From: arcanix AT erols DOT com (Francisco Gochez)
Newsgroups: comp.os.msdos.djgpp
Subject: Re: "Are Allegro's routines fast enough to write Quake-like games?" - No. HUH?
Date: Mon, 04 Aug 1997 08:26:43 GMT
Organization: None
Lines: 16
Message-ID: <5s3pa8$k0f@winter.news.erols.com>
References: <199705232152 DOT QAA08574 AT rrnet DOT com> <33875EFC DOT 2306 AT imag DOT net> <5m8o7e$mo6 AT freenet-news DOT carleton DOT ca> <338b7ff5 DOT 3171460 AT news DOT cybermax DOT net> <5n24bf$ert AT nr1 DOT toronto DOT istar DOT net> <33953049 DOT 82D8D021 AT alumnos DOT inf-cr DOT uclm DOT es> <33A3F6C0 DOT 305A8DFF AT execulink DOT com> <33A3FD8C DOT 17E3 AT cs DOT com> <Pine DOT GSO DOT 3 DOT 96 DOT 970617181644 DOT 8410A-100000 AT saule> <33A6F23E DOT 28E5 AT cs DOT com> <33e1da4a DOT 1183603 AT news DOT zip DOT com DOT au> <KmVNwRARye4zEwX$@jenkinsdavid.demon.co.uk> <33e24af0 DOT 15845161 AT news DOT btinternet DOT com>
Reply-To: arcanix AT erols DOT com
NNTP-Posting-Host: nyd-as5s18.erols.com
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

T DOT Harte AT btinternet DOT com (Thomas Harte) wrote:

>	If any of you are actually interested, I did a one room demo with
>Allegro's t-mapping functions just to check the speed. It's a convex room, so
>there is no order sorting (as there wouldn't be with a BSP tree), and it is
>completely closed (so every pixel is written to, once), and uses clipping code.
>Not mine though. SillyWabbit is the name in the readme. And who am I to argue.


The thing is, Quake doesn't render polygons (AFAIK), it renders spans
which are emitted from a global edge list, used to eliminate overdraw.
The only things in Quake which are rendered as polygons to my
knowledge are the alias models, which are z-buffered.  This means that
no matter how fast Allegro's routines are, you probably couldn't reach
Quake's speed.