The most popular conservative parallel simulation approach is the Chandy-Misra (CM) approach. In this approach, two waiting rules are used to synchronize the executions of processes. This paper studies the effects of waiting overheads via a simulation model. The experiments indicate the following. (i) It is more important to reduce the waiting overheads for a lightly loaded system than a heavily loaded system. (ii) Reducing the input waiting overhead is more important than reducing the output waiting overhead for a network where processes have many input channels. (iii) For most cases, the interaction between the input optimization and the output optimization is positive. The authors also study the effects of waiting overheads on CM simulation with limited processors. It may not be worthwhile to reduce waiting overheads for simulation applications with high potential concurrency. On the other hand, it is important to use optimizations to reduce waiting overheads for simulation applications with low potential concurrency. For a small number of processors, it may not be worthwhile to reduce waiting overheads, because the concurrency provided by CM simulation is large enough to keep the processors busy. For a large number of processors, it is important to reduce waiting overheads to prevent processors from extra blocking.<<ETX>>
[1]
K. Mani Chandy,et al.
Distributed Simulation: A Case Study in Design and Verification of Distributed Programs
,
1979,
IEEE Transactions on Software Engineering.
[2]
David M. Nicol.
Parallel discrete-event simulation of FCFS stochastic queueing networks
,
1988,
PPoPP 1988.
[3]
Edward D. Lazowska,et al.
Parallel simulation of queueing networks: limitations and potentials
,
1989,
SIGMETRICS '89.
[4]
K. M. Chandy,et al.
Conditional Knowledge as a Basis for Distributed Simulation
,
1987
.
[5]
Y.-B. Lin,et al.
Exploiting Lookahead in Parallel Simulation
,
1990,
IEEE Trans. Parallel Distributed Syst..
[6]
Edward D. Lazowska,et al.
The limited performance benefits of migrating active processes for load sharing
,
1988,
SIGMETRICS 1988.
[7]
Wayne M. Loucks,et al.
Null Message Cancellation in Conservative Distributed Simulation
,
1991
.
[8]
Richard M. Fujimoto,et al.
Performance Measurements of Distributed Simulation Strategies.
,
1987
.