Floating-point unit
Circuitry that does arithmetic on floating-point numbers, which store very large and very small values with a fixed number of significant digits.
Also: FPU, maths coprocessor, x87
Floating-point numbers hold a value as digits and an exponent, like scientific notation, so they can represent fractions and huge ranges. Doing this arithmetic in software is slow. Early PCs could add a separate maths coprocessor, such as the Intel 8087 for the 8086 or the 80387 for the 386. The 486DX of 1989 was the first mainstream x86 chip with the FPU built in. Today every CPU core has floating-point and vector units, and GPUs are built around them. Supercomputer speed is measured in floating-point operations per second.