Parallel and distributed simulation: fast cell level ATM network simulation

ABSTRACT

This paper presents performance results for cell level ATM network simulations using both sequential and parallel discrete event simulation kernels. Five benchmarks are used to demonstrate the performance of the simulation kernels for different types of model. The results demonstrate that for the type of network models used in the benchmarks, the TasKit simulation kernel is able to outperform all of the other kernels tested both sequentially and in parallel. For one benchmark TasKit is shown to outperform a conventional sequential simulation kernel by a factor of 3. For the same benchmark TasKit is shown to outperform the best of the other parallel kernels tested by a factor of 6. The paper explains how this performance advantage is achieved and cautions that additional research into automatic model partitioning will be essential to make this technology accessible to the general simulation community.

REFERENCES

• 1: R. E. Bryant, SIMULATION OF PACKET COMMUNICATION ARCHITECTURE COMPUTER SYSTEMS, Massachusetts Institute of Technology, Cambridge, MA, 1977
• 2: Chandy K. M. and J. Misra. 1979. Distributed simulation: A case study in design and verification of distributed simulation. IEEE Transactions on Software Engineering, 5(5):440--452.
• 3: John Cleary , Fabian Gomes , Brian Unger , Zhonge Xiao , Raimar Thudt, Cost of state saving & rollback, Proceedings of the eighth workshop on Parallel and distributed simulation, p.94-101, July 06-08, 1994, Edinburgh, Scotland, United Kingdom
• 4: John G. Cleary , Jya-Jang Tsai, Performance of a conservative simulator of ATM networks, Proceedings of the eleventh workshop on Parallel and distributed simulation, p.142-145, June 10-13, 1997, Lockenhaus, Austria
• 5: Richard M. Fujimoto, Time warp on a shared memory multiprocessor, Transactions of the Society for Computer Simulation International, v.6 n.3, p.211-239, Jul. 1989
• 6: Fabian Gomes , John Cleary , Alan Covington , Steve Franks , Brian Unger , Zhong-e Ziao, SimKit: a high performance logical process simulation class library in C++, Proceedings of the 27th conference on Winter simulation, p.706-713, December 03-06, 1995, Arlington, Virginia, United States [doi>10.1145/224401.224714]
• 7: David R. Jefferson, Virtual time, ACM Transactions on Programming Languages and Systems (TOPLAS), v.7 n.3, p.404-425, July 1985 [doi>10.1145/3916.3988]
• 8: Daniel Dominic Sleator , Robert Endre Tarjan, Self-adjusting binary search trees, Journal of the ACM (JACM), v.32 n.3, p.652-686, July 1985 [doi>10.1145/3828.3835]
• 9: Brian Unger , Zhonge Xiao , John Cleary , Jya-Jang Tsai , Carey Williamson, Parallel shared-memory simulator performance for large ATM networks, ACM Transactions on Modeling and Computer Simulation (TOMACS), v.10 n.4, p.358-391, Oct. 2000 [doi>10.1145/369534.369537]
• 10: Xiao Z., and B. Unger. 1995. Report on warpkit - performance study and improvement. Technical Report Technical Report 98-628-19, Computer Science Department, University of Calgary.
• 11: Z. Xiao , B. Unger , R. Simmonds , J. Cleary, Scheduling critical channels in conservative parallel discrete event simulation, Proceedings of the thirteenth workshop on Parallel and distributed simulation, p.20-28, May 01-04, 1999, Atlanta, Georgia, United States