Using the PlayStation3 for speeding up metaheuristic optimization

Traditional computer software is written for serial computation. To solve an optimization problem, an algorithm or metaheuristic is constructed and implemented as a serial stream of instructions. These instructions are executed on a central processing unit (CPU) on one computer.Parallel computing uses multiple processing elements simultaneously to solve a problem. This is accomplished by breaking the problem into independent parts so that each processing element can execute its part of the algorithmsimultaneously with the others. The processing elements can be diverse and include resources such as a single computer with multiple processors, several networked computers, specialized hardware, or any combination of the above.Today most commodity CPU designs include single instructions for some vector processing on multiple (vectorized) data sets, typically known as SIMD (Single Instruction, Multiple Data). Modern video game consoles and consumer computer-graphics hardware rely heavily on vector processing in their architecture. In 2000, IBM, Toshiba and Sony collaborated to create the Cell Broadband Engine (Cell BE), consisting of one traditional microprocessor (called the Power Processing Element or PPE) and eight SIMD co-processing units, or theso-called Synergistic Processor Elements (SPEs), which found use in the Sony PlayStation3 among other applicationsThe computational power of the Cell BE orPlayStation3 can also be used for scientific computing. Examples and applications have been reported in e.g. Kurzak et al. (2008), Bader et al. (2008), Olivier et al. (2007), Petrini et al. (2007).In this work, the potential of using the PlayStation3 for speeding up metaheuristic optimization is investigated.More specifically, we propose an adaptation of an evolutionary algorithm with embedded simulation for inspection optimization, developed in Van Volsem et al. (2007), Van Volsem (2009a) and Van Volsem (2009b)Traditional computer software is written for serial computation. To solve an optimization problem, an algorithm or metaheuristic is constructed and implemented as a serial stream of instructions. These instructions are executed on a central processing unit (CPU) on one computer.Parallel computing uses multiple processing elements simultaneously to solve a problem. This is accomplished by breaking the problem into independent parts so that each processing element can execute its part of the algorithmsimultaneously with the others. The processing elements can be diverse and include resources such as a single computer with multiple processors, several networked computers, specialized hardware, or any combination of the above.Today most commodity CPU designs include single instructions for some vector processing on multiple (vectorized) data sets, typically known as SIMD (Single Instruction, Multiple Data). Modern video game consoles and consumer computer-graphics hardware rely heavily on vector processing in their architecture. In 2000, IBM, Toshiba and Sony collaborated to create the Cell Broadband Engine (Cell BE), consisting of one traditional microprocessor (called the Power Processing Element or PPE) and eight SIMD co-processing units, or theso-called Synergistic Processor Elements (SPEs), which found use in the Sony PlayStation3 among other applicationsThe computational power of the Cell BE orPlayStation3 can also be used for scientific computing. Examples and applications have been reported in e.g. Kurzak et al. (2008), Bader et al. (2008), Olivier et al. (2007), Petrini et al. (2007).In this work, the potential of using the PlayStation3 for speeding up metaheuristic optimization is investigated.More specifically, we propose an adaptation of an evolutionary algorithm with embedded simulation for inspection optimization, developed in Van Volsem et al. (2007), Van Volsem (2009a) and Van Volsem (2009b)C