论文信息 - Libraries and Methods for Parallel Particle Simulations

Libraries and Methods for Parallel Particle Simulations

We used the high performance systems at HLRS in the last year for two different works. One was the hybrid implementation and optimization of ParaSPH a library for the parallel computation of particle simulations. We give details of the parallelization of ParaSPH for hybrid architectures (clustered SMPs) using MPI and OpenMP and discuss performance results for single node and speedups of the code on the Opteron Cluster and the NEC SX-6. The other work we present is the enhancement of an object-oriented framework for messages passing called TPO++ to support object-oriented parallel I/O.

[1] Rajeev Thakur,et al. Optimizing noncontiguous accesses in MPI-IO , 2002, Parallel Comput..

[2] Rajeev Thakur,et al. Users guide for ROMIO: A high-performance, portable MPI-IO implementation , 1997 .

[3] Erik Asphaug,et al. Impact Simulations with Fracture. I. Method and Tests , 1994 .

[4] Andrew A. Chien,et al. I/O requirements of scientific applications: an evolutionary view , 1996, Proceedings of 5th IEEE International Symposium on High Performance Distributed Computing.

[5] Wolfgang Rosenstiel,et al. Parallel I/O in an Object-Oriented Message-Passing Library , 2004, PVM/MPI.

[6] Jun Qin,et al. Parallel Motif Search using ParSeq , 2005, Parallel and Distributed Computing and Networks.

[7] W Rosenstiel,et al. ParSeq: searching motifs with structural and biochemical properties. , 2004, Bioinformatics.

[8] L. Libersky,et al. High strain Lagrangian hydrodynamics: a three-dimensional SPH code for dynamic material response , 1993 .

[9] L. Libersky,et al. Smoothed Particle Hydrodynamics: Some recent improvements and applications , 1996 .