Description; Programming Interface - IBM 5150 Technical Reference

Description
The Math Coprocessor (8087) enables the
IBM
Personal
Computer to perform high-speed arithmetic, logarithmic
~
functions, and trigonometric operations with extreme accuracy.
The 8087 coprocessor works in parallel with the microprocessor.
The parallel operation decreases operating time by allowing the
coprocessor to do mathematical calculations while the
microprocessor continues to do other functions.
The first five bits of every instruction's operation code for the
coprocessor are identical (binary 11011). When the
microprocessor and the coprocessor see this operation code, the
microprocessor calculates the address of any variables in memory,
while the coprocessor checks the instruction. The coprocessor
takes the memory address from the microprocessor if necessary.
To gain access to locations in memory, the coprocessor takes the
local bus from the microprocessor when the microprocessor
finishes its current instruction. When the coprocessor is finished
with the memory transfer, it returns the local bus to the
microprocessor.
The
IBM
Math Coprocessor works with seven numeric data types
divided into the three classes listed below.
• Binary integers (3 types)
• Decimal integers (1 type)
• Real numbers (3 types)
Programming Interface
The coprocessor extends the data types, registers, and instructions
to the microprocessor.
Coprocessor 2-3