Dynamic topology selection for high performance MPI in the grid environments

Kyung Lang Park, Hwang Jik Lee, Kwang Won Koh, Oh Young Kwon, Sung Yong Park, Hyoung Woo Park, Shin-Dug Kim

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

MPI (Message Passing Interface) is getting more popular and important even in the Grid, but its performance still remains a problem, which is caused by the communication bottleneck on wide area links. To overcome such performance wall problem, we propose a dynamic topology selection which provides an effective resource selection service based on the principles of wide area message passing. It attempts to match the communication pattern of application with the topology of resources. Consequently, the proposed method provides an optimized set of resources and improves overall application performance by reducing the communication delay. To demonstrate the proposed method, we executed parallel benchmark programs and measured each execution time in five geometrically distributed clusters by changing resource selection method. When using topology selection method, experimental results show that performance gain can be achieved by up to 200%.

Original languageEnglish
Pages (from-to)595-602
Number of pages8
JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2840
Publication statusPublished - 2003 Dec 1

Fingerprint

Message Passing Interface
Message passing
High Performance
Topology
Grid
Resources
Communication
Service Selection
Communication Delay
Message Passing
Execution Time
Benchmark
Experimental Results
Demonstrate

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

@article{4663a02b8b394ceaade08bdf9e629292,
title = "Dynamic topology selection for high performance MPI in the grid environments",
abstract = "MPI (Message Passing Interface) is getting more popular and important even in the Grid, but its performance still remains a problem, which is caused by the communication bottleneck on wide area links. To overcome such performance wall problem, we propose a dynamic topology selection which provides an effective resource selection service based on the principles of wide area message passing. It attempts to match the communication pattern of application with the topology of resources. Consequently, the proposed method provides an optimized set of resources and improves overall application performance by reducing the communication delay. To demonstrate the proposed method, we executed parallel benchmark programs and measured each execution time in five geometrically distributed clusters by changing resource selection method. When using topology selection method, experimental results show that performance gain can be achieved by up to 200{\%}.",
author = "Park, {Kyung Lang} and Lee, {Hwang Jik} and Koh, {Kwang Won} and Kwon, {Oh Young} and Park, {Sung Yong} and Park, {Hyoung Woo} and Shin-Dug Kim",
year = "2003",
month = "12",
day = "1",
language = "English",
volume = "2840",
pages = "595--602",
journal = "Lecture Notes in Computer Science",
issn = "0302-9743",
publisher = "Springer Verlag",

}

Dynamic topology selection for high performance MPI in the grid environments. / Park, Kyung Lang; Lee, Hwang Jik; Koh, Kwang Won; Kwon, Oh Young; Park, Sung Yong; Park, Hyoung Woo; Kim, Shin-Dug.

In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 2840, 01.12.2003, p. 595-602.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Dynamic topology selection for high performance MPI in the grid environments

AU - Park, Kyung Lang

AU - Lee, Hwang Jik

AU - Koh, Kwang Won

AU - Kwon, Oh Young

AU - Park, Sung Yong

AU - Park, Hyoung Woo

AU - Kim, Shin-Dug

PY - 2003/12/1

Y1 - 2003/12/1

N2 - MPI (Message Passing Interface) is getting more popular and important even in the Grid, but its performance still remains a problem, which is caused by the communication bottleneck on wide area links. To overcome such performance wall problem, we propose a dynamic topology selection which provides an effective resource selection service based on the principles of wide area message passing. It attempts to match the communication pattern of application with the topology of resources. Consequently, the proposed method provides an optimized set of resources and improves overall application performance by reducing the communication delay. To demonstrate the proposed method, we executed parallel benchmark programs and measured each execution time in five geometrically distributed clusters by changing resource selection method. When using topology selection method, experimental results show that performance gain can be achieved by up to 200%.

AB - MPI (Message Passing Interface) is getting more popular and important even in the Grid, but its performance still remains a problem, which is caused by the communication bottleneck on wide area links. To overcome such performance wall problem, we propose a dynamic topology selection which provides an effective resource selection service based on the principles of wide area message passing. It attempts to match the communication pattern of application with the topology of resources. Consequently, the proposed method provides an optimized set of resources and improves overall application performance by reducing the communication delay. To demonstrate the proposed method, we executed parallel benchmark programs and measured each execution time in five geometrically distributed clusters by changing resource selection method. When using topology selection method, experimental results show that performance gain can be achieved by up to 200%.

UR - http://www.scopus.com/inward/record.url?scp=34548206075&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=34548206075&partnerID=8YFLogxK

M3 - Article

VL - 2840

SP - 595

EP - 602

JO - Lecture Notes in Computer Science

JF - Lecture Notes in Computer Science

SN - 0302-9743

ER -