Scalable asynchronous execution of cellular automata

The performance and scalability of cellular automata, when executed on parallel/distributed machines, are limited by the necessity of synchronizing all the nodes at each time step, i.e., a node can execute only after the execution of the previous step at all the other nodes. However, these synchronization requirements can be relaxed: a node can execute one step after synchronizing only with the adjacent nodes. In this fashion, different nodes can execute different time steps. This can be a notable advantageous in many novel and increasingly popular applications of cellular automata, such as smart city applications, simulation of natural phenomena, etc., in which the execution times can be different and variable, due to the heterogeneity of machines and/or data and/or executed functions. Indeed, a longer execution time at a node does not slow down the execution at all the other nodes but only at the neighboring nodes. This is particularly advantageous when the nodes that act as bottlenecks vary during the application execution. The goal of the paper is to analyze the benefits that can be achieved with the described asynchronous implementation of cellular automata, when compared to the classical all-to-all synchronization pattern. The performance and scalability have been evaluated through a Petri net model, as this model is very useful to represent the synchronization barrier among nodes. We examined the usual case in which the territory is partitioned into a number of regions, and the computation associated with a region is assigned to a computing node. We considered both the cases of mono-dimensional and two-dimensional partitioning. The results show that the advantage obtained through the asynchronous execution, when compared to the all-to-all synchronous approach is notable, and it can be as large as 90% in terms of speedup.