Hi Jan,

I think I'll have agree to differ on the topic of topic quality of
Open Source OpenGL drivers.  Fully featured and good quality drivers
take time to develop.  It took us a long time to get the OSG feature
competitive with proprietary scene graphs, but now... well most
proprietary scene graphs are now dead, and we've gone from strength to
strength.   I believe once a good open source code base for drivers is
established with a healthy community about them then they have every
chance of doing a good job.

The proprietary approach that NVidia have taken is likely to be short
lived in the grand scheme of computing, give it 5 or 10 years and
NVidia might not even exist anymore, or become something very
different from what it is today.  SGI were once totally dominant, and
now they are just a bit of graphics history.

I recently have begun exploring the world outside NVidia, and you know
while drivers aren't quite as good as NVidia drivers the ATI drivers
aren't actually too bad now - almost all of the OSG works well under
Linux on my ATI card.  Sure the bleeding edge features of OpenGL
extensions aren't there, but not everyone needs this.   My onboard
Intel graphics system even worked pretty well and this is using the
open sourced drivers.

Now I would dearly wish that NVidia get over their proprietary
approach to hardware specs and artificial market segmentation, it
would be a great boon for OpenGL and the greater computer industry if
they did.  NVidia may find they have to move fast as the industry
changes in response to advancing hardware technologies, or just get
squeezed out the market altogether.

Robert.

On Wed, May 20, 2009 at 1:02 PM, Jan Ciger <[email protected]> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi Robert,
>
> Robert Osfield wrote:
>> Hi All,
>>
>> Just caught up with the overnight discussion of frustration with
>> NVidia's artificial restriction of stereo support in OpenGL.
>> Personally I think that NVidia is hurting OpenGL and the stereo
>> graphics market because of this position.
>>
>> The long solution term to this artificial driver restrictions has to
>> be fully featured open source drivers across all platforms, then
>> finally we'd have the ability to go and solve problems directly.
>> Perhaps Gallium3D might eventually be our savior:
>>
>>    http://www.tungstengraphics.com/wiki/index.php/Gallium3D
>>
>> Having competitive open source drivers would also raise the pressure
>> on proprietary driver developers to deliver good stability,
>> performance and all the features that the hardware is capable of.
>
> The main issue is that you cannot have open source drivers without
> documentation, the framework doesn't really matter so much. Reverse
> engineering NVIDIA hardware is perhaps possible for a highly dedicated
> team of experts with top notch equipment, but completely unrealistic
> otherwise. And NVIDIA is not keen on opening up their documentation -
> even after their main competitors did so.
>
> Even if the documentation was in place (ATI, Intel), that doesn't
> guarantee a high-quality driver. Intel driver is out in the open for a
> long time and it is quite terrible these days. There is also the Via
> Chrome driver and it is quite bad as well, despite being open source -
> and I am not speaking about OpenGL support here, only basic stuff like
> not crashing your system.
>
> Furthermore, it will always play catch up because the development can
> start only when the hardware is on the market already. By the time the
> open driver is developed, the hardware would be obsolete.
>
> So while open source drivers are a nice idea, for a market dominated by
> one major player that keeps everything as closed as they can it is a
> non-starter, in my opinion.
>
> The only way forward is to actually put economical pressure on the
> company - when they see that they are losing business due to their
> boneheaded decisions, they will do something about it.
>
> Unfortunately, stereo is such a niche market that unless a huge customer
> weighs in, it is not likely going to happen. And those probably do not
> care about GeForces - the cost of few Quadros would be a drop in the sea
> in the overall costs of their projects.
>
> So I am not optimistic here. Whenever developing something for stereo
> these days, I am focusing on passive stereo that can be done with any
> graphic card. It doesn't have quad buffer, but that is not such a big
> deal for these applications. If shutter glasses or HMD requires
> sequential signal, then the blue line stereo is good enough (and cheap)
> hack - it would be good to have it supported in OSG, btw.
>
> Regards,
>
> Jan
>
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
> Comment: Using GnuPG with Mandriva - http://enigmail.mozdev.org
>
> iD8DBQFKE/Fmn11XseNj94gRAtpuAKCdkCExklHT+57dYWWG9VcZvFj8NwCgmgb9
> zTY0754C/1WEDETuGFELGKU=
> =3FaB
> -----END PGP SIGNATURE-----
> _______________________________________________
> osg-users mailing list
> [email protected]
> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
>
_______________________________________________
osg-users mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org

Reply via email to