From: Axel Kohlmeyer (akohlmey_at_cmm.chem.upenn.edu)
Date: Mon Jun 26 2006 - 06:28:22 CDT

On Mon, 26 Jun 2006, Hans Martin Senn wrote:

hans,

your observation is not uncommon. things like this
or similar are known happen occasionally with nvidia
hardware and drivers.

HMS> Dear VMDers
HMS>
HMS> When starting up VMD 1.8.4, I get the error message
HMS>
HMS> ERROR) Compile of OpenGL shader /usr/local/lib/vmd/shaders/vmdsphere
HMS> failed.

HMS> I have appended the full VMD console output below. This is with an
HMS> Nvidia Quadro FX540/PCI/SSE2, OpenGL 2.0.2 Nvidia 87.62, Driver
HMS> 1.0-8762. The system is an EMT64/SuSE 10.1 (Kernel 2.6.16.13-4-smp).
HMS> The resulting GLSL rendering looks fine. However, I would appreciate

really? please try loading two cube files (or equivalent) e.g. with
density and electrostatic potential, create an isosurface representation
for the density data set with 'Volume' coloring using the potential.
now please check, if the color map stays the same, when rotating the
molecule. and then try again with replacing the isosurface with
a VDW representation again using the potential with 'Volume' coloring...

HMS> if somebody could elucidate me about this error message. What's going
HMS> wrong and who is to blame?

i guess that john might have more details about it. but i as far as
i understand it, the file causing the error message is a (shader)
program to be compiled and run 'within' the graphics card. the driver
has to contain some kind of compiler, that compiles this code and
then uploads it to the graphics card. in your case (and several others),
the nvidia driver fails to compile the shader programs and as a
consequence _some_ of the GLSL functionality is broken. in my case
it was the color mapping (when i rotate the representation, the
color mapping changes as if the coloring data set is not moved
along the isosurface data set.

software always contains bugs, so does VMD, but since the shading
language code compiles fine on many machines, it is likely that
nvidia is to blame. john has already a list of open issus (for
quite a while, IIRC) with the nvidia/linux drivers.

greetings from rainy philly,
     axel.

HMS>
HMS> Best regards
HMS> Hans
HMS>
HMS>
HMS> Info) OpenGL renderer: Quadro FX 540/PCI/SSE2
HMS> Info) Features: STENCIL MSAA(16) RN MDE CVA MTX NPOT PP PS GLSL(OVF)
HMS> Info) OpenGL shader linkage log:
HMS> Info) Fragment info
HMS> Info) -------------
HMS> Info) Internal error: assembly compile error for fragment shader at
HMS> offset 3688:
HMS> Info) -- error message --
HMS> Info) line 84, column 9: error: invalid state property
HMS> Info) -- internal assembly text --
HMS> Info) !!ARBfp1.0
HMS> Info) OPTION NV_fragment_program2;
HMS> Info) # cgc version 1.5.0000, build date May 15 2006 13:37:55
HMS> Info) # command line args:
HMS> Info) #vendor NVIDIA Corporation
HMS> Info) #version 1.5.0.0
HMS> Info) #profile fp40
HMS> Info) #program main

[...]

HMS>
HMS> ........................................................................
HMS> .
HMS> Dr. Hans Martin Senn
HMS> Max-Planck-Institut für Kohlenforschung
HMS> Kaiser-Wilhelm-Platz 1 Phone +49 (0)208 306
HMS> 2163
HMS> D-45470 Mülheim an der Ruhr Fax +49 (0)208 306
HMS> 2996
HMS> Germany E-mail senn_at_mpi-
HMS> muelheim.mpg.de
HMS>
HMS>
HMS>
HMS>

-- 
=======================================================================
Axel Kohlmeyer   akohlmey_at_cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.