From: "Charles R. Bond" <cbond@ix.netcom.com>
Subject: Re: rootfinding
Date: Sun, 19 Aug 2001 13:13:39 -0700
Newsgroups: sci.math.num-analysis
Summary: Numerical algorithms in Binary-Coded Decimal format

Algorithms for calculators are usually based on BCD number
representation. This strategy sidesteps the need to convert
from  user input format to binary before number-crunching,
and from binary to user output format for display.

BCD algorithms, such as the square root algorithm already
discussed and BCD CORDIC, don't usually end up in
computer math libraries. The reason is that computer
algorithms often require dozens (hundreds, thousands,
millions) of internal operations before the result is
made available. Under these conditions, converting to
binary representation and performing binary operations
is more efficient.

You might try looking up pseudo-division on the Internet
for references to similar algorithms.

Juergen Buchmueller wrote:

> Charles R. Bond wrote:
>
> > This method is described in detail in one of the HP Journals,
> > Hewlett-Packard Journal, May 1977, Vol. 28, no. 9, under
> > "Personal Calculator Algorithms I: Square Roots."
>
> Thank you! Now I wish we'd live in a world where I could take a peek at an
> electronic copy of this paper :-)
>
> Do you perhaps know why this algorithm never made it into computer
> languages math libraries? It seems to be a lot less expensive computation
> power wise than other implementations I've seen out in the wild.
>
> Cheers
> Juergen