Multiple precision

Multiple precision arithmetic (sometimes called extended precision) is arithmetic carried out at higher precision (i.e. using binary representation with more bits than the usual length, usually a word in most early machines). In early computers this was usually not performed by hardware, but rather in software (although often in machines that did that, registers, such as a 'carry' bit, and instructions, such as a 'sign extension' operation, were provided, to make it easier).

Floating point was often supplied in 'normal' and multiple precision forms.