[CPMD-list] CPMD Installation on Mac OSX

[Axel Kohlmeyer]

On Mon, 17 Dec 2007, Chang Y. Won  wrote:

CW> Dear Axel Kohlmeyer and  CPMD users, 


dear chang won,

CW> 
CW> Here are further information of my system. 
CW> 
CW> >did you use the MACOSX-XLF-G5-MPI configuration?
CW> 
CW> Yes, I used the MACOSX-XLF-G5-MPI configuration. ( MACOSX-XLF-G5 works, but MACOSX-XLF-G5-MPI doesn't)
CW> I tried both mpif77 and mpif90.
CW> First, I tried MACOSX-XLF-G5-MPI as it is. The error is /usr/bin/ld: Undefined symbols:
CW> 
CW> Secondly, I changed mpif77 to mpif90.  With mpif90, the error I got is:

no. CPMD assumes fortran 77 semantics with fixed format. most fortran
90 compilers default to free format, so you'll have to add a flag to
change this. furthermore, which fortran compiler was used to compile
the the mpi library? if it is not the XLF compiler, you may be in 
trouble or at least will have to change the linker flags...

CW> 
CW> "./cpmd.f", line 222.7: 1515-019 (S) Syntax is incorrect
CW> similar error for "./prop.inc," "./soft.inc", & "./system.h"
CW> 
CW> >what are the compiler/linker flags that you are using?
CW> 
CW> My system can provide:
CW> IBM VAX and XLF, 
CW> GCC 3.3
CW> Absoft Forstran, 
CW> G95. 
CW>  
CW> Based on MACOSX-XLF-G5-MPI, I guess I used IBM XLF. 
CW> 
CW> 
CW> 
CW> >can you compile and run _any_ fortran based MPI code?
CW> 
CW> I checked with my system compiler, mpif90. It works fine. 
CW> 
CW> 
CW> My officemate installed LAMMPS (fortran based MPI) on the same machine. 

hmmm.... LAMMPS (unless you use an old and obsolete version)
is a C++ code, so i have my doubts whether this is a useful test.

cheers,
   axel.


CW> Thank you very much again. 
CW> 
CW> Regards, 
CW> 
CW> Chang Won 
CW> 
CW> ---- Original message ----
CW> >Date: Mon, 17 Dec 2007 03:53:06 -0500 (EST)
CW> >From: Axel Kohlmeyer <akohlmey at cmm.chem.upenn.edu>  
CW> >Subject: Re: [CPMD-list] CPMD Installation on Mac OSX  
CW> >To: "Chang Y. Won " <won2 at uiuc.edu>
CW> >Cc: CPMD Mailing List <CPMD-list at cpmd.org>
CW> >
CW> >On Sun, 16 Dec 2007, Chang Y. Won  wrote:
CW> >
CW> >CW> Dear CPMD, 
CW> >
CW> >dear chang won,
CW> >
CW> >
CW> >CW> I'm trying to compile CPMD on Mac Os X Ver. 10.3, MPICH version of 
CW> >CW> IBM-XLF and latest version of CPMD. On the same machine, the serial 
CW> >CW> executable CPMD works fine. I read LAM/MPI is suggested to use for 
CW> >CW> linux, but I can't use any other any communications packages, except 
CW> >CW> MPICH or Charm++ for some reasons.
CW> >
CW> >CPMD uses only a small subset of the MPI-1 standard 
CW> >and thus should work with _any_ MPI package. 
CW> >
CW> >LAM/MPI is deprecated by their developers. if i would 
CW> >recommend any MPI package now, it would be OpenMPI.
CW> > 
CW> >CW> Below is what I got:
CW> >CW> 
CW> >CW> Right above the "/usr/bin/ld: Undefined symbols" I had warning and error. I attch those too. 
CW> >CW> ---------------------------------
CW> >CW> /usr/bin/ld: warning suggest use of -bind_at_load, as lazy binding may result in errors or different symbols being used
CW> >CW> symbol _system used from dynamic library /usr/lib/libm.dylib(system.So) not from earlier dynamic library /turing/software/lib/libgm.0.dylib(gm_fork_system.lo) 
CW> >CW> symbol _fork used from dynamic library /usr/lib/libm.dylib(fork.So) not from earlier dynamic library /turing/software/lib/libgm.0.dylib(gm_fork_system.lo)
CW> >CW> /usr/bin/ld: Undefined symbols:
CW> >CW> _hostnm__
CW> >CW> _getcwd__
CW> >CW> _getpid__
CW> >CW> _getuid__
CW> >CW> _malloc_
CW> >CW> _free_
CW> >CW> _fdate__
CW> >CW> _flush__
CW> >CW> _getlog__
CW> >CW> _sleep__
CW> >CW> make: *** [cpmd.x] Error 1
CW> >CW> ---------------------
CW> >CW> 
CW> >
CW> >
CW> >CW> If anyone has any idea how to compile CPMD on Mac OSX, please give 
CW> >CW> me your feedback.
CW> >
CW> >you have to provide more information, before people can
CW> >help you. several people on this list here (and i) have
CW> >successfully compiled and run cpmd on PPC based mac os
CW> >machines, but each machine may need some little tweaks,
CW> >and both apple and IBM keep changing how you have to
CW> >use their compilers and libraries. please check your
CW> >documentation for any special changes needed. 
CW> >
CW> >did you use the MACOSX-XLF-G5-MPI configuration?
CW> >
CW> >what are the compiler/linker flags that you are using?
CW> >
CW> >can you compile and run _any_ fortran based MPI code?
CW> >
CW> >cheers,
CW> >   axel.
CW> >
CW> >CW> 
CW> >CW> Thank you very much. I'd really appreciate it.
CW> >CW> 
CW> >CW> Regards, 
CW> >CW> 
CW> >CW> Chang Won 
CW> >CW> 
CW> >CW> Chang Yeon Won
CW> >CW> -------------------------------------------------------------------------
CW> >CW> Grad. Res. Asst. (Ph.D Candidate)
CW> >CW> Computational MEMS/Nano Lab
CW> >CW> 3213 Beckman Institute 
CW> >CW> University of Illinois at Urbana-Champaign    
CW> >CW> 405 N. Mathews Avenue
CW> >CW> Urbana, IL 61801                            Office (217)244-1964
CW> >CW> -------------------------------------------------------------------------
CW> >CW> _______________________________________________
CW> >CW> CPMD-list mailing list
CW> >CW> CPMD-list at cpmd.org
CW> >CW> http://cpmd.org/mailman/listinfo/cpmd-list
CW> >CW> 
CW> >
CW> >-- 
CW> >=======================================================================
CW> >Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
CW> >   Center for Molecular Modeling   --   University of Pennsylvania
CW> >Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
CW> >tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
CW> >=======================================================================
CW> >If you make something idiot-proof, the universe creates a better idiot.
CW> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.