Implementation of a floating point VHDL arithmetic library

Abstract

Floating point arithmetic is used in many applications and is found in several areas nowadays such as optimisation of mathematical functions, image processing algorithms, artificial neural networks and digital signal processors, especially real-time signal processing. The use of high level languages, such as VHDL and Verilog has made floating point operations easier to implement on Field Programmable Gate Arrays (FPGAs). Due to the increase in use of floating point notation, a ready-made floating point library would significantly aid developers in creating new applications at a lower cost and greater range. The aim of this project is to build a floating point library which supports several arithmetic operations which are used frequently in applications as mentioned previously. Several algorithms for different arithmetic operations were studied and chosen appropriately in order to be implemented into a library. The implementation also includes variable size floating point arithmetic, which lets the developer to choose freely the range and precision of the floating point operation. Xilinx 10.1 was used to implement the library and run the simulations in order to verify the functionality of the floating point operations.