Inside the Intel 8087 chip: floating-point addition and subtraction