Digital Arithmetic

Floating-Point Addition

427

C, Exponent difference )

C Swap )

C )

( Add significands )

C .oo )

C )

C

Special cases

)

FIGURE 8.7

Dependence graph for basic implementation of floating-point addition.

one double-path implementation. Other variations are described in the references

listed at the end of the chapter.

Single-Path Implementation 16

This implementation is shown in Figure 8.8. It results from the following changes

to the basic implementation considered before:

1. The sign-and-magnitude addition is performed using a two's complement

adder. When an effective subtraction is performed, one of the operands is

complemented (bit-inversion ...