Prioritization in parallel symbolic computing

L. V. Kale, B. Ramkumar, V. Saletore, A. B. Sinha

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

It is argued that scheduling is an important determinant of performance for many parallel symbolic computations, in addition to the issues of dynamic load balancing and grain-size control. We propose associating unbounded levels of priorities with tasks and messages as the mechanism of choice for specifying scheduling strategies. We demonstrate how priorities can be used in parallelizing computations in different search domains, and show how priorities can be implemented effectively in parallel systems. Priorities have been implemented in the Charm portable parallel programming system. Performance results on shared-memory machines with tens of processors and nonshared-memory machines with hundreds of processors are given. Open problems for prioritization in specific domains are given, which will constitute fertile area for future research in this field.

Original languageEnglish (US)
Title of host publicationParallel Symbolic Computing
Subtitle of host publicationLanguages, Systems, and Applications - US/Japan Workshop, Proceedings
EditorsRobert H. Halstead, Takayasu Ito
PublisherSpringer
Pages12-41
Number of pages30
ISBN (Print)9783540573968
DOIs
StatePublished - 1993
EventUS/Japan Workshop on Parallel Symbolic Computing, 1992 - Cambridge, United States
Duration: Oct 14 1992Oct 17 1992

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume748 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

OtherUS/Japan Workshop on Parallel Symbolic Computing, 1992
Country/TerritoryUnited States
CityCambridge
Period10/14/9210/17/92

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Prioritization in parallel symbolic computing'. Together they form a unique fingerprint.

Cite this