This article is from the Amiga FAQ, by Ignaz Kellerer with numerous contributions by others.
The first 680x0 processors (upto 68030) could process integers only. Floating point operations had to be emulated by the software. An FPU is a chip (or part of a chip) that can process floating point operations, a mathematical coprocessor.
One separates three FPU types on the Amiga: The 68881, 68882 and the 68040's internal FPU. The 68882 is up to 1.5 times faster than the 68881, because it is splitted in two parts: A conversion unit (the FPU's are using an 80 bit format internally) and the arithmetic unit. The 68040's internal FPU adds a pipeline, but misses the trigonometric instructions of the others. These are still emulated by the software, 68040.library for example.
Special programs (Raytracing, DTP, Mathematics, TeX) are offered in a special coprocessor version which are up to 50 times faster than the original versions.
Michael Kaiser (email@example.com)