Hi,
I can't comment about how the NVidia stereo works with other packages
but there are a few facts about the NVidia drivers that apply equally
to all of us:
- Unlike classic graphics workstations like SGI and Sun, with NVidia
you can have _either_ stereo, or you can have multisample
antialiasing, but you can't have both at the same time (at least on the
cards and drivers I've tested to date).
Since the vast majority of VMD users don't have stereo capable hardware,
the default VMD behavior is to acquire a window with antialiasing.
In order to get VMD to use stereo on an NVidia card, you need to do
this prior to running VMD:
setenv VMDPREFERSTEREO 1
This environment variable tells VMD that you want stereo rather than
antialiasing, and so it will request stereo only. This is probably
functionally equivalent to the stereo settings described below for O.
- NVidia's cards exhibit a huge loss of performance if you acquire a
stereo window, but render to it in a monoscopically.
It would seem that the NVidia drivers do some sort of software framebuffer
copy to present the same image to both eyes, when drawing monoscopically
to a stereo-capable window. On SGI/Sun workstations and on PC video
cards by 3DLabs, there isn't any performance penalty for this,
they run full speed regardless whether the window is stereo-capable
or not. So, on the NVidia cards, you only want to run VMD with a
stereo window if you actually plan to use it. I may consider adding
some NVidia specific code to workaround this performance issue,
but really this is due to the NVidia hardware/software implementation,
and not the fault of O/PyMol/VMD, etc.
John Stone
v...@ks.uiuc.edu
On Wed, Jul 27, 2005 at 04:03:58PM -0700, Cameron Mura wrote:
> Demetres & Roger,
>
> I've had the same problem w/ stereo on FC4 (tested only w/ an nVidia
> Quadro 700 XGL card), and have been meaning to mail the list. The
> problem occurs even after taking the typical steps to set-up the nVidia
> drivers on the system (as described by Roger), and also occurs w/
> hadrware quad-buffered stereo in other molecular graphics software
> (e.g., PyMOL and VMD -- so I've cc'd this email to those lists too). I
> haven't had time to really troubleshoot it yet (by fiddling around w/ X
> and whatnot), so if anyone else sees this stereo/FC4 problem -- or, even
> better, can explain it! -- it'd be great to hear about it...
>
> Thanks,
> Cameron
>
>
> === o-info-requ...@bioxray.dk wrote (on 07/27/2005 03:46 PM): ===
>
> >Subject:
> >[o-info] Problems with stereo emitter
> >From:
> >"Demetres D. Leonidas" <d...@eie.gr>
> >Date:
> >Wed, 27 Jul 2005 16:39:13 +0300
> >To:
> >o-i...@o-info.bioxray.dk
> >
> >To:
> >o-i...@o-info.bioxray.dk
> >
> >
> >Dear Ousers,
> >
> >We have just got a Celsius K330 workstation with an NVIDIA Quadro
> >FX1300 128MB PCI-Ex stereo card and installed FC4. I did all the
> >necessary modifications to the xorg.conf file to enable STEREO with O
> >and when I hit F1 I get in the screen the stereoview. However, the
> >emitter (NuVision 60GX) does not flash and hence the glasses do not
> >work. The emitter is plugged on the 3-pin plug in the card. What do I
> >do wrong ? Is there a special connector that I should use.
> >
> >
> >
> >Many thanks
> >
> >
> >
> >Demetres
> >
> >
> >------------------------------------------------------------------------
> >
> >Subject:
> >RE: [o-info] Problems with stereo emitter
> >From:
> >"Roger Rowlett" <rrowl...@mail.colgate.edu>
> >Date:
> >Wed, 27 Jul 2005 11:05:55 -0400
> >To:
> >"Demetres D. Leonidas" <d...@eie.gr>, <o-i...@o-info.bioxray.dk>
> >
> >To:
> >"Demetres D. Leonidas" <d...@eie.gr>, <o-i...@o-info.bioxray.dk>
> >
> >
> >The following steps are typical for enabling stereo in O 9.0 with an
> >Nvidia card and NuVision 60GX emitter/glasses:
> >
> >1. Install the proprietary Nvidia graphics driver. In the XF86Config (or
> >its equivalent), the following settings should be included under the
> >Videocard0 device:
> > Driver "nvidia"
> > Option "Stereo" "3"
> >
> >2. Plug the NiVision emitter 3-pin mini-DIN connector to the appropriate
> >graphics card socket.
> >
> >3. Start O with stereo enabled, which will require setting the
> >environment variable STEREO to "on". I use a script, ono9-3d:
> >
> >#! /bin/tcsh
> >setenv STEREO on
> >setenv ODAT /usr/local/bin/ono9/data/
> >/usr/local/bin/ono9/lin_ono
> >
> >When O is started, you should see the red light on the stero emitter
> >immediately, whether or not you are using stereo.
> >
> >It's useful to have a script for both 3d-O and 2d-O as for some Nvidia
> >drivers, as graphics response can be slow when the environment variable
> >STEREO is set to "on" but stereo viewing is not enabled. I have a second
> >script, ono9, which is identical to ono9-3d except that the second line
> >(setting STEREO on) is omitted. This allows O to run quickly in 2d mode,
> >for those occasions when I don't want to use the glasses.
> >
> >I'm using RH9, an NVidia Quadro 980XGL, and the Nuvision 60GX and
> >everything works flawlessly, other than the "slow non-stereo response"
> >bug when stereo is "on". I'm still using version 4363 NVidia drivers, as
> >there are issues with some legacy software with the newer drivers. Works
> >fine anyway. Stereo also works flawlessly in VMD, PyMol, MolMol, and
> >DeepView.
> >
> >Cheers,
> >
> >____________________________________________
> >Roger S. Rowlett
> >Professor & Chair
> >Department of Chemistry
> >Colgate University
> >13 Oak Drive
> >Hamilton, NY 13346
> >
> >tel: 315-228-7245
> >ofc: 315-228-7395
> >fax: 315-228-7935
> >email: rrowl...@mail.colgate.edu
> >web: http://departments.colgate.edu/chemistry
> >
> >
> >
> >
>
--
NIH Resource for Macromolecular Modeling and Bioinformatics
Beckman Institute for Advanced Science and Technology
University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
Email: jo...@ks.uiuc.edu Phone: 217-244-3349
WWW: http://www.ks.uiuc.edu/~johns/ Fax: 217-244-6078
