Abstract: An improved genetic algorithm is proposed, which can solve the partitioning problem and the hardware implementation exploration simultaneously. In accordance with the characteristic of hardware delay-area contradiction in implementation, the mutation operator is chosen adaptively based on the Q-learning algorithm and greedy algorithm. The objectives are to avoid the blindness in mutation and to enhance the capability of local search of the genetic algorithm. Experiment results show that the improved genetic algorithm is more efficient than the greedy algorithm and the standard genetic algorithm in terms of searching quality and convergence.

Key words: computer engineering, reconfigurable systems, field-programmable gate array(FPGA), hardware/software partitioning, genetic algorithm, Q-learning algorithm