Selecting threads for workload migration in software distributed shared memory systems

Tyng Yeu Liang, Ce-Kuen Shieh, Jun Qi Li

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

When threads are migrated from heavily loaded nodes to lightly loaded nodes for load balance in software distributed shared memory systems, the communication cost of maintaining data consistency is increased if migration threads are carelessly selected. Program performance is degraded when loss from increased communication exceeds the benefit from load balancing. This study addresses the problem with a novel selection policy called reduction of inter-node sharing costs. The main characteristic of this policy is simultaneously considering thread memory access types and global sharing. The experimental results show that this policy can reduce the communication of benchmark applications by 50% during load balancing.

Original languageEnglish
Pages (from-to)893-913
Number of pages21
JournalParallel Computing
Volume28
Issue number6
DOIs
Publication statusPublished - 2002 Jun 1

Fingerprint

Distributed Shared Memory
Thread
Migration
Workload
Computer systems
Load Balancing
Data storage equipment
Resource allocation
Software
Communication
Vertex of a graph
Cost Sharing
Data Consistency
Load Balance
Communication Cost
Costs
Exceed
Sharing
Benchmark
Experimental Results

All Science Journal Classification (ASJC) codes

  • Software
  • Theoretical Computer Science
  • Hardware and Architecture
  • Computer Networks and Communications
  • Computer Graphics and Computer-Aided Design
  • Artificial Intelligence

Cite this

@article{7db07eebe01a485bb1ffb8f5b00c7c0e,
title = "Selecting threads for workload migration in software distributed shared memory systems",
abstract = "When threads are migrated from heavily loaded nodes to lightly loaded nodes for load balance in software distributed shared memory systems, the communication cost of maintaining data consistency is increased if migration threads are carelessly selected. Program performance is degraded when loss from increased communication exceeds the benefit from load balancing. This study addresses the problem with a novel selection policy called reduction of inter-node sharing costs. The main characteristic of this policy is simultaneously considering thread memory access types and global sharing. The experimental results show that this policy can reduce the communication of benchmark applications by 50{\%} during load balancing.",
author = "Liang, {Tyng Yeu} and Ce-Kuen Shieh and Li, {Jun Qi}",
year = "2002",
month = "6",
day = "1",
doi = "10.1016/S0167-8191(02)00071-6",
language = "English",
volume = "28",
pages = "893--913",
journal = "Parallel Computing",
issn = "0167-8191",
publisher = "Elsevier",
number = "6",

}

Selecting threads for workload migration in software distributed shared memory systems. / Liang, Tyng Yeu; Shieh, Ce-Kuen; Li, Jun Qi.

In: Parallel Computing, Vol. 28, No. 6, 01.06.2002, p. 893-913.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Selecting threads for workload migration in software distributed shared memory systems

AU - Liang, Tyng Yeu

AU - Shieh, Ce-Kuen

AU - Li, Jun Qi

PY - 2002/6/1

Y1 - 2002/6/1

N2 - When threads are migrated from heavily loaded nodes to lightly loaded nodes for load balance in software distributed shared memory systems, the communication cost of maintaining data consistency is increased if migration threads are carelessly selected. Program performance is degraded when loss from increased communication exceeds the benefit from load balancing. This study addresses the problem with a novel selection policy called reduction of inter-node sharing costs. The main characteristic of this policy is simultaneously considering thread memory access types and global sharing. The experimental results show that this policy can reduce the communication of benchmark applications by 50% during load balancing.

AB - When threads are migrated from heavily loaded nodes to lightly loaded nodes for load balance in software distributed shared memory systems, the communication cost of maintaining data consistency is increased if migration threads are carelessly selected. Program performance is degraded when loss from increased communication exceeds the benefit from load balancing. This study addresses the problem with a novel selection policy called reduction of inter-node sharing costs. The main characteristic of this policy is simultaneously considering thread memory access types and global sharing. The experimental results show that this policy can reduce the communication of benchmark applications by 50% during load balancing.

UR - http://www.scopus.com/inward/record.url?scp=0036601968&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0036601968&partnerID=8YFLogxK

U2 - 10.1016/S0167-8191(02)00071-6

DO - 10.1016/S0167-8191(02)00071-6

M3 - Article

VL - 28

SP - 893

EP - 913

JO - Parallel Computing

JF - Parallel Computing

SN - 0167-8191

IS - 6

ER -