[PyMOL] Re: [o-info] Q&A Quadro4 900 XGL under Linux

[Warren L. D. <wa...@su...>]

On Thu, 3 Oct 2002, Alwyn Jones wrote:

> >Quadro4 900 XGL under Linux
>
> we now have this card running on a linux system (AMD XP1700).
>
> the stereo performance is stellar with a suitable high refresh
> monitor, NuVision glasses and emitter.
>
> the only problem comes when you switch back to mono with F1. the
> performance (as judged by spin, for example) then drops by ~10.

There has been some discussion of this on the PyMOL list as well.  All
nVidia stereo-capable cards experience this problem with O and PyMOL when
stereo is enabled (but not activated).  I think the GLUT library is
partly to blame, but the problem is specific to nVidia hardware.

Presumably this is because both programs are opening a GLUT stereo window
under mono mode -- nVidia's hardware or drivers have a unique problem with
this configuration, which competiting options (ATI, SGI, 3DLabs) do
not share.  Note that this problem is not OS-dependent -- it also occurs
under Windows -- and it doesn't matter which OEM manufacturer built the
card -- the problem lies with nVidia chips or low-level drivers.

> if you start with a mono visual (i.e. setenv STEREO OFF), performance
> is again excellent.

> this is the first linux stereo card that i have seen that i can recommend.

I think this "bug" is a major drawback with nVidia cards.  My experience
has been that the FireGL2's are a better choice for stereo under Linux,
with the caveat that you have to be very careful about running a version
of Linux compatible with their fickle drivers.  The new ATI FireGL X1 may
offer the best stereo performance of any card available right now, but I
haven't yet had the privilege of being able to try one.

Considering the importance of stereographics to structural biology,
computational chemistry, and bioinformatics, I really hope one or both
manufacturers will start taking stereo support under Linux (as well
as Windows) more seriously.

The pharmaceutical industry as well as thousands of academic researchers
have a significant amount of money to waiting to be spent on solid PC
stereo-graphics solutions, if and when such solutions become available.

Cheers,
Warren