Abstract: Hardware/Software partitioning is one of the critical step in Hardware/Software Codesign flow,and has very important influence on the final design. In terms of the number of optimizing objects, it can be classified as single object partitioning and multi objects partitioning. Multi objects partitioning is NP-hard problem,a multi objective partitioning algorithm usually gets irrelevant Pareto results,no traditional optimum results. Genetic algorithm benfits to solve multi objective partitioning for its parallel colony research. Fitness function was investigated and a redefined fitness function was proposed, which adopts self-adaptive parameter and penalization function to escape from the premature convergence and improve evolution speed.The practical experiment results demonstrate that this algorithm is more efficient to balance all the parameters to optimize multi system objects under some constraints, for instance, executing time, cost, hardware area power, etc.

Key words: Hardware/software partitioning,Genetic algorithm,Multi objective partitioning,Fitness function