Adapting an evolutionary algorithm with embedded simulation and pseudo-random number generation for the cell broadband engine

For the problem of optimizing inspection strategies in multi-stage production systems, a metaheuristic consisting of an evolutionary algorithm with embedded simulation was developed in Van Volsem et al. (2007), Van Volsem (2009) and Van Volsem (accepted for publication, 2009). The metaheuristic requires normally distributed pseudo-random numbers; the time needed for this random number generation is a substantial fraction of the total computation time. In an effort to reduce the computation time, the metaheuristic was adapted for computation on the Cell Broadband Engine. The proposed adaptation is twofold: we propose a way to make the metaheuristic suitable for fast multicore computation, and secondly, the potential of SIMD computation for speeding up the random number generation process and the metaheuristic is investigated