
254 r N ~, pT E ~ 5 Division by Digit Recurrence
,
iteration is done using a carry-free adder (and is, therefore, fast). 6 Its
disadvantages are that it complicates somewhat the quotient-digit selection
and that it increases the number of register bits required to store the
residual. Moreover, if the remainder is needed, the final residual has to be
converted to conventional representation. Even if the remainder is not
needed, a sign detection has to be implemented for the correction step,
indicated in Section 5.2.2.
Quotient-digit selection function. The complexity of the implementation of
this function depends on all the previous factors. ...