Exploitation of Cell Multi-Processor Array in Solution of Spatio-Temporal Dynamics

Array computers can be useful in the numerical solution of spatio-temporal problems. IBM has recently introduced a topographic array processor called the Cell Broadband Engine (Cell BE). Researchers at the Cellular Sensory Wave Computers Laboratory of SZTAKI in collaboration with the Department of Information Technology, Pazmany Peter Catholic University, Budapest and the Department of Image Processing and Neurocomputing, Pannon University, Veszprem, have implemented a Cellular Neural Network (CNN) simulation kernel on the Cell BE. The CNN simulation is optimized according to the special requirements of the Cell BE and implements both linear and nonlinear (piecewise linear) templates. We have used the CNN simulation kernel to solve Navier-Stokes partial differential equations (PDEs) on the Cell architecture.