Co-Processor For Genetic Algorithms

Speed and accuracy are two important parameters in solving complex problems. However, there are some problems with no polynomial time solutions. To solve these problems, an exhaustive search in the solution space should be used. Genetic Algorithm (GA) can solve NP problems in a more efficient manner by doing an enhanced random search. In GA, we have some basic operations which are executed many times. The intrinsic complexity of the operations prevents us from having fast speed on conventional hardware. This paper proposes an architecture which can be used with a CPU as a co-processor to better solve GA problems. Implementation results are provided for this processor. As well, elaborate comparison between our solution and a conventional software solution is made.