Scalable Algorithms for Distributed-Memory Adaptive Mesh Refinement
International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD) 2012
Publication Type: Paper
This paper describes a novel mesh restructuring algorithm for adaptive mesh refinement computations that uses a constant number of collectives regardless of the refinement depth. To further increase scalability, we describe a localized hierarchical coordinate-based block indexing scheme in contrast to traditional linear numbering schemes, which incur unnecessary synchronization. With these optimizations as well as an efficient mapping scheme, our algorithm is scalable and suitable for large, highly-refined meshes. We present strong-scaling experiments up to 2048 ranks on Cray XK6, and 32768 ranks on IBM Blue Gene/Q.