Hardware Modules for Safe Integer and Floating-Point Arithmetic
Amrita Ratan · OhioLink ETD Center (Ohio Library and Information Network) · 2013
Integer and floating-point data types are widely used to represent numerical data in computer arithmetic.Since the range of values representable in a computer are limited, arithmetic operations on these data types can lead to overflows and underflows.More often than not, silent overflows and underflows are permitted by software and even though hardware records these errors, it may not be designed to handle them.These errors can be a threat to security and can reduce the reliability of software.In this work we present a hardware design for an Arithmetic and Logic Unit (ALU), and for a Floating-Point Unit (FPU).The ALU and FPU record and gracefully handle overflows and underflows.They have three modes of operation and handle overflows and underflows differently in each mode.The mode may be selected based on the requirement of the application.The designs have been implemented in Verilog and have been verified on the Altera DE2-115 Board.This method can be modified for use in any hardware design.