Compiler techniques for effective communication on distributed-memory multiprocessors

Author

Navarro, Angeles G. ; Paek, Yunheung ; Zapata, Emilio L. ; Padua, David

Author_Institution

Dept. of Comput. Archit., Malaga Univ., Spain

fYear

1997

fDate

11-15 Aug 1997

Firstpage

74

Lastpage

77

Abstract

The Polaris restructurer transforms conventional Fortran programs into parallel form for various types of multiprocessor systems. This paper presents the results of a study on strategies to improve the effectiveness of Polaris´ techniques for distributed-memory multiprocessors. Our study, which is based on the hand analysis of MDG and TRFD from the Perfect Benchmarks and TOYCATV and SWIM from SPEC benchmarks, identified three techniques that are important for improving communication optimization. Their application produces almost perfect speedups for the four programs on the Cray T3D

Keywords

distributed memory systems; parallelising compilers; performance evaluation; Fortran programs; MDG; Perfect Benchmarks; Polaris restructurer; SPEC benchmarks; SWIM; TOYCATV; TRFD; almost perfect speedups; compiler techniques; distributed-memory multiprocessors; effective communication; Communication effectiveness; Computer architecture; Computer science; Data analysis; Distribution strategy; Parallel processing; Phase detection; Phased arrays; Polarization; Privatization;

fLanguage

English

Publisher

ieee

Conference_Titel

Parallel Processing, 1997., Proceedings of the 1997 International Conference on

Conference_Location

Bloomington, IL

ISSN

0190-3918

Print_ISBN

0-8186-8108-X

Type

conf

DOI

10.1109/ICPP.1997.622559

Filename

622559