Er… this may be a stupid question, but how is this actually different to ordinary floating-point arithmetic? Ordinary IEEE-754 floating-point numbers already st

Er… this may be a stupid question, but how is this actually different to ordinary floating-point arithmetic? Ordinary IEEE-754 floating-point numbers already store the mantissa and exponent in separate fields, don’t they?