Optimization of a Three-Dimensional Diagnostic Flow Solver

Abstract

In this study, we investigate the performance of a three-dimensional diagnostic flow solver. The numerical solver is profiled to find the "hot spots" of the code. These hot spots are then optimized for performance using techniques such as loop reordering and cache blocking. OpenMP is used to parallelize the code execution, further enhancing its performance. Finally, preliminary efforts to use NVIDIA Inc. Compute Unified Device Architecture (CUDA) programming model are discussed.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 01, 2010
Accession Number
ADA564452

Entities

People

  • Chatt C. Williamson
  • Yansen Wang

Organizations

  • United States Army Research Laboratory

Tags

Communities of Interest

  • Weapons Technologies

DTIC Thesaurus Topics

  • Algorithms
  • Central Processing Units
  • Compilers
  • Computational Fluid Dynamics
  • Computational Science
  • Computations
  • Computer Programming
  • Computer Programs
  • Computer Science
  • Computers
  • Differential Equations
  • Graphics Processing Unit
  • Hot Spots
  • Linear Systems
  • Operating Systems
  • Optimization
  • Three Dimensional

Readers

  • Finite Element Method (FEM) for solving Partial Differential Equations (PDEs)
  • Parallel and Distributed Computing.
  • Thin Film Deposition Science.