From: Tudor <tudor AT cam DOT org>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Floating Point vs. Fixed Point
Date: Tue, 04 Mar 1997 14:01:30 -0800
Organization: Communications Accesibles Montreal
Lines: 23
Message-ID: <331C9BBA.4C95@cam.org>
References: <5fhlm2$du6 AT alpha DOT hcst DOT com>
Reply-To: tudor AT cam DOT org
NNTP-Posting-Host: dynppp-22.hip.cam.org
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

Bryan Murphy wrote:
> 
> I'm curious as to what advantages fixed point arithmatic has
> over floating point in relation to today's Pentiums and Pentium
> pro's.  I know traditionally Fixed point has been quicker, but
> is it losing that advantage?  Does MMX also promote the use of
> floating point over fixed point?  A lot of programs are being
> written using floating point now a days, and I'm curious if I
> should follow.
Yes, initially fixed point was faster than floating point (on 486's and
lower and especially on machines without a FPU/maths coprocessor).
But on a Pentium floating point is faster (ie. Shawn added floating
point opperations to Allegro; if you tartget pentium & up then use
floating point, else use fixed point).
Another example is Quake.
About MMX: I dunno.I heard you cannot use MMX instructions and floating
point instructions at the same time; they use the same pipe or
something.
cheers,
-- 
tudor 'at' cam 'dot' org
http://www.cam.org/~tudor
'This is Scott Nudds of the Borg. C is irrelevant.'