1:30 PM-2:30 PM
Chair: Petter E. Bjørstad, University of Bergen, Norway
Experience with several different types of parallel machines over the last decade in performing very large simulations of turbulent compressible fluid flows will be discussed with an emphasis on the parallel computing lessons learned. Ease of use, power, and flexibility of parallel vector, distributed memory, SMP cluster, and distributed shared memory (DSM) machines and the associated programming paradigms will be contrasted. Effective techniques for "embarassingly parallel" applications are by now well understood, and sample fluid flow simulations illustrating their power in addressing fluid flow problems of scientific interest will be presented. The challenge of implementing dynamically irregular computations on parallel machines is, however, a principal focus of current work. The special potential of SMP cluster and DSM computer architectures in addressing these new challenges will be discussed.
Paul R. Woodward
University of Minnesota, Minneapolis
PP97 Homepage | Program Updates|
Registration | Hotel Information | Transportation | Program-at-a-Glance