Summary on PC Cluster Software
Thanks to all that supplied information on software for parallel processing
PC clusters. Several people show interest in a summary so here it is.
I was very pleased with the number of responses.
It appears that there are several commercial products.
GAUSSIAN with the GaussView GUI (not required)
HTTP://WWW.gaussian.com
Schrodinger has TITAN (GUI) and Jaguar. TITAN is a SPARTAN GUI.
HTTP://WWW.SCHRODINGER.COM
Non-commercial products include
GAMESS
CHARM
Peter Pulay's Parallel Quantum Solutions
www.pqs-chem.com
GROMACS parallel MD package
http://md.chem.rug.nl/~gmx
NWChem
http://www.emsl.pnl.gov:2080/docs/nwchem/nwchem.html
There are many useful links to pages discussing Beowulf clusters. Again,
Thanks to all,
Jim
=================
http://www.beowulf-underground.org
http://sal.kachinatech.com/
These are two great sites that list a lot of applications that can run
on a beowulf cluster. You might also want to search the boewulf/extreme
linux mailing list archives as many announcements of applications have
been made over both.
--
Brian D. Haymore
University of Utah
Center for High Performance Computing
155 South 1452 East RM 405
Salt Lake City, Ut 84112-0190
Email: brian()at()chpc.utah.edu - Phone: (801) 585-1755 - Fax: (801) 585-5366
===============================
Hi Jim,
I started a web site a couple of months back ( and haven't worked on it
since )
dealing with "Beowulf clusters for chemists" you may find some useful
information there.
http://www.chem.arizona.edu/theochem/beowulf
check the "what chemists are doing with their clusters" link
Hope this helps
-Don
--
++++++++++++++++++++++++++++++++++++++++++++++++++
Dr. Donald B. Kinghorn
Theoretical and Computational Chemistry Group
University of Arizona
Tucson AZ 85721
kinghorn()at()u.arizona.edu
(520)621-6761
++++++++++++++++++++++++++++++++++++++++++++++++++
=======================================
Hi Jim,
for Quantum Chemistry software running under Linux on our Cluster,
take a look at http://www.chm.tu-dresden.de/edv/soft-lin.html
(sorry, in german; but in most cases Links to Homepages are included).
For real parallel computation, GAUSSIAN and GAMESS should work
(but with different parallelization tools; LINDA vs. MPI or TCGMSG).
Instead of this, we are using all the programs sequentially,
managed by an batch system in our Linux cluster.
Matthias
--
Dr. Matthias Mann
Fachrichtung Chemie der TU Dresden
D-01069 Dresden, Mommsenstr. 13
Tel.: +49 (351) 463-4286 / Fax: -7030
Email: Matthias.Mann()at()chemie.tu-dresden.de
=========================================
Dr. Macmillan,
Gaussian 98 is supported on this type of cluster. We utilize the
LINDA software from Scientific Computing Associates(SCA) to simulate shared
memory and interprocess communication. The combination of G98 + Linda
scales quite well for HF, DFT, CI-singles and even MP2 energies and
gradients. We published a report about G94 on Cray T3E systems which
demonstrates the scalability, of course the actual per CPU time varies
by platform, in J. Comp. Chem. vol. 19, pp 1053 (1998) which will give
you a good idea of what you can expect.
The LINDA software is normally licensed on a per-process basis but
we have made an arrangement with SCA which permits use of G98+Linda
on any number of processors so it is a cost effective combination for
this type of Parallel Cluster.
If you would like a proper quote please send a request to
fo()at()gaussian.com. If you have more specific technical questions let me know
at help()at()gaussian.com.
=======================================================
Schrodinger markets JAGUAR for Linux. Wavefunction markets TITAN, sort of
Spartan with a hook to Jaguar,.. for NT, however. JAGUAR V4/Linux will be
out at end of month. There is a Gui for it on the way.
John McKelvey
=======================================
Well for semi-empirical and ab initio modeling there's GAMESS, Gaussian,
and Jaguar that all run under linux. You might check out the site
http://chpc06.ch.unito.it/chem_linux.html, this gives a good
listing of
software running on linux.
hope this helps,
-david
David Reichert, Ph.D.
Washington University School of Medicine
510 S. Kingshighway, Campus Box 8225
St Louis, MO 63110
e-mail: reichertd()at()mir.wustl.edu
voice: (314) 362-8461
fax: (314) 362-9940
===========================================================
Hello, James:
For a good parallelized quantum package that is already optimized for PVM
under Linux, I suggest the code from Peter Pulay's Parallel Quantum
Solutions (www.pqs-chem.com).
Best wishes,
Brian Teppen
=====================================
Dear Dr. Macmillan:
Since you are working in the Dept. of Chemistry I thinks at least you can
use
CHARMM and GAMESS with the Beowulf cluster!
Please check the link listed below,
http://www.lobos.nih.gov/apps_frame.html
Joe
HCHuang()at()uh.edu
==============================
GROMACS parallel MD package
http://md.chem.rug.nl/~gmx
Groeten, David.
________________________________________________________________________
Dr. David van der Spoel Biomedical center, Dept. of Biochemistry
s-mail: Husargatan 3, Box 576, 75123 Uppsala, Sweden
e-mail: spoel()at()xray.bmc.uu.se www: http://zorn.bmc.uu.se/~spoel
phone: 46 18 471 4205 fax: 46 18 511 755
==================================================
If the cluster is reasonably big, try NWChem. Single-node performance is
not impressive, but it's very efficiently parallelized. Runs on Linux,
and is free (subject to a user agreement). Have a look at
http://www.emsl.pnl.gov:2080/docs/nwchem/nwchem.html
Don't know if parallel Gaussian is available for Linux. Check their web
page (www.gaussian.com).
Btw., I think it IS of general interest, so you might summarize some of
the replies you get.
Regards,
Herbert
==============================================
***********************************************
James G. Macmillan
1419 Laurel Cir.
Cedar Falls, IA 50613
E-Mail macmillan()at()uni.edu
Web http:\\www.cns.uni.edu\~macmilla
*********************************************
"Making good wine is a science, making fine wine is an art" - Robert
Mondavi