[CPMD-list] Compling CPMD with IFC and MPI

[Axel Kohlmeyer]

On Tue, 29 Jul 2003 20:22:58 -0700 (PDT)  Keta Jones wrote:

> Dear All,
> 
> Actually I want a parallel version of cpmd.
> When I run ./Configure it make a Makefile, but 
> it does not work to give a cpmd executive.
> I realize that my problem is in compilers.
> I am using ifc for cpmd, but my all libraries
> are compiled with gnu.So is there any source
> to get the optimized libraries, so that work
> will be easier instead of installing all the things.

dear keta,

if you get the proper libraries in rpm format,
the installation effort is usually nothing more 
than opening a root shell and using the rpm command
to install the rpm archives (or using one of the 
GUI rpm frontends). 

i have just gotten access to a very similarly set up
redhat 9 installation. i plan to prepare suitable
rpms for that installation next week. so if you are 
willing to wait a bit, i can put the rpms up for 
download as soon as i've created and tested them.

compiling an mpi library from sources is no rocket
science (try <http://www.lam-mpi.org/> or 
<http://www-unix.mcs.anl.gov/mpi/mpich/>). you just 
have to look at the compiling instructions to find out
how to override the default fortran compiler with ifc.

the intel MKL (providing pentium-III and pentium-IV optimized
blas and lapack) is available in rpm format. if you have
an amd athlon cpu, you can use the pentium-iii version or
you can try my athlon-xp optimized combined atlas/blas/lapack 
library at
<http://www.theochem.ruhr-uni-bochum.de/~axel.kohlmeyer/libatlas.a>

alternatively, you can try searching the various
rpm archives on the net (e.g. <http://freshrpms.net/
or <http://rpmfind.net/>) to find what you need.
chances are slim though.

> or how can I overcome these things, so that  I will
> get all the proper libraries.
> 

short of using somebody else's (statically linked) cpmd binaries,
(which is only likely to work on different linux installations 
for the serial version), i don't see any alternatives.

take care,
	axel.


> regards,
> Keta
> 


> 
> --- Axel Kohlmeyer
> <axel.kohlmeyer at theochem.ruhr-uni-bochum.de> wrote:
> > On Sat, 26 Jul 2003 23:42:35 -0700 (PDT)  Keta Jones
> > wrote:
> > 
> > > 
> > > I am using following configurations :
> > > 
> > > Operating system : Redhat 9.0
> > > CPMD : cpmdv3.5_3
> > > Lapack : lapack-3.0-20
> > > blas	: blas-3.0-20
> > > IFC	: ifc 7.1
> > > 
> > > For mpi whatever supplied with RedHat 9.0.
> > > 
> > > I can not get a proper Makefile.Should I go
> > > for separate MPI cofiguration ?
> > 
> > hi keta,
> > 
> > what do you mean by that? are you not able to
> > successfully compile cpmd or are you not able
> > to get a makefile at all?
> > 
> > if i assume the former - which seems likely - you'll
> > 
> > have gotten many 'undefined reference' errors on the
> > 
> > final linking stage. this is likely from the specs
> > you 
> > have given. all of those libraries supplied by
> > redhat -
> > lapack, blas, and mpi - were compiled with the GNU 
> > fortran compiler (g77). unfortunately you cannot mix
> > 
> > code from this compiler with the intel fortran
> > compiler 
> > without some trickery and using specific flags. so
> > you 
> > have to either compile everything by yourself using 
> > the intel compiler (be very careful with
> > lapack/blas,
> > some routines have to be compiled _without_ any 
> > optimization) or get proper binaries/rpms from some
> > o
> > ther source. as for blas/lapack using either the
> > atlas 
> > library <http://math-atlas.sourceforge.net/>) or the
> > 
> > intel math kernel library (mkl)
> > <http://www.intel.com/software/products/perflib/>
> > is highly recommended for performance reasons (more 
> > than twofold speedup in some cases).
> > 
> > hope this helps,
> > 	axel 
> > 	
> > > 
> > > Please help me.
> > > 
> > > regards,
> > > Keta
> > > 
> > > =====
> > > Keta
> > >
> >
> *****************************************************************************
> > *********
> > > 
> > > __________________________________
> > > Do you Yahoo!?
> > > Yahoo! SiteBuilder - Free, easy-to-use web site
> > design software
> > > http://sitebuilder.yahoo.com
> > > _______________________________________________
> > > 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.
> > _______________________________________________
> > CPMD-list mailing list
> > CPMD-list at cpmd.org
> > http://www.cpmd.org/mailman/listinfo/cpmd-list
> 
> 
> __________________________________
> Do you Yahoo!?
> Yahoo! SiteBuilder - Free, easy-to-use web site design software
> http://sitebuilder.yahoo.com
> _______________________________________________
> 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.