Performance Analysis of Domain Decomposition Applications Using Unbalanced Strategies in Grid Environments

In this paper, we compare the performance of some mesh-based applications in a Grid environment using the domain decomposition technique and unbalanced workload strategies. We propose unbalanced distributions in order to overlap computation with remote communications. Results are presented for typical cases in car crashing simulation where finite element schemes are applied in fine mesh. The expected execution time is basically the same when two unbalanced techniques are used, but it is up 34% smaller that the one requires by the classical balanced strategy. We also analyze the influence of the communication pattern on execution times using the Dimemas simulator.

[1]  Shahid H. Bokhari,et al.  A Partitioning Strategy for Nonuniform Problems on Multiprocessors , 1987, IEEE Transactions on Computers.

[2]  William Gropp,et al.  Latency, bandwidth, and concurrent issue limitations in high-performance CFD. , 2000 .

[3]  Brian Vinter,et al.  The Impact on Latency and Bandwidth for a Distributed Shared Memory System Using a Gigabit Network Supporting the Virtual Interface Architecture , 2000 .

[4]  Michelle R. Hribar,et al.  Balancing load versus decreasing communication: exploring the tradeoffs , 1996, Proceedings of HICSS-29: 29th Hawaii International Conference on System Sciences.

[5]  Michael M. Resch,et al.  Towards Efficient Execution of MPI Applications on the Grid: Porting and Optimization Issues , 2003, Journal of Grid Computing.

[6]  Rupak Biswas,et al.  HARP: a fast spectral partitioner , 1997, SPAA '97.

[7]  Jesús Labarta,et al.  A Domain Decomposition Strategy for GRID Environments , 2004, PVM/MPI.

[8]  G. Voelker,et al.  On the scale and performance of cooperative Web proxy caching , 2000, OPSR.

[9]  Thomas Ertl,et al.  Visualization and pre-processing of independent finite-element meshes for car crash simulations , 2002, The Visual Computer.

[10]  Jack Dongarra,et al.  Recent Advances in Parallel Virtual Machine and Message Passing Interface, 15th European PVM/MPI Users' Group Meeting, Dublin, Ireland, September 7-10, 2008. Proceedings , 2008, PVM/MPI.

[11]  Jesús Labarta,et al.  Validation of Dimemas Communication Model for MPI Collective Operations , 2000, PVM/MPI.

[12]  Matthias S. Müller,et al.  Performance Prediction in a Grid Environment , 2003, European Across Grids Conference.