Results 1 
7 of
7
Statistical Properties of the Simulated Time Horizon
 in Conservative Parallel DiscreteEvent Simulations”, accepted for the Proceedings of ACM Symposium On Applied Computing
, 2002
"... We investigate the universal characteristics of the simulated time horizon of the basic conservative parallel algorithm when implemented on regular lattices. This technique [1, 2] is generically applicable to various physical, biological, or chemical systems where the underlying dynamics is asynchro ..."
Abstract

Cited by 4 (1 self)
 Add to MetaCart
(Show Context)
We investigate the universal characteristics of the simulated time horizon of the basic conservative parallel algorithm when implemented on regular lattices. This technique [1, 2] is generically applicable to various physical, biological, or chemical systems where the underlying dynamics is asynchronous. Employing direct simulations, and using standard tools and the concept of dynamic scaling from nonequilibrium surface/interface physics, we identify the universality class of the time horizon and determine its implications for the asymptotic scalability of the basic conservative scheme. Our main finding is that while the simulation converges to an asymptotic nonzero rate of progress, the statistical width of the time horizon diverges with the number of PEs in a power law fashion. This is in contrast with the findings of Ref. [3]. This information can be very useful, e.g., we utilize it to understand optimizing the size of a moving “time window ” to enforce memory constraints. Keywords conservative parallel discreteevent simulation, scalability, nonequilibrium surface growth, Monte Carlo, stochastic processes 1.
2001. Selforganized criticality in simulated correlated systems. Computer Physics Communications 142 (13): 7681. K.A. Iskra is a PhD candidate at the Computing, System Architecture, and Programming
 the Computing, System Architecture, and Programming Laboratory, Universiteit van
, 2001
"... In this paper we study the influence of spatiotemporal correlations on the dynamic runtime behavior of the optimistic parallel Time Warp simulation method. By means of Ising spin simulation, we show experimentally that the probability distribution of the number of rolled back events behaves as a po ..."
Abstract

Cited by 2 (1 self)
 Add to MetaCart
(Show Context)
In this paper we study the influence of spatiotemporal correlations on the dynamic runtime behavior of the optimistic parallel Time Warp simulation method. By means of Ising spin simulation, we show experimentally that the probability distribution of the number of rolled back events behaves as a powerlaw distribution over a large range of subcritical Ising temperatures and decays exponentially for supercritical Ising temperatures. The experimental results indicate that for critical Ising temperatures, where longrange correlations occur, the computational complexity of Time Warp and physical complexity of the Ising spin model are entangled and contribute both to the runtime behavior in a nonlinear way. © 2001 Elsevier Science B.V. All rights reserved.
SEE PROFILE
"... Statistical properties of the simulated time horizon in conservative parallel discreteevent simulations ..."
Abstract
 Add to MetaCart
(Show Context)
Statistical properties of the simulated time horizon in conservative parallel discreteevent simulations
Going through Rough Times: from NonEquilibrium Surface Growth to Algorithmic Scalability
, 2002
"... E#cient and faithful parallel simulation of large asynchronous systems is a challenging computational problem. It requires using the concept of local simulated times and a synchronization scheme. We study the scalability of massively parallel algorithms for discreteevent simulations which employ co ..."
Abstract
 Add to MetaCart
E#cient and faithful parallel simulation of large asynchronous systems is a challenging computational problem. It requires using the concept of local simulated times and a synchronization scheme. We study the scalability of massively parallel algorithms for discreteevent simulations which employ conservative synchronization to enforce causality. We do this by looking at the simulated time horizon as a complex evolving system, and we identify its universal characteristics. We find that the time horizon for the conservative parallel discreteevent simulation scheme exhibits KardarParisiZhanglike kinetic roughening. This implies that the algorithm is asymptotically scalable in the sense that the average progress rate of the simulation approaches a nonzero constant. It also implies, however, that there are diverging memory requirements associated with such schemes.
Rollbacks in Time Warp  Analysis and Modelling
, 2002
"... This paper presents a study of the interactions between the random number generator used and the runtime behaviour of the parallel Time Warp simulation kernel APSIS. A different rollback length distribution, with a far larger chance of long rollbacks taking place, is observed when the state of the ..."
Abstract
 Add to MetaCart
(Show Context)
This paper presents a study of the interactions between the random number generator used and the runtime behaviour of the parallel Time Warp simulation kernel APSIS. A different rollback length distribution, with a far larger chance of long rollbacks taking place, is observed when the state of the random number generator is not preserved across the rollbacks. An explanation for this phenomenon is provided. An analytical model of the rollback behaviour in Time Warp is developed, for rollback length expressed in either the simulation time or the number of events to be rolled back. The hope is that, once the model is complete, it will be possible to determine under what circumstances it is profitable (not) to preserve the state of the random number generator.
Going through Rough Times: from NonEquilibrium Surface Growth to Algorithmic Scalability
"... Efficient and faithful parallel simulation of large asynchronous systems is a challenging computational problem. It requires using the concept of local simulated times and a synchronization scheme. We study the scalability of massively parallel algorithms for discreteevent simulations which employ ..."
Abstract
 Add to MetaCart
Efficient and faithful parallel simulation of large asynchronous systems is a challenging computational problem. It requires using the concept of local simulated times and a synchronization scheme. We study the scalability of massively parallel algorithms for discreteevent simulations which employ conservative synchronization to enforce causality. We do this by looking at the simulated time horizon as a complex evolving system, and we identify its universal characteristics. We find that the time horizon for the conservative parallel discreteevent simulation scheme exhibits KardarParisiZhanglike kinetic roughening. This implies that the algorithm is asymptotically scalable in the sense that the average progress rate of the simulation approaches a nonzero constant. It also implies, however, that there are diverging memory requirements associated with such schemes.