The Second Approach to Improving the Performance of the Lagrangian Particle Dispersion Model (LPDM) Using Graphics Processing Unit (GPU) Computing

Abstract

Certain atmospheric transport and dispersion models can take many hours to complete execution, such as the Lagrangian Particle Dispersion Model (LPDM) that simulates the ensemble average transport of aerosols and gases in turbulent wind conditions. The execution time needs to be reduced for these models, so they can be useful for rapid release and planning purposes. Performance improvements were achieved during the initial approach by using one technique of graphics processing unit (GPU) computing technology. This report documents the second approach of the implementation and integration processes using GPU computing technology to improve the LPDM codes performance. The execution time of GPU-accelerated LPDM application implementing the second approach was faster than both the original LPDM application without GPU computing and the GPU-accelerated LPDM application implementing the initial approach.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 2018
Accession Number
AD1059392

Entities

People

  • Leelinda P. Dawson
  • Yansen Wang

Organizations

  • United States Army Research Laboratory

Tags

DTIC Thesaurus Topics

  • Atmospheric Motion
  • Computer Programming
  • Computer Programs
  • Department Of Defense
  • Dispersions
  • Graphics
  • Graphics Processing Unit
  • Information Operations
  • Information Science
  • Military Research
  • Normal Distribution
  • Particles
  • Random Number Generators
  • Standards

Fields of Study

  • Computer science

Readers

  • Aerosol Science/Aerosol Physics
  • Computational Fluid Dynamics (CFD)
  • Parallel and Distributed Computing.