Research On Hardware/Software Partitioning Algorithm Based On SOPC

The hardware/software partitioning is a key technology of HW/SW co-design. The task of the hardware/software partitioning is to decide implementation method of the functional modules. The traits of using software are flexible and low cost, and the traits of using hardware are fast and high cost. How to keep the balance among the system's speed, cost,size, power, and obtain the maximum of performance cost ratio,is the purpose of HW/SW partitioning. The result of the hardware/software partitioning has a significant impact on the system performance.During the whole process of HW/SW co-design, the HW/SW partitioning algorithm is a key technology. A good algorithm can get a better solution in a short time. This paper mainly does two contributions. Firstly, it proposes an improved genetic algorithm and an improved simulated annealing algorithm under depth study. Secondly, it proposes an improved genetic-annealing algorithm which combined the simulated annealing algorithm and genetic algorithm, and is applied into hardware/software partitioning which based on IP Core.The advantages of using genetic algorithms in hardware/software partitioning are that it need little specific information, highly fault-tolerant capability and good efficiency of convergence. But it is easy to premature. Simulated annealing algorithm (SA) derived from statistical physics, which simulate the physical process of gradually cooling to achieve the crystallization of the final state. It is proved, as long as the initial temperature is high enough and the cooling process is slow enough, the algorithm will absolutely constringe to the global optimal solution. If the precondition is satisfied, the running time will be long enough to beyond the patience.In this paper, the improved genetic-annealing algorithm which combines the simulated annealing algorithm and genetic algorithm effectively combines the advantage of the two algorithms by introducing the circular strategy, the adaptive probability changes to the mutation, adaptive optimal preservation, and so on.Under the test of example, at the same condition, the improved genetic-annealing algorithm costs less time than the simulated annealing algorithm and genetic algorithm, but it can have greater probability to constringe to the global optimal solution. It is proved that the algorithm has strong stability by running many times, improves the efficiency of convergence, shortens the time of convergence, and has good performance.