Performance testing of a parallel multiblock CFD solver

A distributed-memory version of the OVERFLOW computational fluid dynamics c ode was evaluated on several parallel systems and compared with other appro aches using test cases provided by the NASA-Boeing High-Speed Civil Transpo rt program. A principal goal was to develop partitioning and load-balancing strategies that led to a reduction in computation time. We found multipart itioning, in which the aspect ratio of the multipartition is close to the a spect ratio of the grid zone, offered the best performance. The (uniprocess or) performance of the GRAY vector systems was superior to all other system s tested. However, the distributed-memory version when run on an SGI Origin system offers a price performance advantage over the GRAY vector systems. Performance on personal computer systems is promising but faces several hur dles.