Efficient distributed simulation through dynamic load balancing

With recent advances in parallel computation, distributed simulation has be come a viable way of dealing with time-consuming simulations. For distribut ed simulations to run efficiently, care must be taken in assigning the task s (work) in the simulated system to the available physical processors in th e computer system. An inefficient assignment can result in excessive commun ication times between processors and unfavorable load conditions. This lead s to long run times, possibly giving performance worse than that with a uni processor sequential event-list implementation. This paper establishes the feasibility, and in some cases the necessity, of using dynamic task allocat ion (rather than a-priori static allocation) in distributed simulation. A d ynamic reallocation strategy is developed, and experiments on an iPSC/2 Hyp ercube indicate that significant improvements in run time can be achieved a t low cost.