From:	"Asbjørn" <lordcrc AT hotmail DOT com>
Newsgroups:	comp.os.msdos.djgpp
Subject:	Re: PLEASE! Help optimize my floorcaster....
Date:	Tue, 29 Jul 1997 21:32:39 +0200
Organization:	Cyber Science Technologies
Lines:	37
Message-ID:	<33DE4557.8916F241@hotmail.com>
References:	<33DB24DC DOT 156 AT effect DOT net DOT au> <19970728 DOT 193359 DOT 7127 DOT 0 DOT fwec AT juno DOT com>
Reply-To:	lordcrc AT hotmail DOT com
NNTP-Posting-Host:	nm12-11.ppp.sn.no
Mime-Version:	1.0
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp

Mark T Logan wrote:
> 
> On Sun, 27 Jul 1997 21:37:16 +1100 Josh McDonald <os-dev AT effect DOT net DOT au>
> writes:
> >Hi,
> >       I have a floorcaster going, with bump-mapping implemented by
> >tiles. It runs at a measley 21fps on my p-133 without bump-mapping,
> >and at 2fps with the bump-mapping implemented...
> >
> >       I desperately need this faster, and without assembler :(
> >
> >Does anybody have any ideas, could anybody out there help? I'm not
> >going to use lookup tables for the distance, because it fucks up,
> >and lookup tables for the maximum height a voxel(ish) pixel could
> >be don't speed it up at all......
> >
> >PLEEEEEEEEEEEEEEEEEEEEEEASE!!!!
> 
> Well, there's a lot of little things you could do to speed up your code,
> but the one that jumps out at me is all your floating point math.  You
> want to use fixed point math.  Pentiums do have good floating point
> coprocessors, but I don't know if its fast enough.  Anyone care to
> comment?  Anyway, I can't explain fixed point math right now, mostly
> because I barely understand it.  You can look at the allegro code for
> help, or you can look at almost any game programming book.

the pentium is in some cases faster on floats than ints :) But, if you
have
to do a lot of float->int conversion, you're better off using fixed
point

-- 
- Asbjørn / Lord Crc

http://home.sn.no/~bheid/
lordcrc AT hotmail DOT com

- Raw text -