Intel 8087

The Intel 8087 were the mathematical first coprocessors designed by Intel in 1980.

They were built to be used with the Microprocesseur S Intel 8088 and 8086. The goal of the 8087, the first of the x87 family, was to accelerate calculations for applications requiring a treatment with numbers of floating decimal point. The performances of execution increased by 20% to 500% according to the applications.

This coprocessor introduced approximately 60 new instructions available for the programmer, all starting with “F”; to differentiate them from the internal arithmetic instructions (in Integer) of the 8086/88. For example, for instructions ADD/MUL, the 8087 provided instructions FADD/FMUL.

The 8087 (and, in fact, all the family of the x87) do not provide a linear whole of Registre S such as registers AX/BX/CX/DX of processors 8086/88 and 80286 -- the registers x87 are structured in a certain form of pile (however it is not exactly like a typical structure of data of pile) extending from ST0 with ST7. The instructions in floating decimal point of the coprocessors x87 operate push and the pop ones of values on this pile.

When Intel conceived the 8087, it aimed at making of it a general format of floating decimal point for future chips. In fact, one of the most successful things of this coprocessor from a historical point of view was the introduction of the first standard of floating decimal point for the PC containing x86: the IEEE 754. The 8087 provided two types of source data in floating decimal point into 32/64 bits and an internal support extended to the data 80 bits to improve the precision on large large and complex calculations. Independently of this, the 8087 offered a format 80 bits to 17 digits packed in BCD and a format of data 16,32 and 64 bits in the entireties.

The 8087, announced in 1980, were replaced by the 80287, 80387DX/SX and the 487SX. The Intel 80486, the Pentium and the processors later include a coprocessor integrated in the middle of the processor.

Random links: