
522 c N ApT E Fto Digit-Serial Arithmetic
In terms of the notation of Section 9.3.1,
-8
gl -- xj+i+sr -- q[j]dj+l+sr
-8
H2 =-d[j + 1]qj+l
9.74
The bound ofw[j] is
2ar -8 - ad[j + 1]
co = - = p(d[j + 1]- 2r -8) 9.75
r-1
The selection intervals on
v[j] are
U k = p(d[j + 1]- 2r -~) + kd[j + 1] 9.76
L k -- -p(d[j + 1]- 2r -8) + kd[j + 1]
Using a carry-save representation for
w[j] and v[j ], the grid-restricted intervals
are
Uk = [p(d[j + 1]- 2r -8) + kd[j + 1]- 2-tit
9.77
L k = [-p(d[j + 1]- 2r -8) +
kd[j + 1]It
The expression to determine t and 8 is
d[j + 1]min([p(d[j + 1]- 2r -8) + (k- ...