A synthesis of memory mechanisms for distributed architectures

J. Zhu, J. Hoeflinger, D. Padua

Research output: Contribution to conferencePaper

Abstract

Producing efficient parallel programs for distributed memory multiprocessors is a difficult task. Hand-coding efficient parallel programs for these systems can be extremely difficult, time consuming and error-prone, so people have turned to the shared memory abstraction and automatic parallelizing compilers to ease the task. The two main approaches to this are using compilers that 1) generate message passing code, or 2) generate code for a distributed shared memory software layer. Neither has been completely successful for all types of programs. In this paper, we discuss the use of a combination of these mechanisms to produce a compiler code generation paradigm that can be successful for many user programs. The experimental results indicate that our new paradigm would be able to support both regular and irregular code efficiently.

Original languageEnglish (US)
Pages13-22
Number of pages10
StatePublished - Jan 1 2001
Event2001 International Conference on Supercomputing - Sorento, Italy
Duration: Jun 17 2001Jun 21 2001

Other

Other2001 International Conference on Supercomputing
CountryItaly
CitySorento
Period6/17/016/21/01

Fingerprint

Data storage equipment
Message passing
Code generation

ASJC Scopus subject areas

  • Computer Science(all)

Cite this

Zhu, J., Hoeflinger, J., & Padua, D. (2001). A synthesis of memory mechanisms for distributed architectures. 13-22. Paper presented at 2001 International Conference on Supercomputing, Sorento, Italy.

A synthesis of memory mechanisms for distributed architectures. / Zhu, J.; Hoeflinger, J.; Padua, D.

2001. 13-22 Paper presented at 2001 International Conference on Supercomputing, Sorento, Italy.

Research output: Contribution to conferencePaper

Zhu, J, Hoeflinger, J & Padua, D 2001, 'A synthesis of memory mechanisms for distributed architectures', Paper presented at 2001 International Conference on Supercomputing, Sorento, Italy, 6/17/01 - 6/21/01 pp. 13-22.
Zhu J, Hoeflinger J, Padua D. A synthesis of memory mechanisms for distributed architectures. 2001. Paper presented at 2001 International Conference on Supercomputing, Sorento, Italy.
Zhu, J. ; Hoeflinger, J. ; Padua, D. / A synthesis of memory mechanisms for distributed architectures. Paper presented at 2001 International Conference on Supercomputing, Sorento, Italy.10 p.
@conference{4525d0b3126c472aa6f91a81d0696db8,
title = "A synthesis of memory mechanisms for distributed architectures",
abstract = "Producing efficient parallel programs for distributed memory multiprocessors is a difficult task. Hand-coding efficient parallel programs for these systems can be extremely difficult, time consuming and error-prone, so people have turned to the shared memory abstraction and automatic parallelizing compilers to ease the task. The two main approaches to this are using compilers that 1) generate message passing code, or 2) generate code for a distributed shared memory software layer. Neither has been completely successful for all types of programs. In this paper, we discuss the use of a combination of these mechanisms to produce a compiler code generation paradigm that can be successful for many user programs. The experimental results indicate that our new paradigm would be able to support both regular and irregular code efficiently.",
author = "J. Zhu and J. Hoeflinger and D. Padua",
year = "2001",
month = "1",
day = "1",
language = "English (US)",
pages = "13--22",
note = "2001 International Conference on Supercomputing ; Conference date: 17-06-2001 Through 21-06-2001",

}

TY - CONF

T1 - A synthesis of memory mechanisms for distributed architectures

AU - Zhu, J.

AU - Hoeflinger, J.

AU - Padua, D.

PY - 2001/1/1

Y1 - 2001/1/1

N2 - Producing efficient parallel programs for distributed memory multiprocessors is a difficult task. Hand-coding efficient parallel programs for these systems can be extremely difficult, time consuming and error-prone, so people have turned to the shared memory abstraction and automatic parallelizing compilers to ease the task. The two main approaches to this are using compilers that 1) generate message passing code, or 2) generate code for a distributed shared memory software layer. Neither has been completely successful for all types of programs. In this paper, we discuss the use of a combination of these mechanisms to produce a compiler code generation paradigm that can be successful for many user programs. The experimental results indicate that our new paradigm would be able to support both regular and irregular code efficiently.

AB - Producing efficient parallel programs for distributed memory multiprocessors is a difficult task. Hand-coding efficient parallel programs for these systems can be extremely difficult, time consuming and error-prone, so people have turned to the shared memory abstraction and automatic parallelizing compilers to ease the task. The two main approaches to this are using compilers that 1) generate message passing code, or 2) generate code for a distributed shared memory software layer. Neither has been completely successful for all types of programs. In this paper, we discuss the use of a combination of these mechanisms to produce a compiler code generation paradigm that can be successful for many user programs. The experimental results indicate that our new paradigm would be able to support both regular and irregular code efficiently.

UR - http://www.scopus.com/inward/record.url?scp=0034825253&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0034825253&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:0034825253

SP - 13

EP - 22

ER -