Re: [Dri-users] Lemme get this straight...

[Linux G. <lin...@co...>]

------Original Message------
From: Jon Pennington <dr...@wh...>
To: dri...@so...
Sent: February 23, 2001 5:25:15 AM GMT
Subject: Re: [Dri-users] Lemme get this straight...

[snip]

As far as I have found, and as far as anyone has been able to tell me, there
has NEVER been XFree86-only support for the G*4*5*0 (ASCII email prevents me
from emphasizing the `5' in that model number any more.  The G400 works fine
the G450 does NOT).  Anyone with a G450 must use a Matrox-provided binary or
built XFree86 against their HALlib.  I aim to be involved in an effort to
fix that.  I would forgo the use of the TV-out option on the second head if
the rest of the XFree86 driver worked at all, and would be simply ecstatic
if DRI worked on the primary head in a dual-head configuration.

Do you own a G450?  I don't mean to be rude or irreverant, because I really
do appreciate this banter and useful feedback.  The G450 isn't a very
well-documented piece of hardware just yet, and I'm starting to regret
spending my money on it instead of a real 3D card like a Radeon.  I
primarily wanted this card for it's video production (Xv, DVD, etc)
potential, and will continue to use it for that as soon as it WORKS ;).

--------------------------------------------

Jon,

I have both a G400 and a G450-DDR.  I have used the Matrox binaries and have
had wonderful success with DRI under X-4.0.2.

I have also had success with the X-4.0.99 CVS -- though not tested as
thoroughly with hours of Q3A and Descent.  Note that the X-4.0.99 has *N*O*
(sorry, same ascii limitation) dependence on the proprietary HALlib binary. 
This was tested, albeit limited in scope, on  BOTH cards (i.e. the G450
too).  While it takes a fair amount of time to tweak and configure, it's
hard to imagine that you can't get it to work at all.

If a host.def, XF86Config, kernel config, HOWTO, or any other configuration
file would help you push through a working unit, post a request along with a
description of the present state of your implementation either here or (dare
I suggest) to the Matrox Linux forum.  "Ben"  (the main fire-fighter on the
forum) is usually quick to find your bottlenecks.

As for the differences between the kernel mga.o and the X-sources mga.o --
there was a time when both kernel and the X-4 development would outpace the
other.  Header files and dependencies would change underneath each other and
a lot of things in X would break.  IMO, this is why there are two versions,
so that each could have a working version underneath their own sources. 
However, the differences between the two modules presently is minimal, and I
expect that they'll merge back together now that both the kernel and X-4
development has somewhat leveled off.  I'm not a developer for either, so
take my opinion at face value.

Message void if penguin violated.
Don't mess with the penguin.
______________________________________________
FREE Personalized Email at Mail.com
Sign up at http://www.mail.com/?sr=signup