Grupo de Computación Natural 1Centro de Cálculo

Abstract:- P System computational power lies in its non-deterministic, distributed and massively parallel nature. So, it would be desirable for every implementation of a P System to achieve, as far as possible, these features. In this paper, we suggest a distributed architecture of processors called master-slave, in which communications are directed by a single processor, called ‘master’, and a series of processors called ‘slaves ’ whose task is to apply evolution rules to the multisets they receive from the master. To prevent collision and network congestion, communications between master and slaves occur in an organized way. Furthermore, some membranes are allocated in each processor and, finally, proxies are used to communicate with membranes allocated in different processors. All this yields better parallelism in the system as a whole than in previously published studies. In addition to this, we present an analytic study that establishes a series of equations that allows us to accurately determine the optimum number of processors needed, the required time to execute an evolution step, the number of membranes to be located in each slave processor and the conditions that will determine when it is best to use this distributed solution or the ones that have previously been proposed, and even the sequential one