Characterizing the performance of OVERFLOW on linux beowulf architectures

Linux PC Clusters are a cost effective platform for parallel computational fluid dynamics (CFD) applications. In this paper we compare the performance of the OVERFLOW application on multiple clusters. Different networking architectures, from inexpensive switched Fast Ethernet to high performance networks like Myrinet interconnect, are considered in our performance evaluation. Likewise, a variety of node constructions are considered, from IA32 commodity processors to 64-bit MP (Multi-Processor) configurations. Single processor performance as well as the overall parallel performance using multiple sizes for benchmark cases are compared and evaluated to find the most cost effective parallel platform for OVERFLOW along with where the code's performance can likely be improved on these systems.