Parallelization of Cellular Automata in Graphic Cards with CUDA

• Marcos Paulo Riccioni de Melos UFRRJ

Abstract

Cellular automata is a problem-solving technique containing elements that interact with their near neighbor and require external data to instantiate. The efficiency of the algorithm was tested in several scenarios and compared between the sequential and parallel approaches. The results show that the methodology used improves performance for problems of this nature and can be improved by using the shared memory of the GPU. There is also a limitation on the size of the problem in parallel implementation, so the distribution on more boards and more computers would be a better solution.

Keywords: Autômatos Celulares, CUDA, GPUs