Evaluation of elementary functions without range reduction

摘要: The evaluation of elementary functions can be performed by approximations using minimax polynomials requiring simple hardware resources. general method to calculate an function is composed three steps: range reduction, computation the polynomial in reduced argument and reconstruction. This approach allows a low-degree approximation but reduction reconstruction introduce a computation overhead. This work proposes methodology without steps. Applications that need compute may benefit from avoiding these steps if argument belongs sub-domain function. Particularly context embedded systems, applications related to digital signal processing most times require within specific interval. As a consequence not doing degree approximant increases maintain the required precision. Interval segmentation effective way overcome this issue because approximations are computed smaller intervals. proposed uses non-uniform as way to mitigate problem arising carrying out reduction. benefits that come applying interval technique are limited reconstruction steps only applies step. However, when used proposed methodology it reveals more effective. Some were implemented FPGA device. metric used characterize area occupation corresponding latency. results of each implementation compared with ones method using show latency significantly while approximately the same.