Performance and Scalability Results for an Aggressive Global

Windowing algorithms represent an important class of synchronization protocols for parallel discrete event simulation. In these algorithms, a simulation window is chosen such that all events within the window can be executed concurrently without the possibility of a causality error. Using the terminolog of Chand and Sherman (1989), these are unconditional events. Windowing algorit ms, as al non-aggressive algorithms, have been criticized for not allowing a computation to proceed because there exists the possibility of a causality error. We are interested in the impact of extending the simulation window in order to allow the computation of conditional events, that is, those events that may cause an error. In this paper we develop a model to investigate the benefits of extending the simulation window to admit conditional events into the computation stream. Using this model we demonstrate significant performance gains as a result of aggressive processing. Also we prove that our approach is scalable: Performance is not significantly degraded as the number ofLPs approaches infinity. We validate these results with empirical studies. Performance and Scalability Results for an Aggressive Global Windowing Algorithm
Note: Abstract extracted from PDF file via OCR

Dickens, P, Jr Reynolds, and J Duva. "Performance and Scalability Results for an Aggressive Global." University of Virginia Dept. of Computer Science Tech Report (1992).