Performance analysis and portability of the PLUM load balancing system
The ability to dynamically adapt an unstructured mesh is a powerful tool for solving computational problems with evolving physical features; however, an efficient parallel implementation is rather difficult. To address this problem, we have developed PLUM, an automatic portable framework for performing adaptive numerical computations in a message-passing environment. PLUM requires that all data be globally redistributed after each mesh adaption to achieve load balance. We present an algorithm for minimizing this remapping overhead by guaranteeing an optimal processor reassignment. We also show that the data redistribution cost can be significantly reduced by applying our heuristic processor reassignment algorithm to the default mapping of the parallel partitioner. Portability is examined by comparing performance on a SP2, an Origin2000, and a T3E. Results show that PLUM can be successfully ported to different platforms without any code modifications.
KeywordsSimilarity Matrix Unstructured Mesh Dual Graph Edge Cost Helicopter Rotor
Unable to display preview. Download preview PDF.
- 1.Bhat, K.: An O(n 2.5 log2 n) time algorithm for the bottleneck assignment problems. AT&T Bell Laboratories Unpublished Report (1984)Google Scholar
- 2.Biswas, R., Oliker, L.: Experiments with repartitioning and load balancing adaptive meshes. NASA Ames Research Center Technical Report NAS-97-021 (1997)Google Scholar
- 6.Karypis, G., Kumar, V.: Parallel multilevel k-way partitioning scheme for irregular graphs. University of Minnesota Technical Report 96-036 (1996)Google Scholar
- 7.Oliker, L., Biswas, R.: PLUM: Parallel load balancing for adaptive unstructured meshes. NASA Ames Research Center Technical Report NAS-97-020 (1997)Google Scholar
- 8.Oliker, L., Biswas, R., Strawn, R.: Parallel implementation of an adaptive scheme for 3D unstructured grids on the SP2. Springer-Verlag LNCS 1117 (1996) 35–47Google Scholar