Reducing variations in parallel efficiency for unstructured grid computations

Author

Wörner, B. ; Geuder, U. ; Härdtner, M. ; Zink, R.

Author_Institution

Inst. of Parallel & Distributed High-Perfomance Syst., Stuttgart Univ., Germany

fYear

1994

fDate

2-5 Aug 1994

Firstpage

287

Lastpage

294

Abstract

We present the GRIDS programming system for parallel computations on unstructured grids. The need for adaptive parallelization is discussed. It takes into account the specific reasons for the degradation of parallel efficiency in parallel systems ranging from workstation clusters to parallel supercomputers. Implications are shown on the design of programming models for unified parallel and distributed computing. Some strategies for adaptive optimization that are integrated in the GRIDS system are presented in detail. The effects of balancing the load bundling messages, and dynamically reordering the operations are analysed in the general context of computations on unstructured grids. Performance measurements show the impact on the parallel efficiency of these schemes on a workstation cluster and a parallel computer

Keywords

multiprocessing programs; parallel programming; programming environments; GRIDS programming system; adaptive parallelization; load bundling messages; parallel computations; parallel computer; parallel efficiency; parallel supercomputers; performance measurements; unstructured grid computations; workstation cluster; workstation clusters; Computer architecture; Concurrent computing; Distributed computing; Grid computing; Parallel machines; Parallel processing; Parallel programming; Scalability; Supercomputers; Workstations;

fLanguage

English

Publisher

ieee

Conference_Titel

High Performance Distributed Computing, 1994., Proceedings of the Third IEEE International Symposium on

Conference_Location

San Francisco, CA

Print_ISBN

0-8186-6395-2

Type

conf

DOI

10.1109/HPDC.1994.340233

Filename

340233