cient Load Balancing and Data Remapping for Adaptive Grid

Mesh adaption is a powerful tool for eecient unstructured-grid computations but causes load imbalance among processors on a parallel machine. We present a novel method to dynamically balance the processor workloads with a global view. This paper presents, for the rst time, the implementation and integration of all major components within our dynamic load balancing strategy for adaptive grid calculations. Mesh adaption, repartitioning, processor assignment , and remapping are critical components of the framework that must be accomplished rapidly and eeciently so as not to cause a signiicant overhead to the numerical simulation. Previous results indicated that mesh repartitioning and data remapping are potential bottlenecks for performing large-scale scientiic calculations. We resolve these issues and demonstrate that our framework remains viable on a large number of processors. 1 INTRODUCTION Dynamic mesh adaption on unstructured grids is a powerful tool for computing unsteady three-dimensional problems that require grid modiications to eeciently resolve solution features. By locally reening and coarsening the mesh to capture owweld phenomena of interest, such procedures make standard computational methods more cost eeective. Highly reened meshes are required to accurately capture shock waves, contact discontinuities, vortices, and shear layers. Local mesh adaption provides the opportunity to obtain solutions that are comparable to those obtained on globally-reened grids but at a much lower cost. Unfortunately, the adaptive solution of unsteady problems causes load imbalance among processors on a parallel machine. This is because the computational intensity is both space and time dependent. An eecient parallel implementation of such methods is extremely diicult to achieve, primarily because of the dynamically-changing nonuniform grid. Various methods on dynamic load balancing have been reported to date 5-9,11-14,16-18,24-26]; however, most of them either lack a global view of loads across processors or