high performance computing on graphics processing units: hgpu.org

This article presents a method of implementation genetic algorithm in CUDA. Used algorithm operat on a large population and a complex genotype, so that it exceeded the size of the cache memory. It is not completely transferred to the graphics card. It consists of modules that run on the CPU and are synchronized through it. Calculations were based on weak, but widely available graphics cards to test the ability of acceleration algorithms at low cost.