Effectiveness of Dynamic Load Balancing in Parallel Execution of a Subsurface Flow Cellular Automata Model

In this paper, a subsurface flow Cellular Automata (CA) model, namely the XCA-Flow model, is considered with the aim of optimizing its parallel execution by means of a purposely tailored dynamic load balancing technique. Indeed, a suitable distribution of computational load over different processing elements is particular relevant in the case of parallel execution of CA, where the domain space is partitioned in regions assigned to the parallel computing nodes. In addition, the XCA-Flow model can exhibit very unbalanced distribution of the water flow, and this unbalanced condition also might change during the simulation advancement. As a consequence, a Dynamic Load Balancing technique can be suitably utilized in order to achieve an optimal resource utilization thus reducing the overall execution time. First tests implemented using the MPI technology have demonstrated an appreciable reduction of execution times in comparison with the not-balanced parallel version.