一种考虑电压岛边界约束的多电压布图算法

多电压设计(multiple supply voltage,MSV)是降低SoC功耗的有效方法之一.为便于电压岛供电引脚的放置,提出了一种考虑电压岛边界约束的多电压布图算法.首先,基于切分树表示的布图解特点,提出一种边界检查算法快速确定所有模块的边界信息.其次,以优化功耗为目标采用改进动态规划方法进行多电压分配并构建电压岛.最后,以模拟退火算法作为搜索引擎对芯片的面积、线长和功耗进行协同优化.为减少SA迭代次数,采用了一个两阶段的降温策略.对GSRC电路的实验结果表明,该算法可获得满足边界约束的多电压布图,且和不考虑边界约束时相比,仅在功耗上平均增加5.2％.

Multi-voltage floorplanning algorithm considering voltage islands boundary constraints

MSV is one of the effective methods to reduce SoC power consumption. In order to facilitate the placement of power supply pins of each voltage island, this paper proposed a multi-voltage floorplanning algorithm considering voltage island boundary constraints. Firstly, based on the representation characteristics of slicing tree, this paper proposed a boundary checking algorithm to quickly determine the boundary information of all modules. Then, in order to minimize the power consumption, this paper used the improved dynamic programming method to assign voltage to each module and construct voltage islands. Finally, it took simulated annealing algorithm as search engine to optimize the chip area, wire length and power consumption simultaneously. In addition, to reduce iteration times, the SA algorithm adopted a two-stage cooling strategy. Experimental results on GSRC circuits show that the proposed algorithm can generate multi-voltage floorplan satisfying boundary constraints, and the power consumption increases only by 5.2% on average compared with that without considering boundary constraints.