Floating-Point Hardware Design: A Test Perspective

Abstract

The growing field of Artificial Intelligence research necessitates the development of non-standard bit-width number format arithmetic hardware units to improve the energy efficiency of the underlying hardware. However, building these hardware units using hardware description language is error-prone. It is difficult to catch these errors in the early design stage without having the proper tools or instruments to cross-check the results. Furthermore, floating-point hardware designs contain many stages by which the final result is calculated; therefore, it is essential to identify the erroneous stage for debugging. This paper proposes an easy-to-use Python library for IEEE-754-based floating-point numbers with arbitrary exponent and mantissa width. This library provides not only the result for cross-checking HDL results but also debugging the hardware's intermediate stage results for easier and faster development. The support of this module in converting the numbers to and fro from decimal to binary makes it ideal to use it as a full-fledged calculator to perform the complex arithmetic in the required format and debugger in binary form for the development of hardware to perform these computations on.