R.M. Fujimoto. Parallel discrete event simulation: Will the field survive? ORSA Journal on Computing, 5(3): 213--230, 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the behavior of Time Warp, e.g. induce rollback if the simulation of internal events has progressed further into the future than the time stamp carried by the message etc. The question on the Time Warp protocol performance has often been raised in the literature [Fuji 90b] Gupt 91] Turn 92] Fuji 93] Akyi 93] Das 93] Das 94b] General performance rules cannot be formulated [Fuji 90b] since performance due to a very high degree of interweaving of influencing factors cannot be sufficiently characterized by models. Even full implementations of the Time Warp protocol often prohibit ....

R. M. Fujimoto. "Parallel Discrete Event Simulation: Will the Field Survive?". ORSA Journal of Computing, Vol. 5, No. 3, pp. 218--230, 1993.

....often been criticized as an academic playground, and the community as living in an ivory tower. This pessimism has also diffused into the community and has led to an existential discussion on the chances of survival as reflected in a collection of papers following Fujimoto s position statement in [5]. Despite successful attempts to escape from this image in the spirit of As the mountain is not coming to the Prophet, evidently the Prophet must go to the mountain. 6] expressing the hope that simulation practitioners and simulationists in industry would at least meet half way [6] on their ....

R.M. Fujimoto, Parallel discrete event simulation: will the field survive? ORSA J. Comput. 5 (3) (1993) 218--230.

....that do not have a counterpart in the other scheme (e.g. lazy cancellation) hinder even more a fair comparison. Collected arguments on the performance characteristics and influences are reported in [Fers 95] In an existential discussion within the parallel and distributed simulation community [Fuji 93] the availability of performance prediction methods for DDES protocols has been pointed out as being critical for the future success and general acceptance of DDES methods in practice [Lin 93] e.g. it is important for a simulationist to be able to evaluate the suitability of certain DDES ....

R. M. Fujimoto. "Parallel Discrete Event Simulation: Will the Field Survive?". ORSA Journal of Computing, Vol. 5, No. 3, pp. 218--230, 1993.

....method. 88 discrete event simulation framework but is still capable of simulating practically any Spike Response neuron model. One of our future works is the parallelization of the simulator. Many largescale discrete event simulations are now performed in a parallel computing environment [13, 27]. The discrete event simulation of a pulsed neural network seems a suitable application for parallelization, since every pulse transmission can be treated as an event, and delays between neurons enable us to use a simpler synchronization mechanism. Moreover, the queuing model used in our ....

Richard M. Fujimoto. Parallel discrete event simulation: Will the field survive? ORSA Journal on Computing, 5(3):213--230, 1993. (cited in page 89) 131

....that is based on a discrete event simulation framework but is still capable of simulating practically any Spike Response neuron model. One of our future works is the parallelization of the simulator. Many large scale discrete event simulations are now performed in a parallel computing environment [13, 14]. The discrete event simulation of a pulsed neural network seems a suitable application for parallelization, since every pulse transmission can be treated as an event, and delays between neurons enable us to use a simpler synchronization mechanism. Moreover, the queuing model used in our ....

Richard M. Fujimoto. Parallel discrete event simulation: Will the field survive? ORSA Journal on Computing, 5(3):213--230, 1993.

....the computation granularity of processes. Empirical results show that the SPADES is scalable and can deliver good speedup for large PDES applications. Understanding the performance of parallel simulation is a complex issue and is essential for this technology to be widely accepted in industry [4]. A framework to analyze parallel simulation performance is discussed in [16] Acknowledgements This research has been supported by a grant from the Ministry of Education and the Port of Singapore Authority under grant RP060715. The authors like to thank Siew Theng Kong for implementing the ....

R. M. Fujimoto, "Parallel Discrete Event Simulation: Will the Field Survive?," ORSA Journal of Computing, Vol. 5, No. 3, pp. 213-230, Summer 1993.

....rollback and resimulation costs may severely degrade performance. A number of techniques have been proposed to reduce overheads in conservative and optimistic algorithms. Studies indicate that PDES is a viable technology for studying system behavior. A summary of published results is given in [6]. However, performance is usually dependent on the characteristics of the model being simulated and the architecture of the parallel or distributed system. A consistently e#cient algorithm that performs well for a wide range of applications still eludes researchers. The principal reason is that ....

....attractive, periods of incorrect simulation due to overaggressive event processing can incur large overheads of state saving and rollbacks. Empirical evidence suggests that neither conservative nor optimistic algorithms provide e#cient and consistent performance for a wide variety of models [6]. This is partly due to the fact that it is di#cult to choose a specific algorithm prior to simulation for large, complex models, because the dependencies are irregular and dynamic over the course of the simulation. For example, the simulation of mean cell delay in a large ATM network is ....

R. Fujimoto, "Parallel discrete event simulation: will the field survive?," ORSA Journal on Computing, vol. 5, pp. 213--230, Summer 1993.

....simulation s future. This procedure is repeated until some predefined simulation time is reached or the event list becomes empty. Although DES (of complex systems) presents itself as a viable alternative to analytical evaluation, simulation experiments can take a considerably long time to execute [20]. In some cases, these long simulation times may even render the results, once obtained, useless. For example there is little use in simulating a 3 day weather forecast if the the 7 simulation experiment itself is going to take more than three days. The simulation research community has come up ....

Fujimoto, R. M. Parallel discrete event simulation: Will the field survive ? ORSA Journal on Computing 5, 3 (June 1993).

....they reduce R, the residual cost at the balance point in Figure 10. Yet another measure is their performance relative to that of a protocol that does not control optimism, such as Time Warp. Time Warp is the protocol that has been applied to the most applications with encouraging results [Fuji93a]. However, it suffers the drawbacks of optimistic protocols noted earlier. Consequently, there are several applications for which Time Warp does not perform well. By showing that NPSI protocols consistently outperform Time Warp over a wide range of workloads that includes those for which Time Warp ....

.... focused on designing protocols for specific applications [NiRe84, FuNi92] Among other reasons, we believe the lack of a consistently efficient protocol is a significant contributor to the fact that fifteen years of research in PDES have made almost no impact on the general simulation industry [Fuji93a, NiHe95]. Consequently, we concur with [Reyn93] that a consistently efficient protocol is still one of the most important goals for PDES researchers. The NPSI adaptive approach is a significant step in achieving this goal. The two main strengths of our approach are (i) it is adaptive, and (ii) it uses ....

Fujimoto, R.M., "Parallel discrete-event simulation: Will the field survive?", Feature article in ORSA Journal on Computing, Vol. 5, No. 3, Summer 1993, 213-230.

....performance of the throttling scheme with the conventional TW scheme. Section 6 contains our concluding remarks. 2 Parallel and Distributed Simulation Research on PDES during the last two decades is documented in various landmark survey and tutorial articles at different development stages [11, 12, 13, 23, 26]. Briefly, PDES adopts the process oriented modeling methodology to partition an application system into physical processes that interact at simulated time. The simulator consists of a set of logical processes (LP) one per physical process. LPs are mapped onto a machine for execution based on its ....

R.M. Fujimoto, "Parallel Discrete Event Simulation: Will the Field Survive ?", ORSA Journal of Computing, Vol. 5, No. 3, pp. 218-230, 1993.

....network simulations, computer architectures, digital circuits, and parallel programs. This is due in part by the greater availability of systems which are relatively lower in cost, yet still achieve good performance. However, parallel simulation is still primarily used in the research community [Fuj93]. A major limiting factor for more widespread use is the complexity of implementing efficient parallel simulations. Nonetheless, the potential for significant speedup in the execution time of a simulation has led to the design of several algorithms for parallel simulation. They are broadly ....

R. Fujimoto. "Parallel Discrete event simulation: Will the field survive ?" ORSA Journal on Computing, 5(3):213--230, 1993.

.... simulation users often encounter was the subject of an insightful debate among several noted researchers on the issues and future of the field, with the conclusion that a practical focus is necessary to move the field forward, particularly with regard to developing efficient models for simulation [13]. This paper addresses that challenge to make practical improvements in parallel simulation model development using HDLs. 3. Algorithmic Issues 3.1 Conservative Algorithms In a conservative PDES, simulation activity always corresponds to activity which must occur in the actual system being ....

Fujimoto, R.M., et al. "Parallel Discrete-Event Simulation: Will the Field Survive?" ORSA Journal of Computing, Vol. 5, No. 3, Summer 1993.

....Given this, performing simulation using parallel computers is generally acknowledged as the most promising solution to solve the problem. Unfortunately, despite over two decades of research, the technology of parallel simulation has not significantly impressed the general simulation community [Fuj93b]. Considerable efforts and expertise are still required to develop efficient simulation programs. There are no golden rules that a programmer can follow to guarantee an efficient program. Generally speaking, parallel simulation is very hard [Fuj90] Nevertheless, the field of parallel simulation ....

Richard M. Fujimoto. Parallel discrete event simulation: Will the field survive? ORSA Journal on Computing, 5(3):213--230, Summer 1993.

....models, as compared to the uni processor version. Simulation of large scale models such as a virtual factory [1] would thus become feasible. The widespread interest in PDES in the research community, however, did not bring about the widespread deployment of PDES in real world applications [2]. Most research has been focused on inventing new algorithms and getting performance measurements on specific applications [3] There are two main reasons for this phenomenon. First, parallel processing machines, though in much abundance, are still mainly used by the research community. Most users ....

....objects, it is therefore necessary to have a mechanism to exchange information. The communication between simulation objects can be either explicit or implicit, and implemented by either message passing or sharing of memory. Approaches to the implementation of shared state are discussed in [2]. The advantage of using a PDES library over a PDES language is that the user is given more flexibility in 4 SIMULATION MARCH 1999 controlling the simulation application in terms of the behavior of the underlying synchronization protocol. PDES libraries often come packaged with various options ....

Fujimoto, R.M. "Parallel Discrete-Event Simulation: Will the Field Survive?" ORSA Journal on Computing, Vol. 5, No. 3, pp 213-230, Summer 1993.

....apply to other simulation applications such as battlefield simulation, VLSI simulation, queueing network simulation and computer architecture simulation. However, these optimization techniques may need to be tailored for specific applications. Many studies have devoted to this issue (see [1] 2] [52], 53] 54] and references therein) The PCS example can be seen as being a member of a larger class of simulation model where one first discretizes the spatial domain into a grid, and then simulates moving entities from one grid cell to another. In this sense, the PCS problem is isomorphic to ....

Fujimoto, R.M., "Parallel Discrete Event Simulation: Will the Field Survive?", ORSA Journal on Computing, vol. 5, no. 3, 1993.

....views me and those like me is more the matter at hand. 2. A navel well contemplated The PADS community has done yeoman s work in the area of critical self assessment. A good example of this introspection is a series of articles within the Summer 1993 issue of the ORSA Journal on Computing [1,2,9,10,14, 24, 28]. 2.1. Will the field survive In the lead article of the ORSA JOC series, Richard Fujimoto asked the question, Parallel discrete event simulation: will the field survive The genesis of the question was an observed (or perceived) lack of impact of parallel discrete event simulation (PDES) ....

R. M. Fujimoto. Parallel discrete event simulation: Will the field survive? ORSA Journal on Computing, 5(3):213--230, Summer 1993.

....Given this, performing simulation in parallel is acknowledged as the most promising solution to simulating those intractable systems. Unfortunately, despite over two decades of research, the technology of parallel simulation has not significantly impressed the general simulation community [Fuj93b] Considerable efforts and expertise are still required to develop efficient simulation programs. There are no golden rules that a programmer can follow to guarantee an efficient program. Generally speaking, parallel simulation is very hard [Fuj90a] Nevertheless, the field of parallel ....

....e.g. LK91,Fer78, Jai91] The simulation community, however, is no longer satisfied with the performance of sequential simulators. The systems we desire to simulate today are so complex that the tasks of executing these simulation models are often beyond the capability of sequential simulators [Fuj93b] 2.2 Parallel Simulation Seminal work of DES in parallel dates back roughly 20 years. Hundreds of papers have been published since then, the majority appearing within this decade (see References and Bibliography) Various ways of decomposing a simulation for processing on multiple processors ....

[Article contains additional citation context not shown here]

Richard M. Fujimoto. Parallel discrete event simulation: Will the field survive? ORSA Journal on Computing, 5(3):213--230, Summer 1993.

....is made, and the paradigms prescribe a singular perspective: a contrived form of process interaction. We characterized this typical logical process view as contrived because its definition is motivated more by execution concerns than by an accurate representation of the physical counterpart. Fujimoto (1990, p. 33) notes that this (logical process) view allows application programmers to partition the simulation state variables into a set of disjoint states, and ensure that no simulator event accesses more than one state. This partitioning permits minimal processor synchronization, and thus has ....

Fujimoto, R.M. (1993a). "Parallel Discrete Event Simulation: Will the Field Survive?" ORSA Journal on Computing, 5(3), pp.

....unknown classes without some considerable jiggery pokery in most languages. Reflection makes it rather simple. Because of these characteristics, Java seems to be a language that could be useful in programming parallel simulations. Besides that, Java is a object oriented language, and according to Fujimoto (1993), parallel simulation strategies and object oriented programming have obvious similarities in that a logical process is similar to an object, and sending a message is similar to invoking a method. 5.0 THE JUST CLIENT:SERVER SYSTEM The research discussed here has led to a number of three phase ....

FUJIMOTO, R. M. Parallel Discrete Event Simulation: Will the Field Survive? ORSA Journal on Computing, Vol. 5, No. 3 (1993), pp. 213-230.

....multiple event queues and each processor is unable to trivially determine whether the event at the head of the queue is safe to process, as another event with an earlier time stamp may arrive at any time. The synchronisation protocols developed in Parallel Discrete Event Simulation (PDES) Fujimoto 93] broadly fall into two categories: conservative protocols, in which processors negotiate to determine if events are safe to process; and optimistic protocols, in which events are processed until an out of order event is detected, which causes the state of the simulation to be rewound and event ....

....expressionoriented semantics and garbage collection. The implementation of APOSTLE described in this thesis is based on a compiler and a run time system which reuses an existing optimistic simulator. Generally speaking, conservative protocols are only suitable for a small class of simulations [Fujimoto 93] therefore we choose to focus on optimistic synchronisation protocols in this thesis. State saving is handled using persistent datastructures [Bruce 95] 10 1.2 Thesis statement A simulation oriented language can enable optimisations to be performed that are usually difficult or cumbersome, ....

[Article contains additional citation context not shown here]

R. M. Fujimoto, "Parallel Discrete Event Simulation: Will the Field Survive?", ORSA Journal on Computing, 5(3):213-230.

....continuations avoids the need for stacks in the first place. We similarly obtain cheap and natural thread pre emption. 1 Background Over the last fifteen years or so, parallel discrete event simulation (PDES) has been shown to provide good speed up over a wide range of applications Fujimoto [7] cites a number of examples but it has failed to make a significant impact in the general simulation community [7, x11] Much of this failure can be attributed to the difficulty of developing non trivial parallel simulations. We identify two contributing factors: existing PDES systems tend ....

R. M. Fujimoto. Parallel Discrete Event Simulation: Will the Field Survive? ORSA Journal on Computing, 5(3):213-- 230, Summer 1993.

....for basic research. 4 PARALLEL AND DISTRIBUTED SIMULATION (RICHARD FUJIMOTO) Parallel and distributed simulation technology has reached a crossroads. In past years, researchers have long lamented the limited impact of this technology in the general modeling and simulation community, e.g. see (Fujimoto 1993). This situation has changed considerably in recent years. The latter half of the 1990 s has seen parallel and distributed simulation technology flourish. Perhaps most important is its inclusion in the High Level Architecture (HLA) that has been adopted both by the U.S. Department of Defense and ....

Fujimoto, R. M. 1993. Parallel Discrete Event Simulation: Will the Field Survive? ORSA Journal on Computing, 5(3): 213-230.

No context found.

R.M. Fujimoto. Parallel discrete event simulation: Will the field survive? ORSA Journal on Computing, 5(3): 213--230, 1993.

No context found.

R.M. Fujimoto "Parallel Discrete Event Simulation: Will the Field Survive?", ORSA Journal on Computing (feature article), Vol. 5, No. 3, pp. 213-230, summer 1993.

No context found.

R. M. Fujimoto. Parallel discrete event simulation: Will the field survive? ORSA Journal of Computing, 5(3):213--230, 1993.

First 50 documents

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