A control-parallel programming model implemented on SIMD hardware

H. Dietz; W. Cohen

doi:10.1007/3-540-57502-2_55

A control-parallel programming model implemented on SIMD hardware

H. Dietz, W. Cohen

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-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 language	English
Title of host publication	Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings
Editors	Utpal Banerjee, David Gelernter, Alex Nicolau, David Padua
Pages	311-325
Number of pages	15
DOIs	https://doi.org/10.1007/3-540-57502-2_55
State	Published - 1993
Event	IFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017 - Hamburg, Germany Duration: Sep 3 2017 → Sep 7 2017

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	757 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	IFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017
Country/Territory	Germany
City	Hamburg
Period	9/3/17 → 9/7/17

Bibliographical note

Publisher Copyright:
© Springer-Verlag Berlin Heidelberg 1993.

Funding

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.

Funders	Funder number
National Science Foundation (NSF)	9015696-CDA
Office of Naval Research	N00014-91-J-4013

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/3-540-57502-2_55

Cite this

Dietz, H., & Cohen, W. (1993). A control-parallel programming model implemented on SIMD hardware. In U. Banerjee, D. Gelernter, A. Nicolau, & D. Padua (Eds.), Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings (pp. 311-325). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 757 LNCS). https://doi.org/10.1007/3-540-57502-2_55

Dietz, H. ; Cohen, W. / A control-parallel programming model implemented on SIMD hardware. Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings. editor / Utpal Banerjee ; David Gelernter ; Alex Nicolau ; David Padua. 1993. pp. 311-325 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{1963ff24e3914e6aac87c2ecc2030e54,

title = "A control-parallel programming model implemented on SIMD hardware",

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.",

author = "H. Dietz and W. Cohen",

note = "Publisher Copyright: {\textcopyright} Springer-Verlag Berlin Heidelberg 1993.; IFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017 ; Conference date: 03-09-2017 Through 07-09-2017",

year = "1993",

doi = "10.1007/3-540-57502-2_55",

language = "English",

isbn = "9783540575023",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

pages = "311--325",

editor = "Utpal Banerjee and David Gelernter and Alex Nicolau and David Padua",

booktitle = "Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings",

}

Dietz, H & Cohen, W 1993, A control-parallel programming model implemented on SIMD hardware. in U Banerjee, D Gelernter, A Nicolau & D Padua (eds), Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 757 LNCS, pp. 311-325, IFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017, Hamburg, Germany, 9/3/17. https://doi.org/10.1007/3-540-57502-2_55

A control-parallel programming model implemented on SIMD hardware. / Dietz, H.; Cohen, W.
Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings. ed. / Utpal Banerjee; David Gelernter; Alex Nicolau; David Padua. 1993. p. 311-325 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 757 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - A control-parallel programming model implemented on SIMD hardware

AU - Dietz, H.

AU - Cohen, W.

N1 - Publisher Copyright: © Springer-Verlag Berlin Heidelberg 1993.

PY - 1993

Y1 - 1993

N2 - 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.

AB - 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.

UR - http://www.scopus.com/inward/record.url?scp=0040657566&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0040657566&partnerID=8YFLogxK

U2 - 10.1007/3-540-57502-2_55

DO - 10.1007/3-540-57502-2_55

M3 - Conference contribution

AN - SCOPUS:0040657566

SN - 9783540575023

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 311

EP - 325

BT - Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings

A2 - Banerjee, Utpal

A2 - Gelernter, David

A2 - Nicolau, Alex

A2 - Padua, David

T2 - IFIP WG 5.7 International Conference on Advances in Production Management Systems, APMS 2017

Y2 - 3 September 2017 through 7 September 2017

ER -

Dietz H, Cohen W. A control-parallel programming model implemented on SIMD hardware. In Banerjee U, Gelernter D, Nicolau A, Padua D, editors, Languages and Compilers for Parallel Computing - 5th International Workshop, Proceedings. 1993. p. 311-325. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/3-540-57502-2_55

A control-parallel programming model implemented on SIMD hardware

Abstract

Publication series

Conference

Bibliographical note

Funding

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this