A parallel processing support library based on synchronized aggregate communication

H. G. Dietz, T. M. Chung, T. L. Mattox

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Scopus citations

Abstract

The concept of “data parallelism” is a pervasive force throughout parallel processing. Although a certain level of processing-element autonomy can help performance, the fact is that many parallel algorithms, applications, and compiler analysis techniques focus on identifying a set of data objects that can be processed using loosely synchronous parallelism. Thus, it is not surprising that a large number of communication libraries support at least a few synchronized aggregate operations on data. In this paper, we present an overview of the parallel processing support library for PAPERS clusters. Unlike most other systems, which construct aggregate communications by layering message-passing or shared-memory communication, the PAPERS hardware and software directly implements a model of parallel execution based on synchronized aggregate communications. Asynchronous processor operation is fully supported, but asynchronous communications are not directly supported (although they can be derived using the PAPERS parallel interrupt mechanism). Thus, PAPERS provides highly efficient aggregate communications for MIMD, SIMD, and VLIW execution modes. We demonstrate the effectiveness of this implementation by presenting detailed benchmarks for a variety of other libraries supporting dedicated parallel machines and workstation clusters. This work was supported in part by the Office of Naval Research (ONR) under grant number N00014-91-J-4013 and by the National Science Foundation (NSF) under Parallel Infrastructure Grant number CDA-9015696.

Original languageEnglish
Title of host publicationLanguages and Compilers for Parallel Computing - 8th International Workshop, LCPC 1995, Proceedings
EditorsChua-Huang Huang, Ponnuswamy Sadayappan, Utpal Banerjee, David Gelernter, Alex Nicolau, David Padua
Pages251-268
Number of pages18
DOIs
StatePublished - 1996
Event8th International Workshop on Languages and Compilers for Parallel Computing, LCPC 1995 - Columbus, United States
Duration: Aug 10 1995Aug 12 1995

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume1033
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference8th International Workshop on Languages and Compilers for Parallel Computing, LCPC 1995
Country/TerritoryUnited States
CityColumbus
Period8/10/958/12/95

Bibliographical note

Publisher Copyright:
© Springer-Veriag Berlin Heidelberg 1996.

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'A parallel processing support library based on synchronized aggregate communication'. Together they form a unique fingerprint.

Cite this