“No-Compile-Time Knowledge” distribution of finite element computations on multiprocessors

Author

Erhel, J. ; Hahad, M. ; Priol, T.

Author_Institution

Campus de Beaulieu, INRIA, Rennes, France

Volume

1

fYear

1996

fDate

3-6 Jan 1996

Firstpage

614

Abstract

This paper addresses partitioning and scheduling of irregular loops arising in finite element computations on unstructured meshes. Target computers are distributed memory parallel computers that provide a global address space. We introduce the concept of “conditioned iterations loop” which distributes the iterations dynamically according to a runtime condition. This technique is improved by a learning approach. It is also combined with a renumbering of the mesh to improve data locality. Execution times on KSR1 with up to 64 processors on a model loop show good performances

Keywords

distributed memory systems; finite element analysis; mathematics computing; parallel programming; processor scheduling; program compilers; program control structures; resource allocation; software performance evaluation; KSR1; No-Compile-Time Knowledge; conditioned iterations loop; data locality; distributed memory parallel computers; execution times; finite element computations; global address space; irregular loops; learning; mesh renumbering; multiprocessors; partitioning; runtime condition; scheduling; unstructured meshes; Computational modeling; Concurrent computing; Costs; Distributed computing; Finite element methods; Lead; Parallel programming; Processor scheduling; Runtime; Writing;

fLanguage

English

Publisher

ieee

Conference_Titel

System Sciences, 1996., Proceedings of the Twenty-Ninth Hawaii International Conference on ,

Conference_Location

Wailea, HI

Print_ISBN

0-8186-7324-9

Type

conf

DOI

10.1109/HICSS.1996.495512

Filename

495512