VMD-L Mailing List

From: John Stone (johns_at_ks.uiuc.edu)
Date: Mon Jan 27 2003 - 22:54:53 CST

Hi guys,
  I've had a few emails back and forth with Andy Ritger at
NVidia. I've made his group aware of the problem, along with
a few others I've been aware of for a little while. Andy didn't
promise anything, but he did say that the engineers would discuss
it and get back to me when they get a chance. Since they are now
aware that a number of programs all exhibit the same problems when
using their drivers in a stereo visual, hopefully this will justify
more attention. If I hear anything definitive from them on this
I'll post to VMD-L again with results.

  John

On Fri, Jan 24, 2003 at 05:58:35PM -0600, John Stone wrote:
>
> Michael,
>
> On Fri, Jan 24, 2003 at 02:43:16PM -0600, Michael Redmond wrote:
> > John,
> > I actually do use that command when running scripts, but I usually hadn't
> > included it in the .vmdrc unless I was scripting a specific demo with
> > Windows (at least my current Windows version of VMD does not read scripts).
>
> You mentioned that your current Windows version of VMD doesn't read scripts,
> can you expand on that? It should read scripts just fine. If you need help
> with something there, let me know. On windows, rather than a .vmdrc file
> however, VMD looks for a "vmdrc" file (no ".") which is stored in the VMD
> installation directory rather than in the user's home directory.
>
> > So I know how to do this and, for now, that is a reasonable solution. I can
> > probably even script it to come up with 0.0 eye separation, so to get
> > stereo, I would just need to increase that.
>
> Yes, you could do that too as per Warren's suggestion.
>
> Interesting about the glasses losing sync. I don't have any specific
> suggestion there, hopefully the NVidia staff can offer more ideas on
> that specific issue.
>
> By the way, I heard from some others today that the Quadro speed hit
> you're having occurs on several other molecular viz programs:
> VMD, PyMol (unpatched anyway), O, Ensight, and others.
>
> It would be interesting if you know of any other programs, if we sent NVidia
> a list of the affected programs, maybe they will take the problem more
> seriously, apparently other people have known of this Quadro problem
> for a while. So far I haven't heard anything back from NVidia today.
>
> Thanks,
> John Stone
> vmd_at_ks.uiuc.edu
>
>
> > The NVidia Windows drivers have an option to sync to vertical. I have tried
> > to set it but to no effect. It turns out that I use the cheap eDimensional
> > "in-line" shutter glasses (about $70), and the image shearing initially
> > caused them to lose sync (every 12 seconds, there would be 6 "open shutter"
> > glitches 2 seconds apart that corresponded to the shearing). eDimensional
> > has a resync utility that I assume finds a good sync period and then forces
> > the glasses to ignore signals that are outside that period (not sure, but
> > that's how I would do it). This clears up the "open shutter" flashing of
> > the glasses, but the image shearing is still there. Again, this is on our
> > 550XGL card (real cheap stereo at $250 or less). Out 900XGL also exhibits
> > shearing, but it is harder to spot (for us, on one periodic shear in the
> > top 10% of the screen - can't test with "in-line" glasses because the DVI
> > connector/VGA adapter is not compatible with the eDimensional
> > dongle/connection).
> >
> > So, it appears that the NVidia driver recognizes the need to vertical sync
> > and then, for some reason, doesn't actually do it. That's my speculation
> > anyway...
> >
> > This problem does not show up with clone mode (not too surprising). In our
> > initial tests, it also does not appear to show up under Linux with the same
> > shutter glasses. So someone at NVidia knows how to fix it. Interestingly
> > enough, other glitches have show up with Linux stereo, though not yet fully
> > verified. If you go to full-screen mode in stereo on one of our Linus apps
> > (CEI Enliten), one of the buffers gets corrupted. On the Windows side,
> > full-screen openGL stereo (not in a window) also messes up. We can't run
> > full-screen stereo Cortona as a result.
> >
> > Lots of gotchas in the NVidia driver and/or hardware implementation still.
> > Too bad...
> >
> > Thanks
> > Mike
> > ---
> > At 02:18 PM 1/24/2003 -0600, John Stone wrote:
> >
> > >Michael,
> > > You can force VMD to go into stereo mode by default by adding this
> > >command to your individual .vmdrc (Unix) or vmdrc (Windows), or for all
> > >users that run the program by editing the core VMD startup script and
> > >adding it to the end:
> > > (install dir)/vmd/scripts/vmd/vmdinit.tcl
> > >
> > >Here's the command to enable stereo:
> > > display stereo CrystalEyes
> > >
> > >You can enable one of the other stereo modes as well by default, choose
> > >whichever fits your use the best.
> > >
> > >Interesing that you mention their screen tearing problem.
> > >Do you have the driver set to force syncing to the monitor's vertical
> > >retrace/blank? (does their driver do that by itself?)
> > >
> > >If you do stereo without forcing sync to vertical retrace you will get
> > >all kinds of screen tearing. I assume that their driver would do this
> > >by default when its running in a stereo mode however.
> > >
> > >Regarding the "clone mode" stereo. This can be done with any normal
> > >stereo capable system with one of the CyViz stereo splitter boxes.
> > >They are expensive, but they'll do it generically for any workstation or
> > >PC card with stereo outputs. I know that this sort of thing can also be
> > >done on some of the high-end Unix systems. As far as PC's though,
> > >yeah, NVidia's cards are probably the only ones that will do that
> > >presently, at least as far as I know.
> > >
> > > John Stone
> > > vmd_at_ks.uiuc.edu
> > >
> > >On Fri, Jan 24, 2003 at 01:54:26PM -0600, Michael Redmond wrote:
> > >> My workaround in VMD has always been to immediately go to stereo mode
> > >after
> > >> starting the program. I don't know if there is a command line option to
> > >> start in stereo mode. In VMD this isn't a big deal, but some other apps
> > >> (Ensight by CEI in our case), it is a hassle to make have scripts
> > >include a
> > >> stereo command before doing anything else. All operations are slowed by
> > >the
> > >> mono performance hit because of slow image update speeds.
> > >>
> > >> The NVidia cards are also giving me problems in shutter glasses mode
> > >(only
> > >> under Windows at this point) in that the image "shears" at periodic
> > >> intervals (6 shears 2 seconds apart followed by stable image for 12
> > >> seconds...this with 120Hz refresh and any type of shutter glasses on
> > >> 550XGL). It is a noticeable glitch but one that we are living with for
> > >now.
> > >> Hopefully NVidia will fix it if they first will acknowledge it is an
> > >actual
> > >> problem (or undocumented feature I guess).
> > >>
> > >> Since I work with clone mode stereo for dual projector passive stereo
> > >> display, the NVidia cards appear to be the only choice. I haven't seen
> > >this
> > >> feature in other cards yet.
> > >>
> > >> Thanks for the input. I sent John's message on the the PNY/NVidia folks
> > >and
> > >> hopefully something will come of it.
> > >>
> > >> Mike
> > >> ---
> > >...
> > >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: johns_at_ks.uiuc.edu Phone: 217-244-3349
> > > WWW: http://www.ks.uiuc.edu/~johns/ Fax: 217-244-6078
>
> --
> 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: johns_at_ks.uiuc.edu Phone: 217-244-3349
> WWW: http://www.ks.uiuc.edu/~johns/ Fax: 217-244-6078 

-- 
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: johns_at_ks.uiuc.edu                 Phone: 217-244-3349              
  WWW: http://www.ks.uiuc.edu/~johns/      Fax: 217-244-6078