IBM partially has a hold on banking because they have supported decimal floating point. By switching binary floating-point from radix 2 to radix 10 (decimal ex

IBM partially has a hold on banking because they have supported decimal floating point. By switching binary floating-point from radix 2 to radix 10 (decimal exponents) they get rid of some of the problems. IEEE has had these Decimal types for a while, but IBM is the only company shipping CPUs with them. Fixed decimal is still used where appropriate, but Decimal floating point is a critical feature for some processes. While personal taxes typically drop the cents, tariffs and other taxes are often small percentages as an example. There are software implementations if you don't need performance. But Decimal64 has a lot more represented values and it is easier to trap on some edge cases. As Decimal floating point types were added into C23 we will see if hardware support follows. Note this is also why HP calculators were more accurate for a given bit size, because the engineering ones used radix 10 floats.