A new approach to distribute program workload on software DSM clusters

Tyng Yeu Liang, Yen Tso Liu, Ce-Kuen Shieh, Chun Yi Wu

Research output: Contribution to conferencePaper

1 Citation (Scopus)

Abstract

The workload distribution approaches used by software distributed shared memory (DSM) clusters always distribute the working threads of applications according to the computational power of processors. However, in addition to computational cost, the cost of memory accesses is an important factor for determining program performance. Neglecting this cost will result in making wrong decisions in workload distribution and then degrading program performance. To address this problem, we propose a new approach with simultaneously considering the memory capability and the computational power of processors for workload distribution on software DSM clusters in this paper. We have implemented the proposed approach on a test bed. Our experimental results show that the proposed approach can provide more performance improvement for the applications compared to the others with considering only computational powers or memory capabilities.

Original languageEnglish
Pages201-206
Number of pages6
Publication statusPublished - 2004 Aug 16
EventProceedings on the International Symposium on Parallel Architectures, Algorithms and Networks, I-SPAN - Hong Kong, China
Duration: 2004 May 102004 May 12

Other

OtherProceedings on the International Symposium on Parallel Architectures, Algorithms and Networks, I-SPAN
CountryChina
CityHong Kong
Period04-05-1004-05-12

Fingerprint

Data storage equipment
Costs
Decision making

All Science Journal Classification (ASJC) codes

  • Computer Science(all)

Cite this

Liang, T. Y., Liu, Y. T., Shieh, C-K., & Wu, C. Y. (2004). A new approach to distribute program workload on software DSM clusters. 201-206. Paper presented at Proceedings on the International Symposium on Parallel Architectures, Algorithms and Networks, I-SPAN, Hong Kong, China.
Liang, Tyng Yeu ; Liu, Yen Tso ; Shieh, Ce-Kuen ; Wu, Chun Yi. / A new approach to distribute program workload on software DSM clusters. Paper presented at Proceedings on the International Symposium on Parallel Architectures, Algorithms and Networks, I-SPAN, Hong Kong, China.6 p.
@conference{336df7cde16543f18501db4f9c77e82f,
title = "A new approach to distribute program workload on software DSM clusters",
abstract = "The workload distribution approaches used by software distributed shared memory (DSM) clusters always distribute the working threads of applications according to the computational power of processors. However, in addition to computational cost, the cost of memory accesses is an important factor for determining program performance. Neglecting this cost will result in making wrong decisions in workload distribution and then degrading program performance. To address this problem, we propose a new approach with simultaneously considering the memory capability and the computational power of processors for workload distribution on software DSM clusters in this paper. We have implemented the proposed approach on a test bed. Our experimental results show that the proposed approach can provide more performance improvement for the applications compared to the others with considering only computational powers or memory capabilities.",
author = "Liang, {Tyng Yeu} and Liu, {Yen Tso} and Ce-Kuen Shieh and Wu, {Chun Yi}",
year = "2004",
month = "8",
day = "16",
language = "English",
pages = "201--206",
note = "Proceedings on the International Symposium on Parallel Architectures, Algorithms and Networks, I-SPAN ; Conference date: 10-05-2004 Through 12-05-2004",

}

Liang, TY, Liu, YT, Shieh, C-K & Wu, CY 2004, 'A new approach to distribute program workload on software DSM clusters' Paper presented at Proceedings on the International Symposium on Parallel Architectures, Algorithms and Networks, I-SPAN, Hong Kong, China, 04-05-10 - 04-05-12, pp. 201-206.

A new approach to distribute program workload on software DSM clusters. / Liang, Tyng Yeu; Liu, Yen Tso; Shieh, Ce-Kuen; Wu, Chun Yi.

2004. 201-206 Paper presented at Proceedings on the International Symposium on Parallel Architectures, Algorithms and Networks, I-SPAN, Hong Kong, China.

Research output: Contribution to conferencePaper

TY - CONF

T1 - A new approach to distribute program workload on software DSM clusters

AU - Liang, Tyng Yeu

AU - Liu, Yen Tso

AU - Shieh, Ce-Kuen

AU - Wu, Chun Yi

PY - 2004/8/16

Y1 - 2004/8/16

N2 - The workload distribution approaches used by software distributed shared memory (DSM) clusters always distribute the working threads of applications according to the computational power of processors. However, in addition to computational cost, the cost of memory accesses is an important factor for determining program performance. Neglecting this cost will result in making wrong decisions in workload distribution and then degrading program performance. To address this problem, we propose a new approach with simultaneously considering the memory capability and the computational power of processors for workload distribution on software DSM clusters in this paper. We have implemented the proposed approach on a test bed. Our experimental results show that the proposed approach can provide more performance improvement for the applications compared to the others with considering only computational powers or memory capabilities.

AB - The workload distribution approaches used by software distributed shared memory (DSM) clusters always distribute the working threads of applications according to the computational power of processors. However, in addition to computational cost, the cost of memory accesses is an important factor for determining program performance. Neglecting this cost will result in making wrong decisions in workload distribution and then degrading program performance. To address this problem, we propose a new approach with simultaneously considering the memory capability and the computational power of processors for workload distribution on software DSM clusters in this paper. We have implemented the proposed approach on a test bed. Our experimental results show that the proposed approach can provide more performance improvement for the applications compared to the others with considering only computational powers or memory capabilities.

UR - http://www.scopus.com/inward/record.url?scp=3543099877&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=3543099877&partnerID=8YFLogxK

M3 - Paper

SP - 201

EP - 206

ER -

Liang TY, Liu YT, Shieh C-K, Wu CY. A new approach to distribute program workload on software DSM clusters. 2004. Paper presented at Proceedings on the International Symposium on Parallel Architectures, Algorithms and Networks, I-SPAN, Hong Kong, China.