A comparative study of several 2D cellular automata implementations in FPGA

This work reports FPGA implementations of several two-dimensional cellular automata (CA) structures using both Altera and Xilinx target chips, thus providing a high performance computing platform with many applications (random number generation, modeling of various physical phenomena and image processing). Our study compares various, typical CA structures in terms of hardware efficiency concluding upon the best choice for a given application. While different VHDL descriptions may be provided for the same architecture, our study reveals that the choice for all independent inputs cells instead of outer-totalistic ones gives the most efficient implementation. Also interesting conclusions regarding the specific problems of each particular FPGA developer tools and chips were derived.