Techniques to improve the performance of large-scale discrete-event simulation
Simulation;GPU
Swenson, Brian Paul ; Riley, George F. Electrical and Computer Engineering Blough, Douglas M. Lanterman, Aaron D. Beyah, Raheem A. Fujimoto, Richard F. ; Riley, George F.
Discrete-event simulation is a commonly used technique to model changeswithin a complex physical systems as a series of events that occur at discrete pointsof time. As the complexity of the physical system being modeled increases, thesimulator can reach a point where it is no longer feasible for it to run efficiently on onecomputing resource. A common solution is to break the physical system into multiplelogical processes. When breaking a simulation over multiple computing nodes, caremust be taken to ensure the results obtained are the same as would be obtained froma non-distributed simulation. This is done by ensuring that the events processedin each individual logical process are processed in chronological order. The task iscomplicated by the fact that the computing nodes will be exchanging timestampedmessages and will often be operating at different points of simulation time. Therefore,highly efficient synchronization methods must be used. It is also important that thelogical processes have a capable means to transport messages among themselves orthe benefits of parallelization will be lost.The objective of this dissertation is to design, develop, test, and evaluate tech-niques to improve the performance of large-scale discrete-event simulations. Thetechniques include improvements in messaging passing, state management, and timesynchronization. Along with specific implementation improvements, we also examinetechniques on how to effectively make use of resources such as shared memory andgraphical processing units.
【 预 览 】
附件列表
Files
Size
Format
View
Techniques to improve the performance of large-scale discrete-event simulation