Book chapter
Математическая модель алгоритмов синхронизации времени для распределенного имитационного моделирования
In book
We study synchronization aspects in parallel discrete event simulation (PDES) algorithms. Our analysis is based on the recently introduced model of virtual times evolution in an optimistic synchronization algorithm. This model connects synchronization aspects with the properties of the profile of the local virtual times. The main parameter of the model is a “growth rate” q = 1/(1 + b), where b is a mean rollback length. We measure the average utilization of events and the desynchronization between logical processes as functions of the parameter q. We found that there is a phase transition between an “active phase”, i.e. when the utilization of the average processing time is finite, and an “absorbing state” with zero utilization, vanishing at a critical point qc ≈ 0.136. The average desynchronization degree (i.e. the vari- ance of local virtual times) grows with the parameter q. We also investi- gate the influence of the sparse distant communications between logical processes and found that they do not change drastically the synchronization properties in the optimistic synchronization algorithm, which is the sharp contrast with the conservative algorithm [1]. Finally, we compare our results with the existing case-study simulations.
We address question of synchronisation in parallel discrete event simulation (PDES) algorithms. We study synchronisation in conservative PDES model adding long-range connections between processing elements. We investigate how fraction of the random long-range connections in the synchronisation scheme influences the simulation time profile of PDES. We found that small fraction of random distant connections enhance synchronisation, namely, the width of the local virtual times remains constant with increasing number of processing elements. At the same time the conservative algorithm of PDES on small-world networks remains free from deadlocks. We compare our results with the case-study simulations.
Nowadays simulation modeling is applied for solving a wide range of problems. There are simulations which require significant performance and time resources. To decrease overall simulation time a model can be converted to a distributed system and executed on a computer network. The goal of this project is to create a library enabling clear and rapid development parallel discrete event models in AnyLogic. The library is aimed for professionals in computer simulation and helps to reduce code amount. The project includes a research on different synchronization algorithms. In this paper we present techniques which can be used in creating distributed models. We present comparison of a single threaded model with a distributed model implementing optimistic algorithm. The comparison shows a significant improvement in wallclock time achieved by separating the model into independent submodels with minimal communications.
We extend concept of local simulation times in parallel discrete event simulation (PDES) in order to take into account architecture of the current hardware and software in high-performance computing. We shortly review previous research on the mapping of PDES on physical problems, and emphasise how physical results may help to predict parallel algorithms behaviour.
We investigate synchronisation aspects of an optimistic algorithm for parallel discrete event simulations (PDES). We present a model for the time evolution in optimistic PDES. This model evaluates the local virtual time profile of the processing elements. We argue that the evolution of the time profile is reminiscent of the surface profile in the directed percolation problem and in unrestricted surface growth. We present results of the simulation of the model and emphasise predictive features of our approach.