| Abstract |
|---|
| The ability to dynamically adapt an unstructured mesh is a powerfultool for solving computational problems with evolving features;however, an efficient parallel implementation is rather difficult.We have developed an automatic framework, called PLUM, to addressthis problem. PLUM requires that all data be globally redistributedafter each mesh adaption to achieve load balance. We present an algorithm for minimizing this remapping overhead, and demonstratethat the data redistribution cost can be significantly reduced by applying it to the default mapping of the parallel partitioner. Portability is examined by comparing performance on a SP2, an Origin2000, and a T3E. |
| Contact |
| Rupak Biswas NASA Ames Research Center,Mail Stop T27A-1,Moffett Field, CA 94035-1000,U.S.A. rbiswas@nas.nasa.gov |