Issue No.05 - May (2012 vol.23)
Alfred J. Park , Oak Ridge National Laboratory, Oak Ridge
Richard M. Fujimoto , Georgia Institute of Technology, Atlanta
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/TPDS.2011.207
The master/worker (MW) paradigm can be used as an approach to parallel discrete event simulation (PDES) on metacomputing systems. MW PDES applications incur overheads not found in conventional PDES executions executing on tightly coupled machines. We introduce four optimization techniques in MW PDES systems on public resource and desktop grid infrastructures. Work unit caching, pipelined state updates, expedited message delivery, and adaptive work unit scheduling mechanisms in the context of MW PDES are described. These optimizations provide significant performance benefits when used in tandem. We present results showing that an optimized MW PDES system using these techniques can exhibit performance comparable to a traditional PDES system for queueing network and particle physics simulation applications while providing execution capability across metacomputing systems.
Discrete event simulation, metacomputing systems, parallel and distributed simulation, simulation support systems, master/worker.
Alfred J. Park, Richard M. Fujimoto, "Efficient Master/Worker Parallel Discrete Event Simulation on Metacomputing Systems", IEEE Transactions on Parallel & Distributed Systems, vol.23, no. 5, pp. 873-880, May 2012, doi:10.1109/TPDS.2011.207