基于改进SF-CORDIC的指数和对数函数求值算法

摘要

提出一种改进的SF-CORDIC（Scaling-free-Coordinate Rotation DIgital Computer）算法用于实现指数函数和对数函数的硬件计算。在双曲坐标系下，算法通过适当选取麦克劳林展开式的近似阶数，可完全省去扩展因子的计算，并利用重复基本迭代和数据预处理以扩展收敛域和计算范围。同时给出算法在双曲坐标系下旋转模式和向量模式的迭代结构。仿真实验表明，在相同精度要求下该算法相比常规CORDIC算法可减少12％面积开销。%An enhanced scaling-free CORDIC algorithm is proposed for implementing the hardware computation of exponential and logarith-mic functions.By proper selection of the approximated order of Maclaurin series the algorithm can completely eliminates the scale-factor com-putation,and uses repeated basic iteration and data pre-processing to extend the domain of convergence and computation scope.Meanwhile, both the rotation mode and vectoring mode of the proposed algorithm in hyperbolic coordinate system is presented.Simulation experiments demonstrate that 12% area consumption is able to be reduced comparing with the conventional CORDIC for same computation accuracy re-quirement.