作者: Charbel Farhat , Stéphane Lanteri
DOI: 10.1016/0045-7825(94)00075-1
关键词:
摘要: Abstract We report here on our effort in simulating unsteady viscous flows the iPSC-860, CM-5 and KSR-1 MPPs (Massively Parallel Processors), using a Monotonic Upwind Scheme for Conservation Laws finite volume/finite element method fully unstructured fixed moving grids. advocate mesh partitioning with message passing as portable paradigm parallel processing. present discuss several performance results obtained all three MPP systems terms of interprocessor communication costs, I/O, scability sheer performance.