Distributed simulation of networks

Chandy, K. Mani; Holmes, Victor P.; Misra, Jayadev

doi:10.1016/0376-5075(79)90009-6

Cited by 39 publications

(7 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CPS is designed primarily for experimental comparison of the Bryant-Chandy-Misra (BCM) [3,4], time-warp (TW) [9], and sequential simulation protocols. We believe that CPS may be extended to support simulation protocols proposed in the future.…”

Section: Introductionmentioning

confidence: 99%

a common programming structure for Bryant-Chandy-Misra, time-warp, and sequential simulators

Abrams

1989

Proceedings of the 21st Conference on Winter Simulation - WSC '89

View full text Add to dashboard Cite

CPS expresses event-driven, parallel simulation programs. Rather than being a programming language, CPS is a structure imposed on an existing language, C++. CPS programs contain sufficient information to be implemented as Bryant-Chandy-Misra, time-warp, or sequential simulations. CPS is independent of any problem domain. We describe the main concepts of CPS, illustrating them using the problem of logic circuit simulation. We identify what information a parallel simulation must contain that is not required for sequential simulation.

show abstract

Section: Introductionmentioning

confidence: 99%

a common programming structure for Bryant-Chandy-Misra, time-warp, and sequential simulators

Abrams

1989

Proceedings of the 21st Conference on Winter Simulation - WSC '89

View full text Add to dashboard Cite

show abstract

“…The MULTICM-FIFO& is used when there are multiple nodes sending messages to a node to merge the input streams and to calculate the channel time (as defined by Chandy and Misra [5]). …”

Section: Object Library Componentsmentioning

confidence: 99%

“…A number of techniques have been proposed in the literature for carrying out simulation in parallel, for example [3, 7 , 24,4,5,8,10,11,14,18,231. OLPS, rather than implementing a new simulation technique, provides a framework within which a variety of proposed techniques are and can be implemented.…”

mentioning

confidence: 99%

The Object Library for Parallel Simulation (OLPS)

Abrams¹

1988 Winter Simulation Conference Proceedings

View full text Add to dashboard Cite

3. factor out common operations among the techniques (e.g., interprocess communication, message routing, .~ flow control, random number generation) and provide a single implementation for these, andThe Object Library for Parallel Simulation (OLPS) is a set of C++ objects from which a parallel, discrete event simulation or simulation language can be built and run on multiprocessor workstations. OLPS provides a common interface to the simulation programmer for sev-4. facilitate experinientation through: (a) modifvine existinz algorithms, and \ , * v u era1 parallel simulation algorithms, including implementations of the Chandy-Misra and Time-Warp algorithms. (b) adding new algorithms to the library. . -OLPS also provides instrumentation t o compare the performance of each parallel algorithm on the same simulation problem. The system currently runs on the V distributed system on the DEC Firefly.OLPS is not, however, intended to be a simulation language. Rather, we expect that OLPS will provide the run-time support needed to port existing simulation languages to multiprocessors. Keywords:parallel simulation, object-OLPS currently implements two algorithms: Chandyoriented programming, Chandy-Misra simula-Misra [5] with deadlock avoidance and Time-Warp [ll]. The Object Library is implemented in C++ using the tion algorithm, Time-Warp simulation algorithm Argonne National Library macro package [16] to permit its porting to various operating systems. Our implementation runs on the V operating system [6] on a five pro-1 MOTIVATION cessor DEC Firefly [22].This paper describes the initial implementation of and performance measurements from a system for parallel simulation called the Object Library for Parallel Simulation (OLPS). The system is targeted t o multiprocessor workstations, and is suitable for general purpose, discrete event simulation.

show abstract

“…i t i t i s !i c. The coli ser va t,ive scheme, whose founds.tion was laid by R. E. Bryant(l977) and independently by M . Chandy(1979), suitably partitions a physical system to be simulated into several components called logical processes ( l p 's). Each lp simulates itself autonomously while coordinating with other lp's in the system by sending and receiving time-stamped messages.…”

Section: Introductionmentioning

confidence: 99%

An efficient and scalable parallel algorithm for discrete-event simulation

Prasad¹,

Deo²

1991 Winter Simulation Conference Proceedings.

View full text Add to dashboard Cite

We describe a new parallel algorithm for discreteevent simulation on exclusive-read exclusive-write parallel random-access machine (EREW PRAM). This algorithm is based on a recently developed parallel data structure, namely parallel heap, which when used as a parallel event-queue, allows deletion of O ( p ) earliest messages and insertion of O ( p ) new messages each in O(1og n ) time, using p processors, where n is the number of messages scheduled. Our algorithm can simulate up to p independent messages in O(1og n ) timethus achieving O ( p ) speedup. The number of processors, p , can be optimally varied in the range 1 5 p 5 n . To our knowledge, such a theoretical efficiency in a parallel simulat.ion a l g o r i h i has been achieved for the first time.

show abstract

Distributed simulation of networks

Cited by 39 publications

References 2 publications

a common programming structure for Bryant-Chandy-Misra, time-warp, and sequential simulators

a common programming structure for Bryant-Chandy-Misra, time-warp, and sequential simulators

The Object Library for Parallel Simulation (OLPS)

An efficient and scalable parallel algorithm for discrete-event simulation

Contact Info

Product

Resources

About