The very first 16bit Math Coprocessor available for the x86 platform was the 8087. This chip was intended for use in XT Computers with a 8086 or 8088 microprocessor. Intel describes the areas of application as follows
“The 8087 provides functions meant specifically for high performance numeric processing requirements. Trigonometric, logarithmic, and exponential functions are buils into the coprocessor hardware. These functions are essential in scientific, engineering, navigational or military applications”.
The design of the 8087 began in 1976 and the final product was released in 1980. Although the IEEE standard 754 for Binary Floating Point Arithmic was not final until 1985, the 8087 was compatible with this standard for the most part. The reason for this was of course that the 8087 was released five years earlier than IEEE754 became the official industry standard.
Nowadays the Floating Point Unit is integrated into one package with the CPU. Back then this would not have been possible because of the complexity of such a chip and the limitations of the manufacturing process at the time. Intel made the decision to package the FPU as a separate chip. Because there was a need in the market for such a product the idea was also commercial viable. Manufacturing costs would be around $40, but the commercial price in the early days was up to $700 for a 8087.
