Work distribution in parallel programs for distributed memory multiprocessors

Michael Gerndt

doi:10.1145/109025.109053

Work distribution in parallel programs for distributed memory multiprocessors

Michael Gerndt

Vienna-UNI

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

3 Scopus citations

Abstract

One of the most important issues in parallel programming for distributed memory multiprocessors is to distribute data and work in a cooperative way, so that data are mostly available where they are needed. This paper discusses masking, a technique to distribute work among parallel processes running on a distributed memory multiprocessor. This technique was developed for the par allegation of Fortran 77 programs for the SUPRENUM supercomputer. In the semi-automatic parallelization system SUPERB we implemented masking on the basis of a user defined data distribution. Both concepts, data and work distribution, are very general concepts and can be applied to parallel programs for all kinds of multiprocessors based on memory hierarchies.

Original language	English
Title of host publication	ICS 1991 - Proceedings of the 5th International Conference on Supercomputing
Editors	Edward S. Davidson, Friedel Hossfield
Publisher	Association for Computing Machinery
Pages	96-104
Number of pages	9
ISBN (Print)	0897914341, 9780897914345
DOIs	https://doi.org/10.1145/109025.109053
State	Published - 1 Jun 1991
Externally published	Yes
Event	5th International Conference on Supercomputing, ICS 1991 - Cologne, Germany Duration: 17 Jun 1991 → 21 Jun 1991

Publication series

Name	Proceedings of the International Conference on Supercomputing

Conference

Conference	5th International Conference on Supercomputing, ICS 1991
Country/Territory	Germany
City	Cologne
Period	17/06/91 → 21/06/91

Keywords

Analysis of algorithms
Multiprocessors
Program transformations

Access to Document

10.1145/109025.109053

Cite this

Gerndt, M. (1991). Work distribution in parallel programs for distributed memory multiprocessors. In E. S. Davidson, & F. Hossfield (Eds.), ICS 1991 - Proceedings of the 5th International Conference on Supercomputing (pp. 96-104). (Proceedings of the International Conference on Supercomputing). Association for Computing Machinery. https://doi.org/10.1145/109025.109053

@inproceedings{d61878e39d834f9e9279161e1f98c4d2,

title = "Work distribution in parallel programs for distributed memory multiprocessors",

abstract = "One of the most important issues in parallel programming for distributed memory multiprocessors is to distribute data and work in a cooperative way, so that data are mostly available where they are needed. This paper discusses masking, a technique to distribute work among parallel processes running on a distributed memory multiprocessor. This technique was developed for the par allegation of Fortran 77 programs for the SUPRENUM supercomputer. In the semi-automatic parallelization system SUPERB we implemented masking on the basis of a user defined data distribution. Both concepts, data and work distribution, are very general concepts and can be applied to parallel programs for all kinds of multiprocessors based on memory hierarchies.",

keywords = "Analysis of algorithms, Multiprocessors, Program transformations",

author = "Michael Gerndt",

note = "Publisher Copyright: {\textcopyright} 1991 ACM.; 5th International Conference on Supercomputing, ICS 1991 ; Conference date: 17-06-1991 Through 21-06-1991",

year = "1991",

month = jun,

day = "1",

doi = "10.1145/109025.109053",

language = "English",

isbn = "0897914341",

series = "Proceedings of the International Conference on Supercomputing",

publisher = "Association for Computing Machinery",

pages = "96--104",

editor = "Davidson, {Edward S.} and Friedel Hossfield",

booktitle = "ICS 1991 - Proceedings of the 5th International Conference on Supercomputing",

}

Gerndt, M 1991, Work distribution in parallel programs for distributed memory multiprocessors. in ES Davidson & F Hossfield (eds), ICS 1991 - Proceedings of the 5th International Conference on Supercomputing. Proceedings of the International Conference on Supercomputing, Association for Computing Machinery, pp. 96-104, 5th International Conference on Supercomputing, ICS 1991, Cologne, Germany, 17/06/91. https://doi.org/10.1145/109025.109053

Work distribution in parallel programs for distributed memory multiprocessors. / Gerndt, Michael.
ICS 1991 - Proceedings of the 5th International Conference on Supercomputing. ed. / Edward S. Davidson; Friedel Hossfield. Association for Computing Machinery, 1991. p. 96-104 (Proceedings of the International Conference on Supercomputing).

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

TY - GEN

T1 - Work distribution in parallel programs for distributed memory multiprocessors

AU - Gerndt, Michael

PY - 1991/6/1

Y1 - 1991/6/1

N2 - One of the most important issues in parallel programming for distributed memory multiprocessors is to distribute data and work in a cooperative way, so that data are mostly available where they are needed. This paper discusses masking, a technique to distribute work among parallel processes running on a distributed memory multiprocessor. This technique was developed for the par allegation of Fortran 77 programs for the SUPRENUM supercomputer. In the semi-automatic parallelization system SUPERB we implemented masking on the basis of a user defined data distribution. Both concepts, data and work distribution, are very general concepts and can be applied to parallel programs for all kinds of multiprocessors based on memory hierarchies.

AB - One of the most important issues in parallel programming for distributed memory multiprocessors is to distribute data and work in a cooperative way, so that data are mostly available where they are needed. This paper discusses masking, a technique to distribute work among parallel processes running on a distributed memory multiprocessor. This technique was developed for the par allegation of Fortran 77 programs for the SUPRENUM supercomputer. In the semi-automatic parallelization system SUPERB we implemented masking on the basis of a user defined data distribution. Both concepts, data and work distribution, are very general concepts and can be applied to parallel programs for all kinds of multiprocessors based on memory hierarchies.

KW - Analysis of algorithms

KW - Multiprocessors

KW - Program transformations

UR - http://www.scopus.com/inward/record.url?scp=84943727487&partnerID=8YFLogxK

U2 - 10.1145/109025.109053

DO - 10.1145/109025.109053

M3 - Conference contribution

AN - SCOPUS:84943727487

SN - 0897914341

SN - 9780897914345

T3 - Proceedings of the International Conference on Supercomputing

SP - 96

EP - 104

BT - ICS 1991 - Proceedings of the 5th International Conference on Supercomputing

A2 - Davidson, Edward S.

A2 - Hossfield, Friedel

PB - Association for Computing Machinery

T2 - 5th International Conference on Supercomputing, ICS 1991

Y2 - 17 June 1991 through 21 June 1991

ER -

Work distribution in parallel programs for distributed memory multiprocessors

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this