A DYNAMIC MPI-OPENMP MODEL FOR STRUCTURED ADAPTIVE MESH REFINEMENT
Abstract
We compare experimentally different parallelization models using MPI and OpenMP for structured adaptive mesh refinement on a shared-memory parallel computer, a SunFire 15K. Due to the dynamic properties of the mesh no static parallelization model with fixed number of processes and threads performs best in all stages. Different combinations of MPI and OpenMP are preferable in different settings of the application and grid hierarchy. We suggest a new dynamic approach using a mixed MPI-OpenMP model that adapts the number of threads during run time and gives good performance in all stages throughout the whole run as the solution state changes, i.e. the resolution in the computational grid changes.
References
- Parallel Computing 27, 37 (2001). Crossref, ISI, Google Scholar
- Journal of Computational Physics 82, 64 (1989). Crossref, ISI, Google Scholar
- Scientific Programming 9, 185 (2001). Crossref, ISI, Google Scholar
- K. Devine et al., Design of dynamic load-balancing tools for parallel applications, technical report, Sandia National Laboratories, Albuquerque, NM, USA, 2000 . Google Scholar
- Journal of Parallel Algorithms and Applications 17(4), 285 (2002). Crossref, Google Scholar
- M. A. Iqbal, J. H. Saltz, S. H. Bokhari, Performance tradeoffs in static and dynamic load balancing strategies, Technical Report 86-13, ICASE, NASA Langley Research Center, Hampton, VA, 1986 . Google Scholar
- G. Karypis, K. Schloegel, and V. Kumar, PARMETIS - parallel graph partitioning and sparse matrix ordering library, version 2.0, University of Miuuesota, Minneapolis, MN, 1998 . Google Scholar
- Z. Lan, V. Taylor, and G. Bryan, Dynamic load balancing for structured adaptive mesh refinement applications, in proceediugs of ICPP 2001 . Google Scholar
- H. Löf, M. Nordén and S. Holmgren. Improving Geographical Locality of Data for Shared Memory Implementations of PDE Solvers, In Proc. of ICCS 2004, part II, LNCS 3037, pages 9-16, 2004 . Google Scholar
- BIT 42(1), 129 (2002). Google Scholar
- Computer physics communications 330 (2000). Google Scholar
- Journal of Parallel aud Distributed Computing 52(2), 150 (1998). Crossref, ISI, Google Scholar
- M. Parashar, J. Browne, System engineering for high performance computing software: The HDDA/DAGH infrastructure for implementation of parallel structured adaptive mesh refinement, IMA Volume on Structured Adaptive Mesh Refinement (SAMR) Grid Methods, pages 1-18, 2000 . Google Scholar
- J. Rantakokko, Comparison of Parallelization Models for Structured Adaptive Mesh Refinement, in proceedings of Europar 2004 . Google Scholar
- K. Schloegel, G. Karypis, and V. Kumar, Multilevel diffusion schemes for repartitioning of adaptive meshes, technical report 97-013, Department of Computer Science, University of Minnesota, Minneapolis, MN, 1998 . Google Scholar
- Computing Systems in Engineering 1(2/3), 135 (1991). Google Scholar
- Scientific Programming 9, 83 (2001). Crossref, ISI, Google Scholar
- J. Steensland, Efficient Partitioning of Dynamic Structured Grid Hierarchies, Department of Information Technology, Uppsala Uuiversity, Sweden, Ph.D. thesis, 2002 . Google Scholar
- , Proceedings of PARA2000, New Paradigms for HPC in Industry and Academia, 5th International Workshop, PARA2000,
Lecture Notes in Computer Science, No. 1947 , eds.T. Sorevik (Springer-Verlag, Berlin, 2001) pp. 160–169. Crossref, Google Scholar - S. Söderberg, A Parallel Block-Based PDE Solver with Space-Time Adaptivity, Department of Information Technology, Uppsala University, Sweden, Licentiate thesis, 2001 . Google Scholar
- , Proceedings of PARA2000, New Paradigms for HPC in Industry and Academia, 5th International Workshop, PARA2000,
Lecture Notes in Computer Science, No. 1947 , eds.T. Sorevik (Springer-Verlag, Berlin, 2001) pp. 177–186. Crossref, Google Scholar - Journal of Parallel and Distributed Computing 47(2), 102 (1997). Crossref, ISI, Google Scholar


