This Article 
   
 Share 
   
 Bibliographic References 
   
 Add to: 
 
Digg
Furl
Spurl
Blink
Simpy
Google
Del.icio.us
Y!MyWeb
 
 Search 
   
Performance Analysis of Time Warp with Multiple Homogeneous Processors
October 1991 (vol. 17 no. 10)
pp. 1013-1027

The behavior of n interacting processors synchronized by the Time Warp protocol is analyzed using a discrete-state, continuous-time Markov chain model. The performance and dynamics of the processes (or processors) are analyzed under the following assumptions: exponential task times and timestamp increments on messages, each event message generates one new message that is sent to a randomly selected process, negligible rollback, state saving, and communication delay, unbounded message buffers, and homogeneous processors. Several performance measures are determined, such as: the fraction of processed events that commit, speedup, rollback probability, expected length of rollback, the probability mass function for the number of uncommitted processed events, the probability distribution function for the virtual time of a process, and the fraction of time the processors remain idle. The analysis is approximate, thus the results have been validated through performance measurements of a Time Warp testbed executing on a shared-memory multiprocessor.

[1] K.M. Chandy and J. Misra, "Asynchronous Distributed Simulation via a Sequence of Parallel Computations,"Comm. ACM, Vol. 24, No. 4, Apr. 1981, pp. 198-206.
[2] R. E. Felderman and L. Kleinrock, "An upper bound on the improvement of asynchronous versus synchronous distributed processing," inProc. SCS Multiconf. Distributed Simulation, Jan. 1990, vol. 22, no. 1, pp. 131-136.
[3] R. E. Felderman, "Speedup for large number of processors," private communication, Dec. 1990.
[4] R. E. Felderman and L. Kleinrock, "Two processor time warp analysis: some results on a unifying approach,"Proc. SCS Workshop on Parallel and Distributed Simulation, Jan. 1991, vol. 23, no. 1, pp. 3-10.
[5] R. M. Fujimoto, "Time warp on a shared memory multiprocessor,"Trans. Soc. Comput. Simul., vol. 6, no. 3, pp. 211-239, July 1989.
[6] R. M. Fujimoto, "Performance of time warp under synthetic workloads," inProc. SCS Multiconf. Distributed Simulation, Jan. 1990, vol. 22, no. 1, pp. 23-28.
[7] R. M. Fujimoto, "Parallel discrete event simulation,"Commun. ACM, vol. 33, no. 10, pp. 30-53, Oct. 1990.
[8] E. Gelenbe, "Product form networks with negative and positive customers,"J. Appl. Prob., to be published.
[9] A. G. Greenberg, B. D. Lubachevsky, and I. Mitrani, "Unboundedly parallel simulations via recurrence relations,"Proc. 1990 ACM SIGMETRICS Conf. on Measure. and Model. Comput. Syst., May 1990, vol. 18, no. 1, pp. 11-12.
[10] D. Jefferson, "Virtual Time,"ACM Trans. Programming Languages, Vol. 7, No. 3, July 1985, pp. 404-425.
[11] D. Jefferson and A. Witkowski, "An approach to performance analysis of timestamp-driven synchronization mechanism," inProc. 3rd ACM Ann. Symp. Principles Distributed Comput., 1984.
[12] L. Kleinrock, "On distributed systems performance,"Comput. Networks ISDN J., vol. 20, nos. 1-5, pp. 209-216, Dec. 1990.
[13] L. Lamport, "Time, clocks, and the ordering of events in a distributed system,"Commun. ACM, vol. 21, no. 7, pp. 558-565, July 1978.
[14] S. Lavenberg, R. Muntz, and B. Samadi, "Performance analysis of a rollback method for distributed simulation,"Performance '83. Amsterdam: North-Holland, 1983, pp. 117-132
[15] Y. B. Lin and E. D. Lazowska, "Optimality considerations for "time warp" parallel simulation," inProc. 1990 SCS Multiconf. Distributed Simulation, Jan. 1990, pp. 29-34.
[16] R. J. Lipton and D. W. Mizell, "Time warp vs. Chandy-Misra: a worst-case comparison," inProc. SCS Multiconf. Distributed Simulation, Jan. 1990, vol. 22, no. 1, pp. 137-143.
[17] B. Lubachevsky, A. Shwartz, and A. Weiss, "Rollback sometimes works ... if filtered," inProc. 1989 Winter Simulation Conf., Dec. 1989, pp. 630-639.
[18] V. Madisetti, J. Walrand, and D. Messerschmitt, "Synchronization in message passing computers: models, algorithms, and analysis," inProc. SCS Multiconf. Distributed Simulation, Jan. 1990, vol. 22, no. 1, pp. 35-48.
[19] J. Misra, "Distributed discrete-event simulation,"Comput. Surveys, vol. 18, no. 1, Mar. 1986.
[20] D. Mitra and I. Mitrani, "Analysis and optimum performance of two-message passing parallel processors synchronized by rollback,"Performance Evaluation J., vol. 7, pp. 111-124, 1987.
[21] D. M. Nicol, "Performance bounds on parallel self-initiating discrete-event simulations,"ACM Trans. Model. Comput. Simulation, vol. 1, no. 1, pp. 24-50, Jan. 1991.
[22] B. D. Plateau and S. K. Tripathi, "Performance analysis of synchronization for two communicating processes,"Performance Evaluation J., vol. 8, pp. 305-320, 1988.

Index Terms:
parallel simulation; interacting processors; Time Warp protocol; discrete-state; continuous-time Markov chain model; exponential task times; timestamp increments; event message; negligible rollback; state saving; communication delay; unbounded message buffers; homogeneous processors; performance measures; processed events; speedup; rollback probability; probability mass function; uncommitted processed events; probability distribution function; virtual time; Time Warp testbed; shared-memory multiprocessor; discrete event simulation; Markov processes; multiprocessing systems; performance evaluation; protocols
Citation:
A. Gupta, I.F. Akyildiz, R.M. Fujimoto, "Performance Analysis of Time Warp with Multiple Homogeneous Processors," IEEE Transactions on Software Engineering, vol. 17, no. 10, pp. 1013-1027, Oct. 1991, doi:10.1109/32.99190
Usage of this product signifies your acceptance of the Terms of Use.