Comments on CFD code performance on scalable architectures

摘要

We comment on the current performance of computational fluid dynamics codes on a variety of scalable computer architectures. The performance figures are derived from both the finite-volume and finite-element methodologies, and encompass shared, virtual shared, and distributed memory architectures, as exemplified by the SGI Origin series, CM5, and the CRAY T3D/E family, respec- tively.