Y. Lin. Parallel independent replicated simulation on a network of workstations. In PADS '94: Proceedings of the eighth workshop on Paral lel and distributed simulation, pages 73--80. ACM Press, 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....a single experiment. A single experiment could, for example, consist of a range of workload parameters. Overall experiment run times can be reduced by being able to run a single model on many machines with different input parameters for each individual run. Parallel Independent Replications (PIR) [5] of a simulation model enables time based parallelization. In timebased parallelization, identical models are distributed to multiple machines where each instance independently simulates a different sample path. The results from the independent runs can be combined into an experiment mean with a ....

....in Figure 8 for both the VOQs in the input ports and for the Cross Point (CP) buffers. Instability occurs when a switch cannot carry an offered load that is less than, or equal to, 1.0 (or 100 utilization) Instability will always occur of the offered load exceeds 1.0 for any output port. In [5] a region of instability for RR RR CICQ switches is demonstrated for a two port switch with unbalanced traffic load and offered load less than 1.0. All offered load is for packet of fixed lengths (called cells) and with a Bernoulli distribution (i.e. a given cell slot has a cell in it with ....

Y. Lin, "Parallel Independent Replicated Simulation on a Network of Workstations," Simulation, Vol. 64, No. 2, pp. 102-110, 1995.

....network this has shown a significant speed up in the simulation. 3. 2 Parallel simulation replica As an alternative to use more than one processor on each simulation run, we might as well divide our experiment into K subsections, and run each subsection on K different processors (e.g. workstation) [4,10,11]. This will give a speed up factor of K. This is as a much simpler and effective approach than the multiprocessor simulation. The main motivation is that in each simulation experiments you need more than one independent observation period, e.g. using the sectioning method and starting K separate ....

....(IS.bal) biasing are found in Tables 6 to 8. a.Exact value 2.968750046x10 7 b.Standard deviation of the mean estimator c.S CPU d.Ratio relative to the direct simulation e.No of regeneration cycles = 200,000 7 f.No of regeneration cycles = 20,000 Table 3 Case 1: N=4, l = 0. 05 Method mean a [10 7 ] S b [10 7 ] CPUsec (ELC) efficiency measure c , 10 7 ] speed up factor d Direct simulation e 11.31 5.984 106.2 635.5 1.0 RESTART f I opt 1=1 2.180 1.329 62.5 83.1 7.7 I opt =2 3.418 0.804 67.4 54.2 11.7 Transition splitting f 2.974 0.00660 35.5 0.23 2712.3 IS f balanced ....

[Article contains additional citation context not shown here]

Y.-B. Lin. "Parallel independent replicated simulation on networks of workstations. " In Proceedings of the 8th Workshop on Parallel and Distributed Simulation (PADS'94), pages 71 -- 81. IEEE Computer Society Press, July 6-8 1994.

....modelling efficiently and accurately. Another approach of parallelism applied to simulation is to run multiple serial simulation programs on multiple processors in parallel and average the results at the end of the runs. This approach is referred to as replicated serial simulation (RSS) 1] 4] [7], which belongs to a wide class of parallelism known as serial program, parallel subsystem (SPPS) 11] The major advantage of this approach is providing a simple implementation to reduce the overall turnaround time A shorter version will appear in the Proc. of European Conference on Parallel ....

....and Grains AB and inform remote CSs to update their status of the RPS. The next simulation run restarts when the grains receive a new grain to workstation mapping from the AB. 3. 3 Scheduling Policies of the RPSs Like the RSS, the first N replications initiated (FNI) scheduling method [1] 4] [7] is applied by the RPSs to obtain statistically accurate simulation results. In this scheduling policy, N results of simulation runs are recorded from the first N replications initiated. The value of N is determined by the termination condition of simulation, for example when a desirable ....

Y.-B. Lin. Parallel independent replicated simulation on a network of workstations. In Proc. of 8th Workshop on Parallel and Distributed Simulation, pages 73--80, 1994.

....and Kuhn 1987; Wagner and Lazowska 1989] have reported that typical speedups are only modest. Parallel simulation in which each processor simulates an independent realization of the stochastic process is an attractive alternative for steady state simulations [see e.g. Rego and Sunderam 1992; Lin 1994]. The attraction arises from the following two facts. Firstly, modern workstations connected through a local area network often have free processor capacity that could be utilized. Secondly, only minor modifications to uniprocessor simulation software are needed. The most notable drawback is that ....

Lin, Y.-B. 1994. Parallel Independent Replicated Simulation on a Network of Workstations. In Proc. of the 8th Workshop on Parallel and Distributed Simulation (PADS'94). ACM SIGSIM, 73--80.

.... factor is normally less than the number of processors, even with smart strategies for handling interactions [3, 4, 5, 6] A better way to exploit a multiprocessor environment is to distribute several independent replicas of the same process on different processors, e.g. a cluster of workstations [3,7,8]. 2. Hybrid technique: It combines the flexibility of simulation with the computational efficiency of analytic models. Hybrid techniques have reported substantial computational saving when applied to communication networks [9] 3. Variance reduction: Variance minimizing exploits the known ....

Y.-B. Lin. "Parallel independent replicated simulation on networks of workstations. " In Proceedings of the 8th Workshop on Parallel and Distributed Simulation (PADS'94), pages 71 -- 81. IEEE Computer Society Press, July 68 1994.

.... under study is usually large (e.g. thousands of cells) A typical sequential PCS simulation run takes over 20 hours, while the corresponding PCS PDES takes less than 3 hours using 8 processors [4] ffl Another popular parallel approach, the parallel independent replicated simulation [27] 28] [29] (running multiple simulation replications concurrently) does not work for PCS simulation. In most cases, the PCS designer only is interested in the behavior of the PCS network at the engineered workload (e.g. the workload at which the blocking probability is 1 ) To calibrate the simulation at ....

Lin, Y.-B., "Parallel Independent Replicated Simulation on A Network of Workstations", To appear in SIMULATION, 1995.

No context found.

Y. Lin. Parallel independent replicated simulation on a network of workstations. In PADS '94: Proceedings of the eighth workshop on Paral lel and distributed simulation, pages 73--80. ACM Press, 1994.

No context found.

Yi-Bing Lin. Parallel independent replicated simulation on networks of workstations. In Proceedings of the 8th Workshop on Parallel and Distributed Simulation (PADS'94), pages 71 -- 81. IEEE Computer Society Press, July 6-8 1994.

No context found.

Y.-B. Lin. "Parallel Independent Replicated Simulation on a Network of Workstations". In: D. K. Arvind, R. Bagrodia, and J. Y.-B. Lin, Eds., Proceedings of the 8th Workshop on Parallel and Distributed Simulation (PADS '94), pp. 73--80, July 1994.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC