The author characterizes the speedup of parallel discrete event simulation (PDES) programs relative to their concomitant critical path estimations. The ability of a parallel implementation to outperform the critical path is related to the level of intra-object event independence in the model. Based on a formalization of intra-object event independence, Gunter describes a new approach to calculating the critical path of a simulation and shows that it provides a tighter lower bound approximation than previously known techniques.
The paper is technically sound and provides insight into a fundamental problem for PDES. The presentation is somewhat terse, however, and assumes the reader is familiar with several basic PDES premises, such as state space partitioning in model development. The interested reader might also refer to Lin [1] for a more detailed treatment of the concepts.