Stands for "Floating Point Unit." The first computer processors were far better at dealing with integers than with real numbers (a.k.a. floating point numbers). So a separate FPU processor was developed to handle the floating point calculations. That way, when the CPU encountered a floating-point expresion (ie. 1.62 * 0.87359), it would send the calculation to the FPU. Since the FPU is specifically designed to handle floating-point math, it computes expressions involving real numbers more efficiently. While the first floating point units used to be manufactured as individual chips, they are now typically integrated into the CPU.