Dynamic randomized simulation of hierarchical PRAMs on meshes

Author

Heywood, Todd ; Leopold, Claudia

Author_Institution

Dept. of Comput. Sci., Edinburgh Univ., UK

fYear

1995

fDate

15-17 Mar 1995

Firstpage

257

Lastpage

263

Abstract

The Hierarchical PRAM (H-PRAM) (T. Heywood, S. Ranka, 1992) model is a dynamically partitionable PRAM, which charges for communication and synchronization, and allows parallel algorithms to abstractly represent general locality. We show that the H-PRAM can be implemented efficiently on a two dimensional mesh. We use the Peano indexing scheme to hierarchically partition the mesh. Multiple subPRAMs of the H-PRAM are simulated on irregular submeshes. For an H-PRAM program of cost T, the overall CRCW H-PRAM simulation runs in time constant in T with high probability. The simulation is dynamic, i.e. it does not depend on prior knowledge of a program´s specific hierarchical configuration, which may be data dependent

Keywords

parallel algorithms; parallel architectures; random-access storage; storage management; virtual machines; CRCW H-PRAM simulation; H-PRAM; Peano indexing scheme; dynamic randomized simulation; dynamically partitionable PRAM; general locality; hierarchical PRAMs; hierarchical configuration; irregular submeshes; parallel algorithms; parallel computing; two dimensional mesh; Computational modeling; Computer architecture; Computer science; Concurrent computing; Cost function; Indexing; Partitioning algorithms; Phase change random access memory; Physics computing; Reflectivity;

fLanguage

English

Publisher

ieee

Conference_Titel

Parallel Algorithms/Architecture Synthesis, 1995. Proceedings., First Aizu International Symposium on

Conference_Location

Fukushima

Print_ISBN

0-8186-7038-X

Type

conf

DOI

10.1109/AISPAS.1995.401331

Filename

401331