
Implementations of the Division Algorithm
269
.
The quotient-digit selection has as arguments the truncated carry-save
shifted residual ~" and the truncated divisor d" (this is in contrast to the
radix-2 case, in which the selection is independent of the divisor). As
presented in detail in Section 5.5, the selection is described in terms of
selection constants m k
(i)
so that
qj+l - k if mk(i) < ~ < mk+l(i) k ~ {-2, -1, 0,1, 2}
where
A A
9 i --
16 d and d is the divisor truncated to the fourth fractional bit.
Sincel/2<d < 1, weget8<i < 15
9 ~" is 4w[j] in carry-save form and truncated to the fourth fractional bit.
Its range ...