Intel had introduced floating point co-processors, the 8087, 80187, 80287 and 80387, for its 8086, 80186, 80286 and 80386. These processors had a totally different design from the CPUs they supported. They merely implemented instructions for floating point operations. The 487 on the other hand was a full-fledged CPU in its own right.
The intel 80486, like earlier intel CPUs, was marketed in two different model. The difference between the 8088 and 8086 and the 80386sx and 80386dx versions of the 80386 was the size of their data buses.
Intel marketed a 80486sx and 80486dx. The 80486dx had full floating point support included. The 80486sx was said to be identical to the 80486dx -- except it had no floating point support.
Motherboards that supported the package intel called the 486sx had an empty socket where users could plug in a 487. In fact the 487 didn't supplement the 486sx, it completely replaced it.
- Omar Alvarado, Thomas J. Fellers, Michael W. Davidson. Intel 487SX Math Coprocessor, Olympus Micro. Retrieved on 2008-09-28. mirror
- Intel and Floating Point: Updating One of the Industry’s Most Successful Standards, intel. Retrieved on 2008-11-28.
- Intel 8087 family, CPU World. Retrieved on 2008-11-28. mirror
- Omar Alvarado, Thomas J. Fellers, Michael W. Davidson. Intel 387 Math Coprocessor, Olympus Micro. Retrieved on 2008-11-27. mirror