Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!pt.cs.cmu.edu!a.gp.cs.cmu.edu!koopman
From: koopman@a.gp.cs.cmu.edu (Philip Koopman)
Newsgroups: comp.lang.forth
Subject: Re: Signed Integer Division
Message-ID: <7411@pt.cs.cmu.edu>
Date: 23 Dec 89 12:54:49 GMT
References: <75.UUL1.3#5129@willett.UUCP>
Organization: Carnegie-Mellon University, CS/RI
Lines: 29

In article <75.UUL1.3#5129@willett.UUCP>, GEnie@willett.UUCP (ForthNet articles from GEnie) writes:
> R.BERKEY [Robert]            at 01:00 PST
> I'd like to see your analysis that symmetrical division is faster than floored
> division.  I've looked at division a lot, and there's an absence of
> programming examples that use the symmetrical algorithm.
> 
> I wonder if my posting describing an efficient algorithm for implementing
> division was passed along to you or possibly someone else at Harris.

The "normal" division methods based on non-restoring shift and add
seem to favor symmetrical division.  In this paradigm, you require a post-
pass fudge factor to be applied if the division is to be floored.
This takes either time, or extra hardware, or both.
If your machine implements unsigned division as the universal
division primitive (UM/MOD), this is certain to be the case.

By the way, if floored division is implemented as the primary 
division algorithm on the RTX series of processors, it will
break a lot of C code -- and having both algorithms fully supported
by hardware is likely to be difficult.

No, I never saw your algorithm -- please send it to me directly
and I will be happy to look it over -- Thanks.

  Phil Koopman                koopman@greyhound.ece.cmu.edu   Arpanet
  2525A Wexford Run Rd.
  Wexford, PA  15090
Senior Scientist at Harris Semiconductor.
I don't speak for them, and they don't speak for me.