Beowulf Cluster Construction at TRIUMF

Aaron Lees

XXXXXX
University of Victoria
Department of Computer Science

January 10, 2000

In partial fulfillment of the requirements of the Computer Science/Mathematics Co-op Program

This text is distributed according to the General Public License.

Table of Contents
1. Report Specification
2. Introduction
3. Hardware
4. Software
5. Cluster Setup
6. Programming
7. Resources
Glossary
Appendix A. Source Code for the addnode.pl Perl Script
Appendix B. Source Code for the cpi.c MPI Example Program

1. Report Specification

1.1. Purpose

The purpose of this report is to document my work at TRIUMF in constructing a beowulf cluster. I will begin with a brief introduction to TRIUMF, the general beowulf architecture and parallel programming. The remainder of the report will give an overview of my implementation of the cluster.

1.2. Audience

This report is meant to be read by three groups of people: TRIUMF staff, UVic Computer Science Students and UVic Computer Science Co-op staff. TRIUMF employees will be able to learn the details on how I implemented the cluster so that further work can be done on it. Students at UVic can get an idea of what a co-op term at TRIUMF would be like. The co-op staff can evaluate my performance on this work term.

1.3. Prerequisites

I will try to limit the amount of prerequisite knowledge required to understand this report as much as possible, however, the reader will gain more from this report if they are at least familiar with the following areas.

Computer hardware.
Computer networking.
Unix. Specifically RedHat linux.
Parallel computing.

		Next
		Introduction