PARALLEL 3-DIMENSIONAL MESH GENERATION ON DISTRIBUTED-MEMORY MIMD COMPUTERS

This paper discusses the development of an automatic mesh generation t echnique designed to operate effectively on multiple instruction multi ple data (MIMD) parallel computers. The meshing approach is hierarchic al, that is, model entities are meshed after their boundaries have bee n meshed. Focus is on the region meshing step. An octree is constructe d to serve as a localization tool and for efficiency. The tree is also key to the efficient parallelization of the meshing process since it supports the distribution of load to processors. The parallel mesh gen eration procedure repartitions the domain to be meshed and applies on processor face removals until all face removals with local data have b een performed. The portion of the domain to be meshed remaining is dyn amically repartitioned at the octant level using an Inertial Recursive Bisection method and local face removals are reperformed. Migration o f a terminal octant involves migration of the octant data and the octa nt's mesh faces and/or mesh regions. Results show relatively good spee d-ups for parallel face removals on small numbers of processors. Once the three-dimensional mesh has been generated, mesh regions may be sca ttered across processors. Thererfore, a final dynamic repartitioning s tep is applied at the region level to produce a partition ready for fi nite element analysis.