molecular modeling on linux
og32 at columbia.edu
Wed Jun 17 15:33:58 EST 1998
I do modeling work (specifically molecular dynamics and visualization) on
a Linux PC and it works great. I think the other posts to this group
address your question regarding benchmarks. Regarding hardware
compatibility, check http://www.redhat.com -- there you will find a
hardware compatibility list. Also, check www.linux.org for a list of
hardware vendors that provide systems with Linux pre-installed.
One caveat: as far as I know OpenGL acceleration is not available at the
hardware level, so it probably is not worth spending lots of $$ on a video
card which has OpenGL acceleration at the hardware level (I think there
are some commercial X vendors which claim accleration for OpenGL, but this
is done only on the software level). That being said, let me ease your
worries that hardware compatibility is a "big problem," as you will see
from the above RedHat web site. (for graphics under Linux using X, if you
want a recommendation, try a Matrox Millenium II for your graphics board
-- it is well supported and has a reputation for excellent performance; I
currently use an S3 Virge based card (Diamond 3D 3000, 4 MB memory) and it
works quite well).
Two excellent (and free) sources for molecular dynamics on Linux are:
Theoretical Biophysics Group
Check out their program VMD, which runs under RedHat 5.0 and uses a free
version of the OpenGL library called Mesa to provide graphics like you
would see on an SGI
Ponder Group at U. Washington, St. Louis
Fortran 77 code for molecular dynamics, mechanics, etc. (compiles
without trouble using g77 18.104.22.168)
Best of luck,
p.s.: Linux is also available on the Alpha architecture (by Digital, which
is now owned by Compaq, I believe). The Alpha has the fastest floating
point unit on the planet right now, but unfortunately, the math library
for Linux is not very good on Alpha's at this point, so the Alpha's full
potential cannot be realized (although someone is re-writing the library
and it should be ready for use on Alpha in a couple of weeks from what I
have heard...). So, if you are running Linux, it is probably better to
stick with Intel if you will be compiling your own code using gcc.
On Mon, 8 Jun 1998, Michel Seigneuret wrote:
> Dear Collegues,
> I am thinking of purchasing a 2 processors Pentium2 350 or 400Mhz PC
> to perform bioinformatics and molecular modelling studies under Linux.
> I would therefore be interested to hear from people already using
> similar machines for similar works.
> 1)Are the performances satisfactory for "standard" modeling tasks
> (e.g. sequence or structure database searching, molecular dynamics or
> conformational searching on peptides)? Comparison of performances with
> e.g. an SGI R10000 would be helpful.
> 2)It seems that a big problem with Linux is hardware compatibility.
> I would thus be interested if some people would advise me on specific
> hardware configurations that work (specially on high-end 3D OpenGL
> compatible graphic cards).
> Thank you in avance for your help.
> Michel Seigneuret
> Universite Paris 6, Lab. de Physicochimie Biomoleculaire et Cellulaire
> 4, place Jussieu, 75252 Paris cedex 05, France
> tel 33 1 44277545, fax 33 1 44277560
More information about the Molmodel