Discrete-event simulation on the bulk-synchronous parallel model

The bulk-synchronous parallel (BSP) model of computing has been proposed to enable the development of portable software which achieves scalable performance across diverse parallel architectures. A number of applications of computing science have been demonstrated to be e ciently supported by the BSP model in practice. In this thesis we study the e cient realisation of parallel discrete-event simulation (PDES) on the BSP model. We focus on the design, analysis and implementation of BSP synchronisation protocols for PDES. We show that it is indeed possible to achieve e cient performance in important classes of irregular systems. Our practical results enable the development of portable and scalable parallel-simulation kernels for discrete-event simulation. The topics studied involve the (i) e cient administration of the pending events arising in large scale simulations (sequentially and in parallel), (ii) analysis of the relevant factors leading to e cient and scalable PDES on BSP computers, (iii) development of optimistic synchronisation protocols which adaptively tune their operation to the evolution of the simulation work-load, (iv) comparative evaluation of optimistic and conservative synchronisation protocols, and (v) e cient parallel simulation of a class of complex systems. We contribute to each topic with new insights and techniques. The main result of this thesis is a new optimistic synchronisation protocol for general purpose PDES which we call BSP Time Warp.