The Automated Partitioning of Simulations for Parallel Execution

Report
Authors:Nicol, David, Department of Computer ScienceUniversity of Virginia Reynolds, Paul, Department of Computer ScienceUniversity of Virginia
Abstract:

One of the most active areas in computer science research today is the study of parallel computation. A critical problem in this field is that of partitioning: the decomposition of a problem, program, or algorithm into pieces suitable for parallel execution. We consider the problem of partitioning simulations automatically for parallel execution. Before We can study the partitioning of simulations, we must be able to analyze simulations. Towards this end, we construct a formal model of simulations, study a simu1ation's run-time behavior, and explore the eifect of synchronization on run~time performance. We develop static and statistical methods of analyzing a distributed simulatioifs performance. We then propose a simulation partitioning algorithm and examine the performance of partitions it creates. Finally, We develop an optimal dynamic repartitioning decision process.
Note: Abstract extracted from PDF file via OCR

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

Nicol, David, and Paul Reynolds. "The Automated Partitioning of Simulations for Parallel Execution." University of Virginia Dept. of Computer Science Tech Report (1985).

Publisher:
University of Virginia, Department of Computer Science
Published Date:
1985