From: John Stone (johns_at_ks.uiuc.edu)
Date: Tue Dec 08 2009 - 12:34:57 CST

Hi Rob,
  Appearances can be deceiving. I've got a lot of experience debugging
broken graphics drivers, and I can tell you that the odds currently lay
heavily in favor of your video driver being broken. The most complex parts
of the VMD graphics code are not engaged until one actually loads a
molecule and starts rendering. From the log you provided in your previous
email, it looks to me like you're not even getting completely through
startup before you're getting a crash, which is a strong indicator of
buggy drivers. The only things that VMD is doing at that early point
in the startup are:
  1) creating X11 window and associated OpenGL context
  2) compiling GLSL shaders and enabling them if possible

My feeling is that you're crashing due to bugs in your driver's
implementation of GLSL, because that's the only code that changed
much from VMD 1.8.6 to VMD 1.8.7. The GLSL code isn't even compiled
into VMD in fact, it is loaded by VMD and compiled by the OpenGL driver
at runtime, and if I were to guess, this is probably where your problem
is occuring.

Rather than guess however, you can do a couple of simple tests and
see how far VMD got when the crash occured, by running it like this:
vmd -debug

Once it crashes, type the command "where" into your gdb session, and
that should display a stack trace that shows where the crash is occuring.
We know from your previous log that the crash is occuring inside of one
of the X11 libraries, but we don't know at what point in the VMD code this
was occuring, so running the debugger and getting the traceback with
"where" should also tell us what routine we were in within the VMD code
when the crash occurs, which should tell me what I need to know to track
down the cause of your problem.

I still think you have a 95% chance of a bad video driver as I said before,
but I'm happy to help narrow this down so you can get past this problem.

Cheers,
  John Stone
  vmd_at_ks.uiuc.edu

On Tue, Dec 08, 2009 at 06:05:26AM -0800, Rob wrote:
>
> John Stone wrote on Mon Dec 07 2009:
> >
> > You need to install the proper NVIDIA drivers for your
> > machine. The OpenGL that's currently active is a software
> > rasterizer, and that's definitely _not_ what you want to be using...
>
> I have two PCs with same Nvidia cards "GeForce 8600 GTS (rev a1)".
> One PC is 32 bits with Fedora 12 and here VMD 1.8.7 runs just fine.
> The other PC is 64 bits with Fedora 12 and here VMD 1.8.7 segfaults.
>
> On the 64 bits system, VMD version 1.8.6 runs fine.....
>
> See below for more details.
>
> Does all this make sense?
> So to me it appears that it's not, as you say, that the Fedora NVidia
> drivers are always bad.....
>
> Could it be that there's a bug in 1.8.7 that causes the segfault?
> Or will you tell me that specifically the 64 bits system needs the
> proper NVIDIA drivers?
>
> Thanks.
> Rob.
>
>
> Here is what I get with VMD 1.8.7 32bits Fedora 12:
>
> $ vmd
> Info) VMD for LINUX, version 1.8.7 (August 1, 2009)
> Info) http://www.ks.uiuc.edu/Research/vmd/
> Info) Email questions and bug reports to vmd_at_ks.uiuc.edu
> Info) Please include this reference in published work using VMD:
> Info) Humphrey, W., Dalke, A. and Schulten, K., `VMD - Visual
> Info) Molecular Dynamics', J. Molec. Graphics 1996, 14.1, 33-38.
> Info) -------------------------------------------------------------
> Info) Multithreading available, 4 CPUs detected.
> Info) Free system memory: 1337MB (66%)
> Info) No CUDA accelerator devices available.
> Warning) Detected X11 'Composite' extension: if incorrect display occurs
> Warning) try disabling this optional X server feature.
> Info) OpenGL renderer: Software Rasterizer
> Info) Features: STENCIL MDE CVA MTX NPOT PP PS GLSL(OVF)
> Info) Full GLSL rendering mode is available.
> Info) Textures: 2-D (4096x4096), 3-D (256x256x256), Multitexture (8)
> vmd >
>
> Here is what I get with VMD 1.8.6 on 64bits Fedora 12:
>
> Info) http://www.ks.uiuc.edu/Research/vmd/
> Info) Email questions and bug reports to vmd_at_ks.uiuc.edu
> Info) Please include this reference in published work using VMD:
> Info) Humphrey, W., Dalke, A. and Schulten, K., `VMD - Visual
> Info) Molecular Dynamics', J. Molec. Graphics 1996, 14.1, 33-38.
> Info) -------------------------------------------------------------
> Info) Multithreading available, 8 CPUs detected.
> Info) Free system memory: 5237MB (87%)
> Info) OpenGL renderer: Software Rasterizer
> Info) Features: STENCIL MDE CVA MTX NPOT PP PS GLSL(OVF)
> Info) Full GLSL rendering mode is available.
> Info) Textures: 2-D (4096x4096), 3-D (256x256x256), Multitexture (8)
> vmd >
>
>
> Here is what I get with VMD 1.8.7 on 64bits Fedora 12:
>
> Info) VMD for LINUXAMD64, version 1.8.7 (August 1, 2009)
> Info) http://www.ks.uiuc.edu/Research/vmd/
> Info) Email questions and bug reports to vmd_at_ks.uiuc.edu
> Info) Please include this reference in published work using VMD:
> Info) Humphrey, W., Dalke, A. and Schulten, K., `VMD - Visual
> Info) Molecular Dynamics', J. Molec. Graphics 1996, 14.1, 33-38.
> Info) -------------------------------------------------------------
> Info) Multithreading available, 8 CPUs detected.
> Info) Free system memory: 5145MB (86%)
> Info) No CUDA accelerator devices available.
> Warning) Detected X11 'Composite' extension: if incorrect display occurs
> Warning) try disabling this optional X server feature.
> Info) OpenGL renderer: Software Rasterizer
> Info) Features: STENCIL MDE CVA MTX NPOT PP PS GLSL(OVF)
> Info) Full GLSL rendering mode is available.
> Info) Textures: 2-D (4096x4096), 3-D (256x256x256), Multitexture (8)
> Segmentation fault
>
>
>
>

-- 
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