Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!apple!brutus.cs.uiuc.edu!uakari.primate.wisc.edu!uflorida!stat!fsu!geomag!prem
From: prem@geomag.fsu.edu (Prem Subrahmanyam)
Newsgroups: comp.sys.amiga.tech
Subject: Re: Video Files
Message-ID: <348@fsu.scri.fsu.edu>
Date: 9 Nov 89 21:41:44 GMT
References: <1585@ultb.UUCP>
Sender: news@fsu.scri.fsu.edu
Reply-To: prem@geomag.UUCP (Prem Subrahmanyam)
Organization: Florida State University Computing Center
Lines: 24

In article <1585@ultb.UUCP> dds3769@ultb.UUCP (D.D. Simmons) writes:
>	I'd appreciate the help with this and if you could send
>	me the player uuencoded I'd be a happy camper. To the person 
>	who posted the ftp address for uRay. Thanks. And do you
>	have any suggestions on which libraries I should compile
>	it with for shortest rendering time? I'm running with a 
>	stock Amiga 2000 and have plenty of ram to spare.
>
>		_ Derek Simmons _

   As always, for the fastest results (running minus a coprocessor), use
   the lcmffp.lib (Motorola Fast Floating Point library).  Compile the
   program with a -ff in the "lc" command, and you're all set.  This 
   library sacrifices a little accuracy for the sake of speed (unnoticeable 
   in Amiga-based ray-tracers).  The standard lcm library included is an
   IEEE-based library and is quite a bit slower, but more accurate.  

   In general, it seems that unless you're doing something chaotic (Mandelbrot
   /Julia sets, other fractals, etc.) that the extra accuracy given by the
   standard library is not useful....the fast floating point seems to do better.

   If you have a coprocessor, this would change--use the specific libaries for
   your FPU.
   ---Prem Subrahmanyam (prem@geomag.gly.fsu.edu)