Re: [AL] Linear Frame Buffer

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 454-5900

On Fri, Aug 03, 2001 at 10:20:10AM +0100, George Foot wrote:
> On Fri, Aug 03, 2001 at 10:03:21AM +0200, Rudolf Polzer wrote:
> > For me it is no real problem - according to Bob now VESA1 will be
> > used on these cards, so the problem disappeared.
> 
> I think Bob just said that VESA1 *should* be used on those cards
> -- because the card shouldn't claim VESA2 support.  We can't
> help it if manufacturers release buggy drivers though.

In this case the non-flashable BIOS is buggy. When using
the TSR from SiS, everything works - but then Win9x cannot
boot any more.

> > It was an older Allegro version which detected VESA2L -
> > according to Bob the current one does not.
> 
> If you have problems with specific drivers because your card is
> buggy (e.g. reports being able to do VESA2 when it can't, or
> when the VESA2 driver in the BIOS is buggy), the correct thing
> to do is make a config file (allegro.cfg) which sets the
> graphics driver you *do* want to use, on autodetect.

Sorry, I forgot that... even if I have one on my Linux box
to use the XWFS driver instead of DGA.

> Alternatively, as I think you already realised, you could declar
> declare your own GFX_DRIVER_LIST with the drivers in a different
> order.  But then your game won't work so well on non-broken
> systems (e.g. it will never detect Vesa2 even on supporting
> cards).

This should not be a big performance difference. Or is it?
>From <all...@ca...> Fri Aug 03 06:23:04 2001
Received: from fruitmachine.brighton.climax.co.uk [212.74.3.86] by canvaslink.com with ESMTP
  (SMTPD32-6.00) id A55A8C803E4; Fri, 03 Aug 2001 05:56:42 -0400
Received: by FRUITMACHINE.brighton.climax.co.uk with Internet Mail Service (5.5.2653.19)
	id <P9J0BM0J>; Fri, 3 Aug 2001 10:53:28 +0100
Message-ID: <28A7C352840AD511AC9400508BE70AC24FF9A1@FRUITMACHINE.brighton.climax.co.uk>
From: Shawn Hargreaves <sha...@cl...>
To: "'al...@ca...'" <al...@ca...>
Date: Fri, 3 Aug 2001 10:53:20 +0100 
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C11C02.20AB8AF0"
Subject: RE: [AL] Linear Frame Buffer
Precedence: bulk
Sender: all...@ca...
Reply-To: al...@ca...
X-UIDL: 297212519
Status: O
Content-Length: 2519
Lines: 70

This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

------_=_NextPart_001_01C11C02.20AB8AF0
Content-Type: text/plain;
	charset="iso-8859-1"

Rudolf Polzer writes:
> > Alternatively, as I think you already realised, you could declar
> > declare your own GFX_DRIVER_LIST with the drivers in a different
> > order.  But then your game won't work so well on non-broken
> > systems (e.g. it will never detect Vesa2 even on supporting
> > cards).
> 
> This should not be a big performance difference. Or is it?

It's a huge difference: VESA 1.x is a 16 bit banked-mode API, while VESA 2.0
allows 32 bit calls without having to switch the CPU mode each time, and
supports linear framebuffers. This gives somewhere like a 2x speed boost for
typical apps, and easily up to 100x in some situations (VESA 1.x is really
badly broken for use by 32 bit programs).

	Shawn.

------_=_NextPart_001_01C11C02.20AB8AF0
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Diso-8859-1">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
5.5.2653.12">
<TITLE>RE: [AL] Linear Frame Buffer</TITLE>
</HEAD>
<BODY>

Rudolf Polzer writes:
 &gt; &gt; Alternatively, as I think you already =
realised, you could declar
 &gt; &gt; declare your own GFX_DRIVER_LIST with the =
drivers in a different
 &gt; &gt; order.&nbsp; But then your game won't work =
so well on non-broken
 &gt; &gt; systems (e.g. it will never detect Vesa2 =
even on supporting
 &gt; &gt; cards).
 &gt; 
 &gt; This should not be a big performance =
difference. Or is it?

It's a huge difference: VESA 1.x is a 16 bit =
banked-mode API, while VESA 2.0 allows 32 bit calls without having to =
switch the CPU mode each time, and supports linear framebuffers. This =
gives somewhere like a 2x speed boost for typical apps, and easily up =
to 100x in some situations (VESA 1.x is really badly broken for use by =
32 bit programs).

<P>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT =
SIZE=3D2>Shawn.</FONT>
</P>

</BODY>
</HTML>
------_=_NextPart_001_01C11C02.20AB8AF0--
>From <all...@ca...> Fri Aug 03 07:01:05 2001
Received: from fruitmachine.brighton.climax.co.uk [212.74.3.86] by canvaslink.com with ESMTP
  (SMTPD32-6.00) id A470A0A03E4; Fri, 03 Aug 2001 07:01:04 -0400
Received: by FRUITMACHINE.brighton.climax.co.uk with Internet Mail Service (5.5.2653.19)
	id <P9J0BNC1>; Fri, 3 Aug 2001 11:57:50 +0100
Message-ID: <28A7C352840AD511AC9400508BE70AC24FF9A3@FRUITMACHINE.brighton.climax.co.uk>
From: Shawn Hargreaves <sha...@cl...>
To: "'al...@ca...'" <al...@ca...>
Date: Fri, 3 Aug 2001 11:57:47 +0100 
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C11C0B.21596810"
Subject: RE: [HTML] RE: [AL] Linear Frame Buffer
Precedence: bulk
Sender: all...@ca...
Reply-To: al...@ca...
X-UIDL: 297212520
Status: O
Content-Length: 3932
Lines: 91

This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

------_=_NextPart_001_01C11C0B.21596810
Content-Type: text/plain;
	charset="iso-8859-1"

Rudolf Polzer writes:
> > It's a huge difference: VESA 1.x is a 16 bit banked-mode API, while VESA
2.0
> > allows 32 bit calls without having to switch the CPU mode each time, and
> > supports linear framebuffers. This gives somewhere like a 2x speed boost
for
> > typical apps, and easily up to 100x in some situations (VESA 1.x is
really
> > badly broken for use by 32 bit programs).
> 
> Yes, you are right. This could be relevant on page flipping,
> but it should not make a big difference double buffering.

That depends on the card. Modern graphics boards (ie. anything made within
the last 5 years) are designed to be accessed through a single huge linear
memory aperture, which VESA 2.0 can map as the linear framebuffer. They also
provide a 64k banked mode mapping at physical address 0xA0000, which is used
by the 16 bit VESA 1.x drivers, but this is only included for backward
compatibility with ancient 16 bit apps, so it is usually a quick/cheap hack
on top of the real hardware that they are concentrating on (it is important
that Wordstar still be able to run on a modern PC, but nobody is likely to
spend time optimising the hardware to make sure it runs quickly :-) The end
result is that you are likely to see much lower bandwidth if you go through
the 16 bit banked aperture: even leaving out the cost of bank switching,
you'll get a huge performance hit on many cards.

	Shawn.

------_=_NextPart_001_01C11C0B.21596810
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Diso-8859-1">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
5.5.2653.12">
<TITLE>RE: [HTML] RE: [AL] Linear Frame Buffer</TITLE>
</HEAD>
<BODY>

Rudolf Polzer writes:
 &gt; &gt; It's a huge difference: VESA 1.x is a 16 =
bit banked-mode API, while VESA 2.0
 &gt; &gt; allows 32 bit calls without having to =
switch the CPU mode each time, and
 &gt; &gt; supports linear framebuffers. This gives =
somewhere like a 2x speed boost for
 &gt; &gt; typical apps, and easily up to 100x in =
some situations (VESA 1.x is really
 &gt; &gt; badly broken for use by 32 bit =
programs).
 &gt; 
 &gt; Yes, you are right. This could be relevant on =
page flipping,
 &gt; but it should not make a big difference double =
buffering.

That depends on the card. Modern graphics boards (ie. =
anything made within the last 5 years) are designed to be accessed =
through a single huge linear memory aperture, which VESA 2.0 can map as =
the linear framebuffer. They also provide a 64k banked mode mapping at =
physical address 0xA0000, which is used by the 16 bit VESA 1.x drivers, =
but this is only included for backward compatibility with ancient 16 =
bit apps, so it is usually a quick/cheap hack on top of the real =
hardware that they are concentrating on (it is important that Wordstar =
still be able to run on a modern PC, but nobody is likely to spend time =
optimising the hardware to make sure it runs quickly :-) The end result =
is that you are likely to see much lower bandwidth if you go through =
the 16 bit banked aperture: even leaving out the cost of bank =
switching, you'll get a huge performance hit on many cards.

<P>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; <FONT =
SIZE=3D2>Shawn.</FONT>
</P>

</BODY>
</HTML>
------_=_NextPart_001_01C11C0B.21596810--
>From <all...@ca...> Fri Aug 03 07:22:51 2001
Received: from post.webmailer.de [192.67.198.65] by canvaslink.com with ESMTP
  (SMTPD32-6.00) id A27C9DC03E4; Fri, 03 Aug 2001 06:52:44 -0400
Received: from www42.durchnull.de (B1bda.pppool.de [213.7.27.218])
	by post.webmailer.de (8.9.3/8.8.7) with ESMTP id MAA09571
	for <al...@ca...>; Fri, 3 Aug 2001 12:49:28 +0200 (MET DST)
Received: from rpolzer by www42.durchnull.de with local (Exim 3.12 #1 (Debian))
	id 15ScNx-0005pG-00
	for <al...@ca...>; Fri, 03 Aug 2001 12:40:53 +0200
Date: Fri, 3 Aug 2001 12:40:53 +0200
To: al...@ca...
Message-ID: <200...@ww...>
References: <28A7C352840AD511AC9400508BE70AC24FF9A1@FRUITMACHINE.brighton.climax.co.uk>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.2.5i
In-Reply-To: <28A7C352840AD511AC9400508BE70AC24FF9A1@FRUITMACHINE.brighton.climax.co.uk>; from sha...@cl... on Fri, Aug 03, 2001 at 10:53:20AM +0100
From: Rudolf Polzer <rp...@du...>
Subject: Re: [HTML] RE: [AL] Linear Frame Buffer
Precedence: bulk
Sender: all...@ca...
Reply-To: al...@ca...
X-UIDL: 297212521
Status: O
Content-Length: 922
Lines: 17

On Fri, Aug 03, 2001 at 10:53:20AM +0100, Shawn Hargreaves wrote:
> Rudolf Polzer writes:
> > > Alternatively, as I think you already realised, you could declar
> > > declare your own GFX_DRIVER_LIST with the drivers in a different
> > > order.  But then your game won't work so well on non-broken
> > > systems (e.g. it will never detect Vesa2 even on supporting
> > > cards).
> > 
> > This should not be a big performance difference. Or is it?
> 
> It's a huge difference: VESA 1.x is a 16 bit banked-mode API, while VESA 2.0
> allows 32 bit calls without having to switch the CPU mode each time, and
> supports linear framebuffers. This gives somewhere like a 2x speed boost for
> typical apps, and easily up to 100x in some situations (VESA 1.x is really
> badly broken for use by 32 bit programs).

Yes, you are right. This could be relevant on page flipping,
but it should not make a big difference double buffering.
>From <all...@ca...> Fri Aug 03 10:36:58 2001
Received: from mail.libertysurf.net [213.36.80.91] by canvaslink.com with ESMTP
  (SMTPD32-6.00) id A7094C003D6; Fri, 03 Aug 2001 10:36:57 -0400
Received: from ybot0001lsurf.fr (213.36.139.110) by mail.libertysurf.net (5.1.053)
        id 3B6A03BD00011C3A for al...@ca...; Fri, 3 Aug 2001 16:33:45 +0200
Message-ID: <004e01c11c29$1d1903a0$6e8b24d5@fr>
From: "Eric Botcazou" <ebo...@li...>
To: "Allegro mailing list" <al...@ca...>
References: <80F65A95CCE1D411BA9300105AA3493202B53D@SSTMAIL> <003101c114ef$9f970b60$4a1724d5@fr> <00c201c11a0c$9df20c20$168224d5@fr> <002501c11a0e$de81d700$0101bfbf@f9c6f0> <018f01c11a14$c0638420$168224d5@fr> <003b01c11a1d$afae6420$0101bfbf@f9c6f0> <008101c11abd$d7e90bc0$621724d5@fr> <002c01c11ac0$1bbe8bc0$e9b386d9@hagatha>
Date: Fri, 3 Aug 2001 16:31:24 +0200
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 5.00.2014.211
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2014.211
Subject: Re: [AL] WIP 3.9.37 - new 'features'
Precedence: bulk
Sender: all...@ca...
Reply-To: al...@ca...
X-UIDL: 297212522
Status: O
Content-Length: 313
Lines: 9

> Are you thinking of CS_BYTEALIGNCLIENT ? :)

Well, it looks like this Class Style flag only does what the docs are saying
it does ;-) More precisely, it probably aligns on a 1-byte boundary because
it has no effects at all on the code.

Ah ! The good old time with monochrome and 4-bit color displays...

-Eric