Research Of SoC Hardware/software Partitioning Methodology Based On Genetic Algorithm And Tabu Search

Integrated circuit enters into the era of system on chip(SoC) as the development of microelectronic techniques, however, as the design complexity increases, traditional design methodology is unable to satisfy the needs of system on chip, hardware/software co-design methodology merges as needed. Hardware/software partitioning is one of the critical techniques in software/hardware co-design. Studying on the hardware/software partitioning methodology, building a reasonable system description model, proposing and improving a partitioning algorithm possess dramatically important value in theory and in application.This dissertation generalizes the domestic and abroad developing situation in the research field of hardware/software co-design, emphasizes on the mathematical model constructed in the process of solving hardware/software partitioning problem and also analyzes the problems commonly existed in current partitioning technology. After the comparison of the properties of genetic algorithm (GA) and tabu search (TS), oriented hardware/software partitioning problem in embedded system and SoC, a hybrid algorithm is innovatively puts forward based on genetic algorithm and tabu search, namely GATS, where the main frame of the algorithm is provided by genetic algorithm and tabu search is taken as the mutation operator,solution space search is accomplished by TS in the process of mutation. Directed acyclic graph produced by TGFF are used in hardware/software bi-partitioning as the mathematical model.At the end of this dissertation, GA, TS and GATS are respectively programmed and verified using the real data generated by TGFF tool. The verification results show that GATS is able to overcome the poor mountain climbing ability of GA and the weak single starting-point capacity of TS, which give a better and more accurate partitioning result.