采用启发式分支定界的软硬件划分

提出一种以任务图为描述方法的软硬件划分方法．首先分别计算芯片所需面积／时间／通信软硬件倾向度，并结合各节点的比重因子获得启发参数；然后采用启发式的分支定界法对系统进行划分，以获得可行解和最优解．通过对文中算法和RECOD和UNRET的划分算法进行编码，并在同一平台上分别计算节点数为10，15，20，25，30的系统的启动间距、最小启动间距及其所需时间，比较各算法之间的性能．文中算法适用于划分粒度较粗和中小规模的系统．

Hardware-Software Partitioning Using Heuristic Branch and Bound

The paper presents a method of Hardware-Software partitioning using task graph. By the method, first we calculate the inclination of area, time, communication and the ratio of all nodes, used to calculate the heuristic parameters. Then according to the heuristic parameters, we adopt heuristic branch and bound algorithm for partitioning the system to obtain both the feasible solution and the best solution. We coded the algorithm described in the paper as well as other two algorithms, RECOD and UNRET. Then we compared the performance of these three algorithms by calculating the minimum initiation internal, the initiation internal and the calculation time for the systems respectively with 10, 15,20,25,30 nodes. The algorithm proposed in the article is suitable for the systems in middle or small scale with a coarse granularity partitioning of task graph.