基于近似逻辑的不完全指定FPRM函数的功耗优化

基于精确逻辑的逻辑综合和优化方法已有广泛的研究. 但有时并非需要精确逻辑, 从而可利用这种特性进行关键指标的综合和优化. 本文提出了基于近似逻辑的不完全指定固定极性Reed-Muller (Incompletely Specified Fixed Polarity RM, ISFPRM)逻辑函数的逻辑优化方法. 首先建立基于二级逻辑的功耗估算模型和近似电路的设计效能评估模型, 然后提出基于近似逻辑的ISFPRM函数的优化算法, 并用遗传算法加以实现. 所提算法应用于MCNC标准电路进行测试, 结果表明在一定的面积约束下, 电路每增加1%的误差率获得12%~18%的功耗优化

Power optimization of incompletely specified FPRM function based on approximate logic

The logic synthesis and optimization methods based on accurate logic have been extensively studied. However, sometimes accurate logic may not be technically needed. As a result, this undesired requirement can be taken advantage of to synthesize and optimize other key issues. In this paper, an optimization method based on approximate logic for the Incompletely Specified Fixed Polarity Reed-Muller (ISFPRM) function is proposed. Firstly, the evaluation model of power consumption based on the two-level logic and the estimation model of design efficiency aiming at the approximate circuit are established. Then an optimized algorithm based on the approximate logic ISFPRM function is proposed and implemented in genetic algorithm. The proposed algorithm is tested under MCNC benchmarks. The experimental results show that under area constraints, every 1% increase in the error rate of the circuit can achieve power saving by 12% to 18%