Performance of a Sequential and Parallel Computational Fluid Dynamic (CFD) Solver on a Missile Body Configuration

摘要

In order to effectively port production codes originally written for vector processors to reduced instruction set (RISC) based parallel computers, different paradigms have been tried by the parallel computing community. Among the techniques used are message passing and loop level parallelization using hand inserted compiler directives or automatic parallelizing compiler flags. The goals of this report are: (1) to investigate the performance of message passing and loop level parallelization techniques, as they were implemented in the computational fluid dynamics (CFD) code overflow, and (2) to validate the sequential and parallel results obtained on a demonstration problem of interest to the Army that is, a generic missile body configuration. The computational simulations were run, and performance data were gathered on a Silicon Graphics Incorporated (SGI) Power Challenge Array (PCA) and Origin 2000.