System Notification:
We are currently experiencing intermittent issues impacting performance. We apologize for the inconvenience.
By Topic

Parallel and Distributed Simulation, 2000. PADS 2000. Proceedings. Fourteenth Workshop on

Date 28-31 May 2000

Filter Results

Displaying Results 1 - 23 of 23
  • Proceedings Fourteenth Workshop on Parallel and Distributed Simulation

    Save to Project icon | Request Permissions | PDF file iconPDF (549 KB)  
    Freely Available from IEEE
  • Author index

    Page(s): 181
    Save to Project icon | Request Permissions | PDF file iconPDF (72 KB)  
    Freely Available from IEEE
  • Advanced distributed simulation: what we learned, where to go next

    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (8 KB)  

    Summary form only given. It has been a little less than ten years since modeling and simulation (M&S) hit the knee on the curve. During these past few years a great deal of marketing of the potential of M&S has occurred, which resulted in a significant influx of funding for research and development (R&D) projects, especially in the area of distributed simulation. One of the significant experiments in this area-the DARPA Synthetic Theater of War program-officially ended, bringing to an end one of the more robust experiments in distributed simulation. We are now in a time where most of the M&S funding is targeted at production programs, with much fewer dollars going into R&D or experimentation. Although it is good that major programs are capitalizing on previous R&D efforts, it would not be true to say that the necessary R&D has been completed to realize the vision of distributed simulation. It would be true to say that we now have a much better understanding of the issues. What is needed now is a period of reflection on the vision, where we are, what was done right, what hasn't worked well, and where we should be headed for the next five to ten years View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Full text access may be available. Click article title to sign in or learn about subscription options.
  • An approach for federating parallel simulators

    Page(s): 63 - 70
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (180 KB)  

    This paper investigates issues concerning federations of sequential and/or parallel simulators. An approach is proposed for creating federated simulations by defining a global conceptual model of the entire simulation, and then mapping individual entities of the conceptual model to implementations within individual federates. Proxy entities are defined as a means for linking entities that are mapped to different federates. Using this approach, an implementation of a federation of optimistic simulators is examined. Issues concerning the adaptation of optimistic simulators to a federated system are discussed. The performance of the federated system utilizing runtime infrastructure (RTI) software executing on a shared memory multiprocessor (SMP) is compared with a native (non-federated) SMP-based optimistic parallel simulator. It is demonstrated that a well designed federated simulation system can yield performance comparable to a native, parallel simulation engine, but important implementation issues must be properly addressed View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Pre-sampling as an approach for exploiting temporal uncertainty

    Page(s): 157 - 164
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (148 KB)  

    We describe an approach to exploit temporal uncertainty in parallel and distributed simulation by utilizing time intervals rather than precise time stamps. Unlike previously published work that proposes new message ordering semantics, our approach is based on conservative, time stamp order execution and enhancing the lookahead of the simulation by pre-drawing random numbers from a distribution that models temporal uncertainty. The advantages of this approach are that it allows time intervals to be exploited using a conventional Time Stamp Order (TSO) delivery mechanism, and it offers the modeler greater statistical control over the assigned time stamps. An implementation of this approach is described and initial performance measurements are presented View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Applying parallel discrete event simulation to network emulation

    Page(s): 15 - 22
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (216 KB)  

    The simulation of wide area computer networks is one area where the benefits of parallel simulation have been clearly demonstrated. We present a description of a system that uses a parallel discrete event simulator to act as a high speed network emulator. With this, real Internet Protocol (IP) traffic generated by application programs running on user workstations can interact with modelled traffic in the emulator thus providing a controlled test environment for distributed applications. The network emulator uses the TasKit conservative parallel discrete event simulation (PDES) kernel. TasKit has been shown to be able to achieve improved parallel performance over existing conservative and optimistic PDES kernels, as well as improved sequential performance over an existing central-event-list based kernel. This paper explains the modifications that have been made to TasKit to enable real-time operation along with the emulator interface that allows the IP network simulation running in the TasKit kernel to interact with real IP clients. Initial emulator performance data is included View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • ROSS: a high-performance, low memory, modular time warp system

    Page(s): 53 - 60
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (220 KB)  

    We introduce a new time warp system called ROSS: Rensselaer's Optimistic Simulation System. ROSS is an extremely modular kernel that is capable of achieving event rates as high as 1,250,000 events per second when simulating a wireless telephone network model (PCS) on a quad processor PC server. In a head-to-head comparison, we observe that ROSS out performs the Georgia Tech Time Warp (GTW) system on the same computing platform by up to 180%. ROSS only requires a small constant amount of memory buffers greater than the amount needed by the sequential simulation for a constant number of processors. The driving force behind these high-performance and low memory utilization results is the coupling of an efficient pointer-based implementation framework, Fujimoto's (1989) fast GVT algorithm for shared memory multiprocessors, reverse computation and the introduction of kernel processes (KPs). KPs lower fossil collection overheads by aggregating processed event lists. This aspect allows fossil collection to be done with greater frequency, thus lowering the overall memory necessary to sustain stable, efficient parallel execution View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Slow memory: the rising cost of optimism

    Page(s): 45 - 52
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (188 KB)  

    Rapid progress in the design of fast CPU chips has outstripped progress in memory and cache performance. Optimistic algorithms would seem to be more vulnerable to poor memory performance because they require extra memory for state saving and anti-messages. We examine the performance of both optimistic and conservative protocols in controlled experiments to evaluate the effects of memory speed and cache size, using a variety of applications View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Repeatability in real-time distributed simulation executions

    Page(s): 23 - 32
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (268 KB)  

    Real-time distributed simulations, such as on-line gaming or military training simulations are normally considered to be non-deterministic. Analysis of these simulations is therefore difficult depending solely on logging and runtime observations. This paper explores an approach for removing one major source of non-determinism in these simulations, thereby allowing repeatable executions. Specifically, we use a synchronization protocol to ensure repeatable delivery of messages. Through limited instrumentation of the simulation code, we maintain a virtual time clock, by which message delivery is governed. The additional overhead imposed by the scheme is shown to be reasonable, although additional reductions in this overhead are anticipated. The results are demonstrated in the context of a simple combat model, whose only source of non-determinism is communications latency. The simulation is shown to be made repeatable, and the perturbation on the execution compared to the non-repeatable execution small. The paper is one step in bridging the gap between the traditional PDES perspective and real-time simulation world View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Optimizing cell-size in grid-based DDM

    Page(s): 93 - 100
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (132 KB)  

    In a large scale distributed simulation with thousands of dynamic objects, efficient communication of data among these objects is an important issue. The broadcasting mechanism specified by the Distributed Interactive Simulation (DIS) standards is not suitable for large scale distributed simulations. In the high level architecture (HLA) paradigm, the Runtime Infrastructure (RTI) provides a set of services, such as data distribution management (DDM) among federates. The goal of the DDM module in RTI is to make the data communication more efficient by sending the data only to those federates that need the data, as opposed to the broadcasting mechanism employed by DIS. Several DDM schemes have appeared in the literature. We discuss grid based DDM and develop a DDM model that uses grids for matching the publishing/subscription regions, and for data filtering. We show that appropriate choice of the grid-cell size is crucial in obtaining good performance. We develop an analytical model and derive a formula for identifying the optimal cell size in grid-based DDM View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Safe timestamps and large-scale modeling

    Page(s): 71 - 78
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (188 KB)  

    The paper examines issues that recur in consideration of simulation time-stamps, in the context of building very large simulation models from components developed by different groups at different times. A key problem here is “safety”, loosely defined to mean that unintended model behavior does not occur due to unpredictable behavior of timestamp generation and comparisons. We revisit the problems of timestamp format and simultaneity, and then turn to the new problem of timestamp interoperability. We describe how a C++ simulation kernel can support the concurrent evaluation of submodels that internally use heterogeneous timestamps, and evaluate the execution time costs of doing so. We find that use of a safe timestamp format that explicitly allows different time scales costs less than 10% over a stock 64-bit integer format, whereas support for completely heterogeneous timestamps can cost as much as 50% in execution speed View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Analytic performance model for speculative, synchronous, discrete-event simulation

    Page(s): 35 - 44
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (340 KB)  

    Performance models exist that reliably describe the execution time and efficiency of parallel discrete-event simulations executed in a synchronous iterative fashion. These performance models incorporate the effects of processor heterogeneity, other processor load due to shared computational resources, application workload imbalance, and the use of speculative computation. This includes modeling the effects of predictive optimism, a technique for improving the accuracy of speculative assumptions. We extend these models to incorporate correlated workloads across the set of processors and validate the models with two different applications View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Grain sensitive event scheduling in time warp parallel discrete event simulation

    Page(s): 173 - 180
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (224 KB)  

    Several scheduling algorithms have been proposed to determine the next event to be executed on a processor in a time warp parallel discrete event simulation. However none of them is specifically designed for simulations where the execution time (or granularity) for different types of events has large variance. We present a grain sensitive scheduling algorithm which addresses this problem. In our solution, the scheduling decision depends on both timestamp and granularity values with the aim at giving higher priority to small grain events even if their timestamp is not the lowest one (i.e. the closest one to the commitment horizon of the simulation). This implicitly limits the optimism of the execution of large grain events that, if rolled back, would produce a large waste of CPU time. The algorithm is adaptive in that it relies on the dynamic recalculation of the length of a simulated time window within which the timestamp of any good candidate event for the scheduling falls in. If the window length is set to zero, then the algorithm behaves like the standard Lowest-Timestamp-First (LTF) scheduling algorithm. Simulation results of a classical benchmark in several different configurations are reported for a performance comparison with LTF: these results demonstrate the effectiveness of our algorithm View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Model structure and load balancing in optimistic parallel discrete event simulation

    Page(s): 147 - 154
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (228 KB)  

    The concept of strong groups is introduced to describe the structure of simulation models. It is shown that logical processes within strong groups process at approximately the same rate and that different strong groups can progress at different rates. An algorithm based on the rates of the strong groups is presented to balance the load among the physical processors and for flow control View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Efficient distributed simulation of a communication switch with bursty sources and losses

    Page(s): 85 - 92
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (180 KB)  

    Algorithms for simulating an ATM switch on a distributed memory multiprocessor are described. These include parallel generation of bursty arrival streams, along with the marking and deleting of lost cells due to buffer overflows. These algorithms increase the amount of computation carried out independently by each processor and reduce the communication between the processors. When the number of cells lost is relatively small, the run time of the simulation is approximately O(N/P), where N is the total number of cells simulated and P is the number of processors. The cells are processed in intervals of fixed length; that length affects the structure and the performance of the algorithms View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Network aware time management and event distribution

    Page(s): 119 - 126
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (220 KB)  

    We discuss new synchronization algorithms for parallel and distributed discrete event simulations (PDES) which exploit the capabilities and behavior of the underlying communications network. Previous work in this area has assumed the network to be a black box which provides a one-to-one, reliable and in-order message passing paradigm. In our work, we utilize the broadcast capability of the ubiquitous Ethernet for synchronization computations, and both unreliable and reliable protocols for message passing, to achieve more efficient communications between the participating systems. We describe two new algorithms for computation of a distributed snapshot of global reduction operations on monotonically increasing values. The algorithms require O(N) messages (where N is the number of systems participating in the snapshot) in the normal case. We specifically target the use of this algorithm for distributed discrete event simulations to determine a global lower bound on time-stamp (LETS), but expect the algorithm has applicability outside the simulation community View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Distributed, parallel simulation of multiple, deliberative agents

    Page(s): 101 - 108
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (252 KB)  

    Multi agent systems comprise multiple, deliberative agents embedded in and recreating patterns of interactions. Each agent's execution consumes considerable storage and calculation capacities. For testing multi agent systems, distributed parallel simulation techniques are required that take the dynamic pattern of composition and interaction of multi-agent systems into account. Analyzing the behavior of agents in virtual, dynamic environments necessitates relating the simulation time to the actual execution time of agents. Since the execution time of deliberative components can hardly be foretold, conservative techniques based on lookahead are not applicable. On the other hand, optimistic techniques become very expensive if mobile agents and the creation and deletion of model components are affected by a rollback. The developed simulation layer of JAMES (a Java Based Agent Modeling Environment for Simulation) implements a moderately optimistic strategy which splits simulation and external deliberation into different threads and allows simulation and deliberation to proceed concurrently by utilizing simulation events as synchronization points View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Load balancing for conservative simulation on shared memory multiprocessor systems

    Page(s): 139 - 146
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (192 KB)  

    Load balancing is a crucial factor in achieving good performance for parallel discrete event simulations. We present a load balancing scheme that combines both static partitioning and dynamic load balancing. The static partitioning scheme maps simulation objects to logical processes before simulation starts while the dynamic load balancing scheme attempts to balance the load during runtime. The static scheme involves two steps. First, the simulation objects that contribute to small lookahead are merged together by using a merging algorithm. Then a partitioning algorithm is applied. The merging is needed to ensure a consistent performance for our dynamic scheme. Our dynamic scheme is tailor-made for an asynchronous simulation protocol that does not rely on null messages. The performance study on a supply chain simulation shows that the partitioning algorithm and dynamic load balancing are important in achieving good performance View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Strategies for success in parallel simulation applications

    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (36 KB)  

    Summary form only given. While the PADS community has traditionally focused on, and done a great job with, the technical aspects of developing simulations that run fast and can be connected to other simulations, it has paid little or no attention to the overall strategies required to produce a marketable, useful, and successful parallel simulation. This lack of market focus has led to many fears of the demise of the PADS community, complaints of its lack of general acceptance by the broader simulation community and predictions that it will become merely another venue for simulation interconnection. These fears, complaints, and predictions are unnecessary. There are several examples of successful parallel simulations, in domains as far apart as aviation modeling and wargames. What can we learn from their successes? How can we translate their general acceptance into other parallel simulation domains? Are there market opportunities that we are missing? In short, what is the parallel simulation community lacking, and what does it need to do in order to be more successful? The purpose of the article is to begin a discussion on the answers to these questions (as opposed to definitively answering them). We draw on numerous examples of successful applications, and make some concrete suggestions for furthering the community View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Parallelizing a sequential logic simulator using an optimistic framework based on a global parallel heap event queue: an experience and performance report

    Page(s): 111 - 118
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (172 KB)  

    We have parallelized the Iowa Logic Simulator a gate-level fine-grained discrete-event simulator by employing an optimistic algorithm framework based on a global event queue implemented as a parallel heap. The original code and the basic data structures of the serial simulator remained unchanged. Wrapper data structures for the logical processes (gates) and the events are created to allow roll-backs, all the earliest events at each logical processes are stored into the parallel heap, and multiple earliest events are simulated repeatedly by invoking the simulate function of the serial simulator. The parallel heap allowed extraction of hundreds to thousands of earliest events in each queue access. On a bus-based shared-memory multiprocessor simulation of synthetic circuits with 250,000 gates yielded speedups of 3.3 employing five processors compared to the serial execution time of the Iowa Logic Simulator and limited the number of roll-backs to within 2,000. The basic steps of parallelization are well-defined and general enough to be employable on other discrete-event simulators View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • An empirical study of conservative scheduling

    Page(s): 165 - 172
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (204 KB)  

    It is well known that the critical path provides an absolute lower bound on the execution time of a conservative parallel discrete event simulation. It stands to reason that optimal execution time can only be achieved by immediately executing each event on the critical path. However, dynamically identifying the critical event is difficult, if not impossible. We examine several heuristics that might help to determine the critical event, and conduct a performance study to determine the effectiveness of using these heuristics for preferential scheduling View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Locality-preserving load-balancing mechanisms for synchronous simulations on shared-memory multiprocessors

    Page(s): 131 - 138
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (208 KB)  

    Many synchronous algorithms have been proposed for parallel and discrete simulations. However the actual performance of these algorithms has been far from ideal, especially when event granularity is small. Barring the case of low parallelism in the given simulation models, one of the main reasons of low speedups is in the uneven load distribution among processors. We present several new locality-preserving load balancing mechanisms for synchronous simulations on shared-memory multiprocessors. We show both theoretically and empirically that some of these mechanisms incur very low overhead. The results confirm that one of the new mechanisms is indeed more efficient and scalable than common existing approaches View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.