论文信息 - Parallel Multigrid Solvers Using OpenMP/MPI Hybrid Programming Models on Multi-Core/Multi-Socket Clusters

Parallel Multigrid Solvers Using OpenMP/MPI Hybrid Programming Models on Multi-Core/Multi-Socket Clusters

OpenMP/MPI hybrid parallel programming models were implemented to 3D finite-volume based simulation code for groundwater flow problems through heterogeneous porous media using parallel iterative solvers with multigrid preconditioning. Performance and robustness of the developed code has been evaluated on the "T2K Open Supercomputer (Tokyo)" and "Cray- XT4" using up to 1,024 cores through both of weak and strong scaling computations. OpenMP/MPI hybrid parallel programming model demonstrated better performance and robustness than flat MPI with large number of cores for illconditioned problems with appropriate command lines for NUMA control, first touch data placement and reordering of the data for contiguous "sequential" access to memory.

Kengo Nakajima

[1] Kengo Nakajima. Parallel multilevel iterative linear solvers with unstructured adaptive grids for simulations in earth science , 2002, Concurr. Comput. Pract. Exp..

[2] Barry F. Smith,et al. Domain Decomposition: Parallel Multilevel Methods for Elliptic Partial Differential Equations , 1996 .

[3] Timothy G. Mattson,et al. Patterns for parallel programming , 2004 .

[4] Clayton V. Deutsch,et al. GSLIB: Geostatistical Software Library and User's Guide , 1993 .

[5] Kengo Nakajima,et al. Flat MPI vs. Hybrid: Evaluation of Parallel Programming Models for Preconditioned Iterative Solvers on “T2K Open Supercomputer” , 2009, 2009 International Conference on Parallel Processing Workshops.