GPGPU Programming and Cellular Automata: Implementation of the SCIARA Lava Flow Simulation Code

This paper presents an efficient implementation of a well-known computational model for simulating lava flows on Graphical Processing Units (GPU) using the Compute Unified Device Architecture (CUDA) interface developed by NVIDIA. GPUs are specifically designated for efficiently processing graphic datasets. However, recently, they are also being exploited for achieving exceptional computational results even for applications not directly connected with the Computer Graphics field. We here show an implementation of the SCIARA Cellular Automata model for simulating lava flows on graphic processors using CUDA. Carried out experiments show that significant performance improvements are achieved, over a factor of 100, depending on the problem size, adopted device and type of performed memory optimization, confirming how graphics hardware can represent a valid solution for the implementation for Cellular Automata models.