Taylor Series Based Architecture for Quadruple Precision Floating Point Division

Abstract

This paper presents an area efficient architecture for quadruple precision division arithmetic on the FPGA platform. Many application demands for the higher precision computation (like quadruple precision) than the single and double precision. Division is an important arithmetic, but requires a huge amount of hardware resources with increasing precision, for a complete hardware implementation. So, this paper presents an iterative architecture for quadruple precision division arithmetic with small area requirement and promising speed. The implementation follows the standard processing steps for the floating point division arithmetic, including processing of sub-normal operands and exceptional case handling. The most dominating part of the architecture, the mantissa division, is based on the series expansion methodology of division, and designed in an iterative fashion to minimize the hardware requirement. This unit requires a 114×114 bit integer multiplier, and thus, a FPGA based area-efficient integer multiplier is also proposed with better design metrics than prior art on it. These proposed architectures are implemented on the Xilinx FPGA platform. The proposed quadruple precision division architecture shows a small hardware usage with promising speed.