Cellular Automata Model Tuned for Efficient Computation on GPU with Global Memory Cache

Models based on Cellular Automata paradigm are inherently parallel and give the opportunity to achieve very high efficiency when they are implemented on GPUs. In previous works, the Author presented two versions of the CA model of water flow: one that uses only the global memory, and a second that uses local memory as a manually managed cache. In this paper, these two algorithms are tested on processors with Fermi architecture. The results of performance evaluations are presented and discussed.