D.R. Je#erson. Virtual time. ACM Trans. Program. Lang. Syst., 7(3):404--425, 1985.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....DES algorithm given in gure 2.1 on page 9. Events are scheduled at other LPs by sending messages for each such event. Special synchronisation (using rollback) is needed to preserve causality (see also section 2.2. 1) In TW this synchronisation mechanism is based on the notion of Virtual Time [14]. Each LP keeps track of its local virtual time (LVT) This is the simulation time to which the LP has progressed and is equal to the timestamp of the last processed event. Because of the rollbacks, the LVTs jump back and forth through simulation time, while each LVT does eventually progress ahead ....

....If it already a message sent to another LP 19 had processed the positive event, it also needs to perform a rollback to the most recent state with a save time smaller than the event s timestamp. All rollback propagations eventually terminate, while all erroneous events are unsent recursively [14]. Another important quantity is the Global Virtual Time (GVT) It is the simulation time to which the simulation as a whole has progressed, i.e. the time up to which the total simulation can be considered to be valid. It is also the smallest time to which any LP could need to rollback, because ....

[Article contains additional citation context not shown here]

D. R. Jeerson. Virtual time. In ACM Transactions on Programming Languages and Systems, volume 7, pages 404-425, July 1985.

....appropriate time. However, all the reasoning about distributed messages and timing is done in the discrete event framework. MPADES is a conservative simulator (such as [1] events are not realized until it can be guaranteed that casual event ordering will not be violated. Optimistic simulations [2], on the other hand, support a back up mechanism in case events are executed out of order. Debates over the merits of conservative and optimistic simulation are common and several surveys discuss the issues [3, 4] Agent based or agent oriented simulation seems to be a relatively new idea for the ....

Jeerson, D.: Virtual time. ACM Trans. Prog. Lang. and Syst. 7 (1985)

....1.2.1.2 Optimistic algorithms Optimistic algorithms execute events without the guarantee of freedom from causality errors. When a causality error occurs, the common approach is to recover from the error using a rollback mechanism. Figure 1. 4 shows a typical event execution loop for Time Warp [5], an optimistic algorithm. During forward execution, an LP executes events unconditionally as they arrive. Messages sent during forward execution are called positive messages. Periodically, an LP saves its simulation state. If a message arrives in an LP s logical past (i.e. a timestamp smaller ....

....which is also a positive message is called a primary rollback. One initiated by the arrival of an antimessage is called a secondary rollback. Because the logical clocks of LPs can move forward as well as roll back, a measure of the progress of the simulation is needed. Global virtual time (GVT) [5] is defined for optimistic PDES as the smallest timestamp among (1) unexecuted events, 2) events being executed, and (3) messages in transit between LPs. GVT is the logical time up to which the simulation is guaranteed to be accurate and free from causality errors. A particularly important ....

[Article contains additional citation context not shown here]

D. Je#erson, "Virtual time," ACM Transactions on Programming Languages and Systems, vol. 7, pp. 404--425, July 1985.

....LPs. The messages are time stamped by the sending LP according to its Local Virtual Time (LVT) At the receiving side, they are put in a local Future Event List (FEL) to be processed. Synchronization of messages for causality constraints can either be performed conservatively [1] or optimistically [2]. Simulation of communication networks usually requires a large amount of events to be simulated as these events typically represent packets in the real system. High speed network models emphasize this property as the number of messages increases dramatically. Therefore, parallel simulations are ....

....is to obtain a correct simulation at any time. Unfortunately, this scheme introduces deadlocks that must be avoided by sending null messages. These messages rely on a lookahead ability which allows for an arti cial propagation of the simulation time. On the other side, optimistic approaches [2] do not search for safety but provisions are made to roll back to an earlier coherent state when they occur. Periodic check pointing and anti messages to cancel bad computations are then needed as a counterpart of more freedom. In addition, a Global Virtual Time (GVT) is required to monitor the ....

Jeerson, D. R. : Virtual Time. ACM Trans. on Prog. Lang. and Sys., 7(3) (July 1985) 405-425.

....are likely to occur. A causality error happens when a message arrives at a receiving LP and is outdated, or old, according to its LVT. These problems arise the need of explicit synchronization mechanisms that are traditionally classi ed in two main categories: conservative [14] and optimistic [24]. Other methods to parallelize a simulation exist but their applications are limited [21] Simulation of communication network models usually requires a large amount of events to be simulated as these events typically represent packets in the real system. High speed network models emphasize this ....

Jeerson, D. R. : Virtual Time. ACM Trans. on Prog. Lang. and Sys., 7(3) (July 1985) 405-425.

....16] However they cannot su#ciently model time based on local clocks which may drift in distributed systems. Also, some researchers have explored methods for agreement on processes time by using clock synchronization [10, 13] and for maintaining local time based on causality by using time stamps [9, 12]. However the former methods cannot cope with systems where communication delay is unpredictable. The latter methods lose real time duration between events. The goal of this paper is to develop a formalism for reasoning about distributed object oriented computations, in particular, local time ....

Je#erson, D. R., Virtual Time, ACM TOPLAS, Vol.7, No.3, 1985.

....is the need for appropriate simulation protocols to guarantee the absence of causality violations, and thus guarantee that DDES produces the same results as sequential simulation. Since a description of synchronization protocols such as the conservative [Chandy and Misra, 1979] or the optimistic [Je erson, 1985] approaches would exceed the scope of this contribution, in the sequel we suppose that the reader is familiar with these issues. A detailed introduction can be found for example in [Ferscha, 1996] 2.2 Fault Types Di erent types of failures may require di erent types of reactions and thus ....

Jeerson, D. R. (1985). Virtual time. ACM Transactions on Programming Languages and Computer Systems, 7(3):404-425.

....is to obtain a correct simulation at any time. Unfortunately, this scheme introduces deadlocks that must be avoided by sending null messages. These messages rely on a lookahead ability which allows for an arti cial propagation of the simulation time. On the other side, optimistic approaches [13] do not search for safety but provisions are made to roll back to an earlier coherent state when they occur. Periodic check pointing and antimessages to cancel bad computations are then needed as a counterpart of more freedom. In addition, a Global Virtual Time (GVT) is required to monitor the ....

Jeerson, D. R. : Virtual Time. ACM Trans. on Prog. Lang. and Sys., 7(3) (July 1985) 405-425.

....is to obtain a correct simulation at any time. Unfortunately, this scheme introduces deadlocks that must be avoided by sending null messages. These messages rely on a lookahead ability which allows for an arti cial propagation of the simulation time. On the other side, optimistic approaches [5] do not search for safety but provisions are made to roll back to an earlier coherent state when they occur. Periodic checkpointing and anti messages to cancel bad computations are then needed as a counterpart of more freedom. In addition, a Global Virtual Time (GVT) is required to monitor the ....

Jeerson, D. R. : Virtual Time. ACM Trans. on Prog. Lang. and Sys., 7(3) (July 1985) 405425.

No context found.

D.R. Je#erson. Virtual time. ACM Trans. Program. Lang. Syst., 7(3):404--425, 1985.

No context found.

Jeerson, D. R.,. 1985. Virtual time. ACM Transactions on Programming Languages and Systems 7 (3): 404-425.

No context found.

D.R. Je#erson, "Virtual Time", ACM Transactions on Programming Languages and Systems, Vol.7, No.3, pp.404425, 1985.

No context found.

D. Je#erson, "Virtual Time", ACM Transactions on Programming Languages and Systems, vol.7, n.3, pp.404-425, 1985. 31

No context found.

D.R. Je#erson. Virtual time. ACM Trans. Prog. Lang. and Syst., 7(3):404--425, July 1985.

No context found.

D.R. Je#erson. "Virtual Time". ACM Trans. Prog. Lang. and Syst., 7(3):404--425, July 1985.

No context found.

D.R. Jeerson. \Virtual Time". ACM Trans. Prog. Lang. and Syst., 7(3):404-425, July 1985.

No context found.

D.R. Je#erson. Virtual time. ACM Trans. Prog. Lang. and Syst., 7(3):404--425, July 1985.

No context found.

D.R. Jeerson. \Virtual Time". ACM Trans. Prog. Lang. and Syst., 7(3):404-425, July 1985.

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