The Good News About Dynamic Object-Oriented Parallel Processing

Report
Authors:Grimshaw, Andrew, Department of Computer ScienceUniversity of Virginia Strayer, W, Department of Computer ScienceUniversity of Virginia Narayan, Padmini, Department of Computer ScienceUniversity of Virginia
Abstract:

Mentat [1] is an object-oriented parallel processing system that addresses the need for easy-to-use tools for the construction of portable medium-grain parallel software. Three of the key features of the Mentat approach are that it is object-oriented, that it uses a layered virtual machine, and that it dynamically constructs data dependence program graphs and dynamically manages system resources and communication. It has been argued that the overhead of these three features would be so large that the performance gains over sequential execution would be negligible. We have evidence to the contrary. In this paper we show that these features do not inherently imply poor performance. We introduce Mentat, its object model, and those features that most strongly affect performance. We then examine the performance of four applications under Mentat on two different platforms. The four applications were selected because they span the spectrum of communication and synchronization requirements. The results show that a dynamic, object-oriented parallel programming environment can offer both a hospitable programming environment as well as good performance for a wide range of applications.
Note: Abstract extracted from PDF text

Rights:
All rights reserved (no additional license for public reuse)
Language:
English
Source Citation:

Grimshaw, Andrew, W Strayer, and Padmini Narayan. "The Good News About Dynamic Object-Oriented Parallel Processing." University of Virginia Dept. of Computer Science Tech Report (1992).

Publisher:
University of Virginia, Department of Computer Science
Published Date:
1992