodular transformations, since they do not produce the exact bounds of the transformed loop nest. In this paper, we present a method for nested loop transformation which gives the exact bounds for both unimodular and nonunimodular transformations. The basic idea is to use the Hermite Normal Form (HNF) of the transformation matrix
Keywords :
matrix algebra; parallel programming; parallelising compilers; program compilers; program control structures; Hermite normal form; communication requirements; linear transformations; loop transformation; memory hierarchy; nonunimodular matrices; transformation matrix; transformed loop nest; Ear; Helium;
fLanguage :
English
Journal_Title :
Parallel and Distributed Systems, IEEE Transactions on
Publisher :
ieee
ISSN :
1045-9219
Type :
jour
DOI :
10.1109/71.406959
Filename :
406959
Link To Document :
بازگشت