I propose a new approach to the implementation of binary arithmetic on decimal numbers, which gives the decimal precision of binary calculations. This allows you to implement a bitwise comparison of binary numbers in float, double and other formats. The new approach makes it possible to calculate exact decimal values, including when subtracting close and equal binary equivalents of decimal numbers, which makes the process of comparing binary numbers trivial.
All this is achieved using the algorithm for the correct rounding of binary floating point numbers, presented at
https://github.com/yur-spiridonov/Decimal-rounding-of-binary-numbers/commit/0e114f5c0a6a0bfc882e6f92...