K.M. Chandy and R. Sherman. "Space-Time and Simulation." In Distributed Simulation Conference, Miami, 1989.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....until all domain simulations collectively reach a convergence to the consistent state of all domains and all models. This method can also be seen as a variant and modification of a general scheme for optimistic simulation referred to as Time Space Mappings proposed by Chandy and Sherman in [2]. The simulation in their general scheme is viewed as proceeding in a multidimensional space in which one dimension represent the simulation time and the others spatial dimension of the application. This multidimensional space is then partitioned and a subset of these partitions is executed ....

Chandy, K. M., and R. Sherman, "Space-time and simulation," Proc. Distributed Simulation, 1989, Society for Computer Simulation, pp. 53--57.

....and demonstrate that it is able to use different network simulators as components of the same simulation supporting interoperability and scalability. The described method can be seen as a variant and modification of a general scheme for optimistic simulation referred to as Time Space Mappings [1] where it occupies sparsely pop ulated place of methods based on iterative approximation of a solution in spatially separate parts at the same simulation time. Our approach partitions the network into parts, called domains. The simulation time is partitioned into disjoint intervals. Each domain ....

Chandy, K. M., and R. Sherman, 1989. "Space- time and simulation." Proceedings of the Workshop on Distributed Simulation. SCS Press, 53- 57.

....its own domain. As a result, all domain simulations collectively reach a convergence to the consistent state of all domains and all models. Our method can also be seen as a variant of a general scheme for optimistic simulation referred to as Time Space Mappings proposed by Chandy and Sherman in [2]. Although all optimistic simulations can be viewed as variants of this scheme, very few apply, as we do, iterations over the same time interval to find a solution. In fact, we are aware of only one other network simulator [3] that uses such an iterative scheme. It, however, simulates a single ....

Chandy, K. M., and R. Sherman, "Space-time and simulation, " Proceedings of Distributed Simulation, 1989, Society for Computer Simulation, pp. 53--57.

....called Genesis, can integrate different network simulators as components of the same large scale simulation supporting interoperability and scalability. The described method can be seen as a variant and modification of a general scheme for optimistic simulation referred to as Time Space Mappings [3]. Although all optimistic simulations can be viewed as variants of this scheme, very few apply, as we do, iterations over the same time interval to find a solution. Our approach partitions the network into parts, called domains. The simulation time is partitioned into disjoint intervals. Each ....

K.M. Chandy and R. Sherman, "Space-time and simulation," in Workshop on Distributed Simulation'89. SCS, 1989, pp. 53-73.

....termination detection algorithms to approximate GVT did already appear in a distributed garbage collection algorithm by Hughes [15] This algorithm, however, which is based on Rana s termination detection scheme [24] requires a global clock. The idea has also been sketched by Chandy and Sherman [7] although their resulting algorithm is not used for GVT approximation. Connections between termination detection and GVT approximation were made by Jefferson [16] and Tel [30, 32] and, as mentioned above, by Schoone and Tel [29] The idea of using acknowledgements to catch the timestamps of ....

Chandy, K. M., and Sherman, R. Space--time and simulation. In Proc. of the SCS Multiconference on Distributed Simulation (1989), pp. 53--57.

....repeats until all domain simulations collectively reach a convergence to the consistent state of all domains and all models. Our method can also be seen as a variant and modification of a general scheme for optimistic simulation referred to as Time Space Mappings proposed by Chandy and Sherman in [2]. The simulation in their general scheme is viewed as progressing in a multi dimensional space in which one dimension represent the simulation time and the others map to spatial dimensions of the application. This multi dimensional space is then partitioned and a subset of these partitions is ....

Chandy, K. M., and R. Sherman, "Space-time and simulation," Proc. Distributed Simulation, 1989, Society for Computer Simulation, pp. 53--57. 14

.... (with routing, transport and application layers on top of medium access) and in more general scenarios (such as sparsely connected network, changing topology for mobility, diverse traffic) This study will also garner interest in exploiting combination mechanisms (both time and space parallel [4]) for more effective exploitation of parallelism. 2. Related work A number of prior work used space parallel, conservative or optimistic PDES techniques and tools to speed up various wireless network simulations. See for example [14, 2] Here, we review in detail only the time parallel ....

....PDES techniques and tools to speed up various wireless network simulations. See for example [14, 2] Here, we review in detail only the time parallel mechanisms, as the rest of our work will concentrate on time parallel mechanisms alone. One of the early work was done by Chandy and Sherman [4]. They observe that simulation space time can be partitioned arbitrarily, with updates being made when inputs to a partition change. Simulation continues until the whole system converges. A very general technique is proposed without any specific examples or indications of performance. ....

K. M. Chandy and R. Sherman. Space-time and simulation. Proceedings of the SCS Multiconference on Distributed Simulation, 21(2):53--57, March 1989.

....kernels have been developed using conservative and optimistic approaches [5, 34, 63] The performance of both approaches is dependent on a large number of interrelated factors. Both approaches can be implemented using a number of different protocols [50, 63] and by using a variety of algorithms [17, 20 22, 50]. Furthermore, various parts of the kernel can use different algorithms for accomplishing the same task. For example different algorithms can be used to implement how memory is allocated and de allocated [88] and there are many ways to manage the event queues [16,81] Each of these options and ....

Chandy, K. M., and Sherman, R. Space-time and simulation. In Distributed Simulation (1989), Society for Computer Simulation, pp. 53--57.

....of violating the local causality rule are called safe events. Conservative approaches rely on some strategy to determine when it is safe to process an event, and block until they have a safe event to process [28] 8 Optimistic approaches, on the other hand, do not avoid causality errors [11, 4]. Instead they assume that event messages generally arrive in time stamp order and so proceed forward, processing the events that are available in their event queue without blocking; whenever they discover a causality violation, they use various techniques to ensure that the simulation properly ....

.... generally arrive in time stamp order and so proceed forward, processing the events that are available in their event queue without blocking; whenever they discover a causality violation, they use various techniques to ensure that the simulation properly reflects the correct event execution order [11, 4]. The main drawback with conservative approaches is that they heavily rely on lookahead for their performance [22] Lookahead (in terms of conservative simulation) refers to the ability to predict what will happen, or more importantly, not happen in the simulated future; if an LP with simulated ....

Chandy, K. M., and Sherman, R. Space-time and simulation. In Distributed Simulation (1989), Society for Computer Simulation, pp. 53--57.

....scheduling of large grains of computation even in very low granularity models. This is achieved through a multi level scheduling algorithm. The performance of CCT is also enhanced by ensuring good program cache behavior and automatic load balancing, and through a limited form of time parallelism [4]. In this section we first outline a network modeling paradigm along with a cell level asynchronous transfer mode (ATM) modeling environment. Next we present motivation for the CCT algorithm that is largely deduced from our experience with developing and testing network models. In section 2 we ....

....is extremely good program cache behavior since each processor moves down the pipe with the same set of event buffers. Notice that viewed from the level of a task, we have multiple processors executing the task, each within non overlappingtime windows. Such time parallelism was first described in [4]. 4. The CCT Algorithm This section explains the CCT algorithm in detail. As with all discrete event simulation algorithms, if event m k with timestamp t k , is generated as a result of executing event m i with timestamp t i , then it is a requirement that t k t i (1) The algorithm places two ....

K. Chandy and R. Sherman, "Space, Time and Simulation", SCS Trans on Distributed Simulation Vol. 21 (2), (PADS89), pp93-99, March 1989.

....algorithm. The conservative algorithm has three different variants including the null message protocol [23] the conditional event protocol [7] and an asynchronous combination of the null message and the conditional event protocols [17] The optimistic algorithm uses the space time protocol [8] with lazy cancelation. Besides these protocols, Maisie also provides an Ideal Simulation Protocol(ISP) 16] runtime which provides a reference point to estimate the overhead of different synchronization protocols. ISP computes the execution time for a given model on a given architecture by ....

K.M. Chandy and R. Sherman. Space-Time and Simulation. In Distributed Simulation Conference, Miami, 1989.

....that it will not subsequently receive a message with an earlier timestamp. This constraint may introduce deadlocks, which are typically handled by incorporating deadlock detection[Mis86] or deadlock avoidance[Mis86, Bry77, Cha89a] mechanism into the simulation algorithm. Optimistic algorithms[Jef85, Cha89b] allow an LP to process messages out of order; causality errors are corrected by using rollbacks and re computations. Implementations of optimistic algorithms are usually more difficult because they require complex mechanisms for detection and handling of causality errors, termination detection, ....

K.M. Chandy and R. Sherman. "Space-Time and Simulation." In Distributed Simulation Conference, Miami, 1989.

....with each other. The feature separating the different approaches is the method used to determine which events are safe to process. Some of the most interesting synchronous algorithms include: the three phase algorithm [5, 4] the bounded lag algorithm [41, 40] the Space Time algorithm [14], the Breathing Time Buckets algorithm [56, 57] and the SPaDES algorithm [36] PDES asynchronous methods fall into two broad categories depending on how they enforce adherence to the local causality constraint: conservative and optimistic. A more detailed taxonomy of simulation mechanisms is ....

K.M. Chandy and R. Sherman. Space-Time and Simulation. In Proceedings of the SCS Multiconference on Distributed Simulation, pages 53--57, March 1989.

....discrete event simulators has increased. This has led to the development of techniques for parallel simulation and to the emergence of distributed synchronization algorithms for parallel simulation [4, 13] Two approaches to distributed synchronization exist: conservative [13,18] and optimistic [2, 9]. In conservative approaches, the Support for this work was provided in part by the Defense Advanced Research Projects Agency, contracts F33615 93 C 1315 and F33615 93 C 1316 monitored by Wright Laboratory and contract J FBI 93 116 monitored by the Department of Justice. Preprint ....

.... rollback simulate (lines 3 4) When condition (iii) holds the axiom straggler ax is introduced as an antecedent (line 5) Removing some unnecessary formulae (line 6) and repeatedly rewriting using the axiom simulate simulatex the sequent simplifies to goal.3 : 1] recvtime(e 1) current 1 [ 2] recvtime(e 1) finish 1 [1] nexteventtime(e 1, q 1) MAXINT [2] nexteventtime(e 1, q 1) finish 1 f3g simulatex(initval 1, current 1, finish 1, add(e 1, q 1) add(e 1, q 1) merge(cancel(simulatex(initval 1, current 1, finish 1, q 1, q 1) simulatex(initval 1, recvtime(e 1) ....

[Article contains additional citation context not shown here]

Chandy, K. M., and Sherman, R. Space-time and simulation. In Distributed Simulation (1989), Society for Computer Simulation, pp. 53--57.

....timestamp to be committed. This problem (for the case where ffl i is zero) has been discussed in [15] End of Proof. The above situation is possible in both, the aggressive and lazy cancellation[10] versions of Time Warp. Also, it applies to other optimistic protocols e.g. space time protocol[2]. The only conditions that an optimistic protocol needs to satisfy in order for the above example to apply are: ffl LP s A and B may be mapped to two different processors (parallelism) ffl A processor never blocks while there are unprocessed events in the queue of any of the LP s mapped to ....

K.M. Chandy and R. Sherman. Space-time and simulation. In Distributed Simulation Conference, Miami, 1989.

....few null messages need to be sent. In general, for a network with a cyclic topology and poor lookahead, null messages can quickly overwhelm the simulator leading to poor parallel performance. Conservative implementations of the Maisie simulation language are described in [7] Optimistic algorithms[17, 12] allow an object to process messages out of order; causality errors are corrected by using rollbacks and recomputation. Implementations of optimistic algorithms are usually more difficult because they require complex mechanisms for detection and handling of causality errors, termination detection, ....

K.M. Chandy and R. Sherman. Space-time and simulation. In Distributed Simulation Conference, Miami, 1989.

....the Maisie simulation language, describes a set of optimizations and illustrates the use of the language in the design of efficient parallel simulations. 1 Introduction Distributed (or parallel) simulation refers to the execution of a simulation program on parallel computers. A number of algorithms[25, 10, 11, 21, 20] have been suggested for distributed simulation and many experimental studies have been conducted to evaluate the speedups that may be obtained from these algorithms and their variants. Experience with parallel simulators suggests that reduction in the completion time of a simulation depends ....

.... The simulation algorithms currently supported by Maisie include a sequential algorithm, parallel conservative algorithms based on null messages[25] and conditional events[10] a new conservative protocol that combines null messages with conditional events[23] and a parallel optimistic algorithm[11, 4]. The initial Maisie model is typically executed using a sequential algorithm. If the completion time of the sequential program is not acceptable, parallel implementations may be explored by the analyst. The first step is to make simple modifications to the program to explicitly distribute the ....

[Article contains additional citation context not shown here]

K.M. Chandy and R. Sherman. Space-Time and Simulation. In Proceedings of the 1989 SCS Multiconference on Distributed Simulation Conference, pages 53--57, Miami, March 1989.

....optimistic parallel algorithm. The conservative algorithm has three different variants including the null message protocol, the conditional event protocol, and an asynchronous combination of the null message and the conditional event protocols. The optimistic algorithm uses the space time protocol [6] with lazy cancellation. The communication libraries under Maisie runtime systems provide a common interface with different communication primitives. Therefore, a Maisie program can be ported to different architectures easily (of course, significant effort may be required to exploit the maximal ....

K.M. Chandy and R. Sherman. Space-Time and Simulation. In Distributed Simulation Conference, Miami, 1989.

No context found.

K.M. Chandy and R. Sherman. "Space-Time and Simulation." In Distributed Simulation Conference, Miami, 1989.

No context found.

K. M. Chandy and R Sherman. Space-time and simulation. In Proc. SCS Multiconference on Distributed Simulation, 1989.

No context found.

K.M. Chandy and R. Sherman, "Space-Time and Simulation, " Proc. Distributed Simulation Conf., Society for Computer Simulation, San Diego, Calif., 1989, pp. 33-57.

No context found.

Chandy, K.M. and Sherman, R., "Space, time and simulation", Proceedings of the 1989 SCS Multiconference on Distributed Simulation, Vol. 21, No.2, March 1989, 53-57.

No context found.

Chandy, K. M. and R. Sherman. 1989b. Space-time and simulation. In Distributed Simulation Conference, Miami.

No context found.

Chandy K. M. and R. Sherman. 1989. "Space-Time and Simulation." In Proceedings of the SCS Multiconference on Distributed Simulation (Tampa, FL, March 28-31). SCS, San Diego, CA: 53-57.

No context found.

K.M. Chandy and R. Sherman, "Space, Time, and Simulation," Proceedings of the SCS Multiconference on Distributed Simulation, March 1989, pp. 53-57.

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