VMD-L Mailing List

From: John Stone (johns_at_ks.uiuc.edu)
Date: Mon May 18 2009 - 00:15:06 CDT

Cameron,
  Excellent. BTW, I will look into changing the default configure
script behavior to favor builds against /usr/local/* by default rather
than what we use locally. We have special paths for everything to have
tight control over what VMD is linked against, and which version of CUDA
it is compiled with, etc (we have 7 different CUDA versions locally
in both 32-bit and 64-bit variants, so as you can guess picking the one
in /usr/local is almost never the right choice.)

Cheers,
  John Stone
  vmd_at_ks.uiuc.edu

On Sun, May 17, 2009 at 08:07:47PM -0400, Cameron Mura wrote:
> ok, finally -- happily Pythonic VMD on F10!!!:
> >struc3:sandbox:$] ./bin/vmd -python
> >Info) VMD for LINUXAMD64, version 1.8.7beta3 (May 17, 2009)
> >Info) http://www.ks.uiuc.edu/Research/vmd/
> >...snip...
> >...snip...
> >Info) Detected 1 available CUDA accelerator:
> >Info) [0] Quadro FX 4600 12 SM_1.0 @ 1.19 GHz, 767MB RAM, KTO, ZCP
> >Info) Creating CUDA device pool and initializing hardware...
> >Info) OpenGL renderer: Quadro FX 4600/PCI/SSE2
> >Info) Features: STENCIL MSAA(16) MDE CVA MTX NPOT PP PS GLSL(OVF)
> >Info) Full GLSL rendering mode is available.
> >Info) Textures: 2-D (8192x8192), 3-D (2048x2048x2048), Multitexture (4)
> >Info) Starting Python...
> >Info) Text interpreter now Python
> >>>>
> Axel -- you mentioned plans to update your rpm builds for vmd-1.8.7 /
> F10, so tomorrow I'll send you my modifications to the vmd src/Makefile
> as an f10-specific patch... These modifications mostly involved
> utilizing system-wide python2.5, fltk, etc., and also a minor edit of
> the location to which the nvidia CUDA stuff (like nvcc) get located by
> default (by nvidia's "cudatoolkit" package).
>
> possibly relevant sidenote: there's also a memory mismanagement issue
> with the RNG() routine in the default release of Numeric that ships with
> F10 (v24.2-11), so I created a patched release (v24.2-12) that
> essentially just grafts on some changes to the r11 (basically to make it
> behave like Hinsen's v23.8, which plays well with mmtk, and which Marcin
> drew to my attention); that patch is at
> http://muralab.org/~cmura/ComputerStuff/Linux/python-numeric
>
> Thanks again Axel, John, Marcin, for your help.
>
> Cam 

-- 
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