| Title: | Mathematics at DEC |
| Moderator: | RUSURE::EDP |
| Created: | Mon Feb 03 1986 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 2083 |
| Total number of notes: | 14613 |
I'm looking for a routine and/or algorithm to do 64 by 64 bit divides
(using the EDIV intrustrution). I've read (as best I could) Knuth but still
havent got it.
Assume the A is to be divide by B. The 32 bit "digits (base 2**32) are
A[0] A[1] and B[0] B[1]. If B[0] is zero its simple. But if B[0] is
small compared to A[0] its hard to "guess" a good Q[0]. Knuth talks
about normalizing B by multiplying A and B by b/2 (2**16??), but this
would cause an overflow of A and thus require a larger A field.
It would seem easier to use the good old shift and subtract (using
FFS).....
I've read most of the notes on n by n divide but they dont provide
what I'm after.
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 1705.1 | c++ code | RDVAX::GRIES | Tue Dec 22 1992 16:00 | 332 | |
| 1705.2 | RUSURE::EDP | Always mount a scratch monkey. | Tue Dec 29 1992 14:26 | 7 | |
Note .1, which contains Digital-confidential information, has been
hidden just so that it is not accidentally included in mass extracts of
the conference. It contains C++ code to perform 64-bit divisions and
is available to any Digital employee; just ask the moderator.
-- edp
| |||||