Deadlock detection and recovery based distributed simulation: a performance study

The authors present the results of an experimental performance study of distributed simulation of closed queuing networks using a minor variation of the deadlock detection and recovery based algorithm of K.M. Chandy and J. Misra (1979,1981). A major part of this study is to determine the effect of overhead on distributed simulation. Moreover, in these experiments the values of certain refined notions of ideal speedup ratio are measured. These ratios are more refined than just the number of processors, in the sense that they capture the potentially achievable speedups of distributed simulation more closely.<
>