POSE: Scalable General-purpose Parallel Discrete Event Simulation
Thesis 2005
Publication Type: PhD Thesis
Repository URL:
Parallel discrete event simulation (PDES) applications encompass a broad range of analytical simulations. Their utility lies in their ability to model a system under study and provide information about the behavior of that system in a timely manner. The most comprehensive models for such systems can be vastly complex, have highly irregular structures and fine granularity, making them challenging problems to parallelize. Current PDES methods provide limited performance improvements over sequential simulation and complicate the modelling process, requiring knowledge of specialized parallel computing practices that may be well outside the application developer's field.

We propose a novel environment for PDES that facilitates the development of highly parallel models and requires minimal understanding of parallel computing concepts. We propose four primary approaches to improving the performance of PDES. We first examine the overhead required for synchronizing events to obtain correct results in parallel and develop a new approach to the structure of model entities and mechanisms for PDES that help to reduce that overhead. Secondly, we design new adaptive synchronization strategies that exploit this new model structure to obtain better cache performance and reduce context switching overhead. We then develop techniques to optimize communication in concert with these new strategies. Finally, we study load balancing in the context of optimistic synchronization and design new approaches to fit with our other techniques. These four approaches form an integrated system for handling non-ideal simulation models. We demonstrate our techniques via a highly flexible synthetic benchmark capable of mimicking a variety of simulation behaviors, as well as with simulations of network models for very large parallel computers.

Wilmarth, Terry L., "POSE: Scalable General-purpose Parallel Discrete Event Simulation", Ph.D. Thesis, Department of Computer Science, University of Illinois at Urbana-Champaign, 2005.
Research Areas