A Data Parallel Implementation of the BRL-Q1D Code.

Abstract

Massively parallel processing computer technology offers the potential to perform complex three dimensional numerical simulations of blast events in a production environment. In an attempt to evaluate this technology in simulating blast phenomena, the U.S. Army Ballistic Research Laboratory (BRL) adapted one of its blast modeling codes to the Thinking Machines CM-2 (and later CM-200) architecture. The code, which was adapted to the Connection Machine (CM), is BRL-01 D, a quasi-one dimensional, finite difference, single material, polytropic gas fluid dynamics code. This code is primarily used for simulating flow in shock tubes and blast simulators and has been reconfigured for the Connection Machine using the CM Fortran programming language. The performance of the code was measured on the CM for a variety of problem sizes and compiler options, and the performance was compared to that of the original Fortran 77 code on a single processor of a Cray X-MP. The optimum performance of the code on one eighth of a maximum configuration CM-200 is 2.5 times greater than that of the original code on a single processor of a Cray X-MP. The code also proved to be perfectly scalable on the CM architecture, allowing BRL to extrapolate a factor of 20 speedup over the Cray X-MP for a maximum configuration of CM-200. nuclear explosion simulation, shock tubes, computer programming.

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 1992
Accession Number
ADA254695

Entities

People

  • Stephen J. Schraml

Organizations

  • Ballistic Research Laboratory

Tags

DTIC Thesaurus Topics

  • Computer Programming
  • Computers
  • Fluid Dynamics
  • Materials
  • Nuclear Explosion Simulation
  • Nuclear Explosions
  • Parallel Computing
  • Parallel Processing
  • Programming Languages
  • Shock Tubes
  • Simulations
  • Simulators
  • Three Dimensional

Readers

  • Combustion Dynamics and Shock Wave Physics.
  • Computational Fluid Dynamics (CFD)
  • Computer Science.