Unified parallel algorithm for grid adaptation on a multiple-instruction multiple-data architecture

A unified parallel algorithm for grid adaptation by local refinement/coarsening is presented. It is designed to be independent from the type of the grid. This is achieved by employing a generic data template that can be configured to capture the data structures for any computational grid regardless of structure and dimensionality. Furthermore, the algorithm itself is specified in terms of generic parallel primitives that are completely independent of the underlying parallel architecture. The unified parallel algorithm is employed for dynamic adaptation of three-dimensional unstructured tetrahedral grids on a partitioned memory multiple-instruction multiple-data architecture. Performance results are presented for the Intel iPSC/860.