Re: GLITN and latest Xfree
On Wed, Mar 12, 2003 at 12:21:23PM +0100, Ivan Porro wrote:
> HI to all,
>
> I'm newbee to this list, and I know I'm going a bit OT 'cause this is not a
> "user support" list. But I'm really in trouble with an Oxygen GVX210 card
> (glint r3 chips) and latest XFree. The trouble is: apparently the card is
> recognized and X running fine, but how can I be sure that hardware
> acceleration is enabled?
> In fact, the same program compiled under windows2000 with proper 3dlabs driver
> has better performances (it's a simple wireframe cube with a lot of point
> inside plotted using OpenGL.
Well, the glint driver supports 2D acceleration and Xv acceleration
only, DRI support is only available for the GMX2000, which used MX
rasterizer and not PM3 ones. I tried to do the port, but hadn't had
enough time, and anyway, i made the mistake to get myself an AGP8x
motherboard where i cannot plug my Jeronimo 2000 board anymore (1 gamma
and 2 pm3 (one for each head), but no pm3 interleaving is supported).
Also i am not sure if the scanline interleaving is supported (that is,
you sue both rasterizer to render to a single head).
If you are interested, you could try porting the gamma DRI module
yourself, i have already done what is needed for both the X driver and
the kernel driver to support the gamma + pm3 combo (the easy part), but
now only the actual gamma accel code in the mesa driver needs to be
updated. The gamma part is already working, the only thing that is
needed is to modify the code that uses the MX rasterizer to use the PM3
instead. You would need docs and such, but if you ask 3Dlabs about it
and sign their NDA, you should be able to get it.
John Gay <johngay@eircom.net> is also interrested on doing this, and i
can help as much as my time permitts.
There is also a tdlabs-0-0-1-branch in the DRI CVS where we can work on
and where my early code can be found. You would need to ask Alan
Hourihane <alanh@users.sourceforge.net> to update it to the 4.3.0
released state though.
Friendly,
Sven Luther
Reply to: