[CPMD-list] cpmd-3.5.2 & ifc-6.0.1
Axel Kohlmeyer
axel.kohlmeyer at theochem.ruhr-uni-bochum.de
Mon Dec 30 23:15:27 CET 2002
On Thu, 05 Dec 2002 13:03:34 +0100 Kai Reichert wrote:
> Dear All,
>
>
> I'm not able to compile CPMD.
>
> As I own an Intel Fortran Compiler I issue the following command :
>
> Configure PC-IFC > Makefile
>
>
> I can then compile CPMD, but in the last linking stage I get dozens of
> undefined references.
>
>
> I tried to compile ATLAS on my own.
> But that did not work :-(
>
> In the newsgroups I found the hint, I should compile ATLAS also with
> ifc, instead of g77.
you could also try 'g77 -fno-second-underscore' atlas.
>
> Thats solve a lot of my undefined references, but it seems, that the
> liblapack library of ATLAS is insufficient ?
>
> I tried to compile the whole LAPACK distribution, but I failed. With my
> ifc, the compile process run in an infinite loop :-(
try compiling lapack with 'g77 -fno-second-underscore' and then
link first atlas then your lapack and then (but only if you need it)
-lg2c (you'll find the directory for the -L option with
'g77 -print-libgcc-file-name'. the speed for anything else than
DGEMM/DGEMV (which is in atlas) does not really matter for cpmd.
also compiling lapack with ifc but without optimization (-O0)
is something you can try.
as a last chance, you can ditch atlas completely an use the
intel math kernel library (should be roughly as fast as atlas,
depending on problem size).
NOTE: the order of the -L and -l arguments _does_ matter.
alternatively, i can offer you that i build a combined atlas/lapack
for your platform (as i described to you in a previous, private email),
but for that i need EXCLUSIVE access to a maschine of your architecture
and a little bit of time (the complete atlas tuning run takes some time
as do the manual library reassembly and the testing of the executables).
to the cpmd.org maintainer(s): since building of cpmd on linux seems to
be a recurring problem, how about putting a selection of precompiled
executables (statically linked) into the contrib section?
that would probably help a lot of people to get started and they don't
have to fight with compilers and input file syntax at the same time.
i can contribute binaries optimized for amd athlon xp (usable on thunderbird
and later) cpus for that.
cheers,
axel kohlmeyer.
>
> I would really appreciate if someone could give detailed instructions on
> how to compile CPMD and all its required libraries on a standard redhat
> 7.3 linux system.
>
>
> Regards,
>
> Kai Reichert
>
>
> -----------------------------------------------
> University of Dortmund
> Department of Chemical Engineering
> Campus Nord, Building G II, 5th floor
> Chair of Thermodynamics
>
> _______________________________________________
> CPMD-list mailing list
> CPMD-list at cpmd.org
> http://www.cpmd.org/mailman/listinfo/cpmd-list
>
--
=======================================================================
Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de
Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673
Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045
D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.
More information about the CPMD-list
mailing list