OpenMP implementation and performance
		Next

OpenMP implementation and performance

Jeongnim Kim

$Revision: 1.5 $

Copyright © 2004 Jeongnim Kim

$Date: 2004/09/14 17:19:50 $

Abstract

We discuss OpenMP parallelization in ohmms and its performance on SMP platforms.

Table of Contents

1. Introduction

2. Implementation details

3. Classical simulations

Summary
Intel compilers on Itanium 2 SMP clusters
KAPPRO on Itanium 2 SMP clusters
KAPPRO on IBM P4 clusters
IBM VisualAge compilers of IBM P4 clusters

4. Improvement by using updates with displacements

Performance improvement
Comparisons of the implementations

List of Figures

2.1. OpenMP performance analysis by guideview
3.1. Performance on SGI Altix at OSC using intel compilers 8
3.2. Performance on SGI Altix at OSC using kappro compilers (analysis by guideview)
3.3. Performance on seaborg using KAPPRO compilers

List of Tables

3.1. Total time for the main parallel and serial regions for 8x8x16
4.1. Relative timing results without/with the update routine.

		Next
		Chapter 1. Introduction