The abbreviation "Floating Point Unit" is FPU. A CPU or component of a processor that conducts floating point computations is known as an FPU. While early FPUs were separate computers, the majority of them are now built into the Microprocessor of a computer.

A Processor can perform both integer and floating point (non-integer) computations even without a floating point unit. However, it is wasteful to use the same CPU to perform both kinds of operations because integer operations use reasoning that is very distinct from floating point operations. Calculations involving non-integer values can be handled more quickly with the aid of an FPU.

Either the FPU or the integer processing unit can carry out any arithmetic function, including addition, subtraction, multiplication, and division. A Microprocessor instantly transmits a command to the appropriate processor when it gets one. For instance, while 1.0023 x 5.789 would be routed to the FPU, 12 Plus 5 would be handled as a numeric computation. Though it is technically feasible, it is typically not essential for a coder to create a command particularly for either working device.

Most CPU evaluations include both integer and floating point calculations because of the wide speed variations between the two. Usually referred to as "integer performance," integer computation speed is designated as "SPECint" in SPEC evaluations. The "floating point efficiency" of an FPU is frequently mentioned and expressed in FLOPS.