Comparison of Floating-Point Numbers Absolute Values when Representation of the Exponents with Sign-Inverse Two’s Complementary Code

摘要

Comparison of floating-point numbers absolute values is implemented by a joint one-step analysis of operands exponents and fractions. Rules for determination of comparison signs for floating-point numbers absolute values, when exponents and fractions are canonically represented, are formulated and mathematically confirmed. An algorithm for a modulo adder usage for initial floating-point values fractions comparison is analytically substantiated. An operation’s implementation with the usage of sign-inverse two’s complementary encoding is proposed for original operands comparison in a scheme of joint adder of exponents and fractions, at the standard outputs of which floating-point numbers absolute values comparison signs are uniquely and consistently formed.