A control-parallel programming model implemented on SIMD hardware

H. Dietz, W. Cohen

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

10 Scopus citations

Abstract

Although “data parallelism” has been shown to be an effective and portable way to express some types of parallel algorithms, there are many other problems for which data parallelism seems awkward and inefficient. For example, recursive decompositions and operations on irregular grids are most readily expressed using control parallelism. The problem is that control parallelism has always been associated with MIMD (Multiple Instruction stream, Multiple Data stream) hardware. In this paper, we describe how to make a MIMD programming model execute efficiently on a SIMD (Single Instruction stream, Multiple Data stream) computer. The efficient execution of control-parallel code on a SIMD machine involves a careful blend of compiler technology and design and semi-automatic construction of the support routines (i.e., the MIMD emulator). This paper discusses how the techniques were applied to give the appearance of a 16;384-processor shared memory barrier MIMD using the hardware of a SIMD MasPar MP-1.

Original languageEnglish
Title of host publicationLanguages and Compilers for Parallel Computing - 5th International Workshop, Proceedings
EditorsUtpal Banerjee, David Gelernter, Alex Nicolau, David Padua
Pages311-325
Number of pages15
DOIs
StatePublished - 1993
EventIFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017 - Hamburg, Germany
Duration: Sep 3 2017Sep 7 2017

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume757 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceIFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017
Country/TerritoryGermany
CityHamburg
Period9/3/179/7/17

Bibliographical note

Funding Information:
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 award number 9015696-CDA.

Publisher Copyright:
© Springer-Verlag Berlin Heidelberg 1993.

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science (all)

Fingerprint

Dive into the research topics of 'A control-parallel programming model implemented on SIMD hardware'. Together they form a unique fingerprint.

Cite this