Design and Analysis of Parallel Hierarchical Battlefield Simulation.

Abstract

The purpose of this research is to determine if hierarchically partitioning a discrete event battlefield simulation reduces runtime and, if reduction exists, to characterize the run time reduction given any particular partition configuration. A hierarchical discrete event simulation of a main battle tank was constructed. Implementations were built for both a single processor and a multiprocessing machine. The implementations used the Message Passing Interface to increase portability to other parallel and distributed configurations. Three test cases were generated and run on three parallel and distributed environments, a network of Sun SparcStation 20's, a Silicon Graphics Power Challenge, and a Paragon XP/S. Three simplistic analytical models were constructed to develop the relationship between partition configurations. The results showed that hierarchically partitioning simulations can produce speedup if a single event causes multiple reactions, and those reactions contain a significant requirement for processing. The analytic models were able to predict which partition configuration was better from two possible configurations if the runtime of the events and the probability of the events occurring were known.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1995
Accession Number
ADA309496

Entities

People

  • Conrad P. Masshardt

Organizations

  • Air Force Institute of Technology

Tags

Communities of Interest

  • Energy and Power Technologies
  • Materials and Manufacturing Processes
  • Weapons Technologies

DTIC Thesaurus Topics

  • Air Force
  • Circuit Boards
  • Computer Programming
  • Computer Simulations
  • Computers
  • Flow Rate
  • Formal Languages
  • Fuel Tanks
  • Language
  • Literature Surveys
  • Operating Systems
  • Parallel Computing
  • Parallel Processing
  • Parallel Processors
  • Probability
  • Programming Languages
  • Reliability

Fields of Study

  • Computer science

Readers

  • Computational Modeling and Simulation
  • Parallel and Distributed Computing.