Stochastic computing implementation of trigonometric and hyperbolic functions

Lian Huai, Peng Li, Gerald E. Sobelman, David J. Lilja

Abstract: High performance implementations of trigonometric and hyperbolic functions are important in many application areas. This paper presents an FPGA implementation of these functions using the stochastic computing methodology. The results are compared to the well-known CORDIC approach. All of the designs are synthesized and implemented on a Xilinx Virtex-5 FPGA. The results are compared in terms of delay and area for various input data widths. The results show that the proposed design method has advantages in the delay-area product and in soft error tolerance compared to CORDIC designs.