
Floating-Point Division and Square Root 451
8~i 6
Floating-Point Division and Square Root
We now consider algorithms and implementations for floating-point division and
square root. The algorithms are given in generic terms, while implementations
are determined by the IEEE standard.
Division: Algorithm and Basic Implementation
The operands are x and d, represented by
(M i, Ex)
and (M~, Ed), with M]
and M~ signed and normalized. The result
q = x/d
8.76
is represented by
(Mq, Eq ),
with
Mq
also signed and normalized. The high-level
description of the floating-point division algorithm is composed of the following
steps:
1. Divide significands ...