基于library-free映射的电路面积快速优化算法

针对library-free映射过程中常用的动态规划算法在求解大电路时覆盖时间过长的问题，提出了一种将动态规划与遗传算法相结合的混合优化算法，用于平衡求解速度和求解质量，并利用“与/或/非”图和逻辑努力实现基于MOS晶体管的电路的面积估算.MCNC电路的测试结果显示，相较于动态规划，混合算法求得的最优解在面积平均增加不到1%的情况下，求解时间可节省35%以上.

A fast area optimization method using library-free mapping

Accounting for the time consuming dynamic programming for the large circuit covering during library-free mapping, a hybrid optimization method consisting of the dynamic programming and the genetic algorithm is proposed as the covering strategy to trade off the quality of the solutions and the cost of running time.Furthermore, an area estimation method based on AND-OR-Inverter Graph and logic effort for CMOS circuits is also proposed for circuit area optimization.The proposed method is tested under MCNC benchmarks.Experimental results show that compared with the reported algorithm, the proposed method can reduce more than 35% running time with less than 1% increase of circuit area on average.