Communicating data-parallel tasks: An MPI library for HPF

Ian T. Foster; David R. Kohr; Rakesh Krishnaiyer; Alok Choudhary

Communicating data-parallel tasks: An MPI library for HPF

Ian T. Foster^*, David R. Kohr, Rakesh Krishnaiyer, Alok Choudhary

^*Corresponding author for this work

Electrical and Computer Engineering

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

Abstract

High Performance Fortran (HPF) has emerged as a standard dialect of Fortran for data-parallel computing. However, HPF does not support task parallelism or heterogeneous computing adequately. This paper presents a summary of our work on a library-based approach to support task parallelism, using MPI as a coordination layer for HPF. This library enables a wide variety of applications, such as multidisciplinary simulations and pipeline computations, to take advantage of combined task and data parallelism. An HPF binding for MPI raises several interface and communication issues. We discuss these issues and describe our implementation of an HPF/MPI library that operates with a commercial HPF compiler. We also evaluate the performance of our library using a synthetic communication benchmark and a multiblock application.

Original language	English (US)
Title of host publication	Proceedings of the 1996 3rd International Conference on High Performance Computing, HiPC
Editors	Anon
Publisher	IEEE
Pages	433-438
Number of pages	6
State	Published - Dec 1 1996
Event	Proceedings of the 1996 3rd International Conference on High Performance Computing, HiPC - Trivandrum, India Duration: Dec 19 1996 → Dec 22 1996

Other

Other	Proceedings of the 1996 3rd International Conference on High Performance Computing, HiPC
City	Trivandrum, India
Period	12/19/96 → 12/22/96

ASJC Scopus subject areas

General Engineering

Cite this

@inproceedings{0856f66d53ce4fe283712323f17ca16e,

title = "Communicating data-parallel tasks: An MPI library for HPF",

abstract = "High Performance Fortran (HPF) has emerged as a standard dialect of Fortran for data-parallel computing. However, HPF does not support task parallelism or heterogeneous computing adequately. This paper presents a summary of our work on a library-based approach to support task parallelism, using MPI as a coordination layer for HPF. This library enables a wide variety of applications, such as multidisciplinary simulations and pipeline computations, to take advantage of combined task and data parallelism. An HPF binding for MPI raises several interface and communication issues. We discuss these issues and describe our implementation of an HPF/MPI library that operates with a commercial HPF compiler. We also evaluate the performance of our library using a synthetic communication benchmark and a multiblock application.",

author = "Foster, {Ian T.} and Kohr, {David R.} and Rakesh Krishnaiyer and Alok Choudhary",

year = "1996",

month = dec,

day = "1",

language = "English (US)",

pages = "433--438",

editor = "Anon",

booktitle = "Proceedings of the 1996 3rd International Conference on High Performance Computing, HiPC",

publisher = "IEEE",

note = "Proceedings of the 1996 3rd International Conference on High Performance Computing, HiPC ; Conference date: 19-12-1996 Through 22-12-1996",

}

TY - GEN

T1 - Communicating data-parallel tasks

T2 - Proceedings of the 1996 3rd International Conference on High Performance Computing, HiPC

AU - Foster, Ian T.

AU - Kohr, David R.

AU - Krishnaiyer, Rakesh

AU - Choudhary, Alok

PY - 1996/12/1

Y1 - 1996/12/1

N2 - High Performance Fortran (HPF) has emerged as a standard dialect of Fortran for data-parallel computing. However, HPF does not support task parallelism or heterogeneous computing adequately. This paper presents a summary of our work on a library-based approach to support task parallelism, using MPI as a coordination layer for HPF. This library enables a wide variety of applications, such as multidisciplinary simulations and pipeline computations, to take advantage of combined task and data parallelism. An HPF binding for MPI raises several interface and communication issues. We discuss these issues and describe our implementation of an HPF/MPI library that operates with a commercial HPF compiler. We also evaluate the performance of our library using a synthetic communication benchmark and a multiblock application.

AB - High Performance Fortran (HPF) has emerged as a standard dialect of Fortran for data-parallel computing. However, HPF does not support task parallelism or heterogeneous computing adequately. This paper presents a summary of our work on a library-based approach to support task parallelism, using MPI as a coordination layer for HPF. This library enables a wide variety of applications, such as multidisciplinary simulations and pipeline computations, to take advantage of combined task and data parallelism. An HPF binding for MPI raises several interface and communication issues. We discuss these issues and describe our implementation of an HPF/MPI library that operates with a commercial HPF compiler. We also evaluate the performance of our library using a synthetic communication benchmark and a multiblock application.

UR - http://www.scopus.com/inward/record.url?scp=0030393241&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0030393241&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:0030393241

SP - 433

EP - 438

BT - Proceedings of the 1996 3rd International Conference on High Performance Computing, HiPC

A2 - Anon, null

PB - IEEE

Y2 - 19 December 1996 through 22 December 1996

ER -

Communicating data-parallel tasks: An MPI library for HPF

Abstract

Other

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this