From: Shawn Hargreaves <Shawn AT talula DOT demon DOT co DOT uk>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Allegro Vesa mode graphics
Date: Fri, 31 Jan 1997 19:39:41 +0000
Organization: None
Distribution: world
Message-ID: <1qOuNAA9pk8yEw0s@talula.demon.co.uk>
References: <gRdbzIAw+P8yEwlA AT talula DOT demon DOT co DOT uk>
 <Pine DOT SGI DOT 3 DOT 93 DOT 970131151924 DOT 1275C-100000 AT gibson DOT eee DOT upd DOT edu DOT ph>
NNTP-Posting-Host: talula.demon.co.uk
MIME-Version: 1.0
Lines: 28
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

Orlando Andico writes:
>> VBE 2.0 is still faster, though, just not by quite as much as you
>> implied :-)
>
>I only use VBE 2.0 with UniVBE, I don't use DOS too much. Under
>Linux/XFree86, the card can be made to use a "linear framebuffer." Is this
>the same functionality as VBE 2.0? the X docs claims that LFB's don't give
>much of a speed improvement, at least on my card (CL5428) (which I realize
>is something of a lamer..)

Under DOS, VBE 2.0 provides an API for accessing linear framebuffers,
and also a protected mode interface that can be used for the standard
bank switching routines. Supporting p-mode directly gives a _huge_ speed
increase over VBE 1.x, which requires you to call a real mode interrupt
function, but obviously that isn't relevant to a real operating system
like Linux :-) The linear framebuffer is only available if the hardware
supports it, and is faster because it removes the need for bank
switching altogether (typically two or three port writes per switch),
and because a lot of recent graphics cards can transfer data faster via
linear framebuffers than with the old VGA-compatible apertures at
0xA0000. Drawing 32x32 sprites to random screen locations on my Matrox
Mystique, linear modes are faster than banked access by a factor of
around 14%.

/*
 *  Shawn Hargreaves - shawn AT talula DOT demon DOT co DOT uk - http://www.talula.demon.co.uk/
 *  Ghoti: 'gh' as in 'enough', 'o' as in 'women', and 'ti' as in 'nation'.
 */