Table of Contents Author Guidelines Submit a Manuscript
Scientific Programming
Volume 3 (1994), Issue 4, Pages 301-324

The Performance of an Object-Oriented, Parallel Operating System

David R. Kohr, Jr.,1 Xingbin Zhang,1 Mustafizur Rahman,2 and Daniel A. Reed1

1Department of Computer Science, University of Illinois, Urbana, IL 61801, USA
2Department of Computer Science, University of Massachusetts at Amherst, Amherst, MA 01003, USA

Received 5 April 1994; Accepted 5 May 1994

Copyright © 1994 Hindawi Publishing Corporation. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


The nascent and rapidly evolving state of parallel systems often leaves parallel application developers at the mercy of inefficient, inflexible operating system software. Given the relatively primitive state of parallel systems software, maximizing the performance of parallel applications not only requires judicious tuning of the application software, but occasionally, the replacement of specific system software modules with others that can more readily respond to the imposed pattern of resource demands. To assess the feasibility of application and performance tuning via malleable system software and to understand the performance penalties for detailed operating system performance data capture, we describe a set of performance instrumentation techniques for parallel, object-oriented operating systems and a set of performance experiments with Choices, an experimental, object-oriented operating system designed for use with parallel sys- tems. These performance experiments show that (a) the performance overhead for operating system data capture is modest, (b) the penalty for malleable, object-oriented operating systems is negligible, but (c) techniques are needed to strictly enforce adherence of implementation to design if operating system modules are to be replaced.