[CP2K:6522] issues in compiling cp2k on IBM-BG.Q

Iain Bethune ibet... at epcc.ed.ac.uk
Wed May 20 08:03:34 UTC 2015
Previous message (by thread): issues in compiling cp2k on IBM-BG.Q
Next message (by thread): issues in compiling cp2k on IBM-BG.Q
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Hi Luca,

I am currently working on installing CP2K on the JUQUEEN system at FZJ.  I can make a couple of comments:

1) The ‘dlopen’ warning can be safely ignored.
2) I also observe the crash at the start of SCF (it’s in the FFT code and occurs even when running on a single core with everything compiled at -O0).

3) The compiler error you showed I have not seen.  I am using the following version:

xlf -qversion
IBM XL Fortran for Linux, V14.1 (5765-J05, 5725-C75)
Version: 14.01.0000.0007

And compiling all source files with the following flags:

-O3 -qhot -g -qsmp=omp -qstrict -qarch=qp -qtune=qp -qmaxmem=-1

The compile goes fine, but I am still debugging the issue above.

I will let you know when I get a working build.

Cheers

- Iain

--

Iain Bethune
Project Manager, EPCC

Email: ibet... at epcc.ed.ac.uk
Twitter: @IainBethune
Web: http://www2.epcc.ed.ac.uk/~ibethune
Tel/Fax: +44 (0)131 650 5201/6555
Mob: +44 (0)7598317015
Addr: 2404 JCMB, The King's Buildings, Peter Guthrie Tait Road, Edinburgh, EH9 3FD

> On 20 May 2015, at 07:07, Luca <luca.d... at gmail.com> wrote:
> 
> Dear Developers
>  
> I am trying to install the release 2.6.1 of CP2K on a IBM/BG-Q system. I did several attempts but without any success. I used the arch file prepared by the system manager to install the old (2.4) version of cp2k (please look at the end of this mail).  Using that file I was able to compile the code with just a warning :
>  
> ###############################
>  /opt/ibmcmp/xlf/bg/14.1/bglib64/libxlopt.a(cblas.64.o):(.text+0xe4): warning: Using 'dlopen' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking
> ###############################
>  
>  
> Notwithstanding  I got the executable, unfortunately it does not run properly in the sense that it always crashes at the beginning of the scf procedure:
>  
>  
> ###############################
> SCF WAVEFUNCTION OPTIMIZATION
>  
>   Step     Update method      Time    Convergence         Total energy    Change
>   ------------------------------------------------------------------------------
> 2015-05-16 23:51:40.820 (WARN ) [0x400013291e0] :3119263:ibm.runjob.client.Job: terminated by signal 11
> 2015-05-16 23:51:40.820 (WARN ) [0x400013291e0] :3119263:ibm.runjob.client.Job: abnormal termination by signal 11 from rank 0
> #############################
>  
>  
> I did a second attempt modifying the arch file IBM-BGQ-MPI.psmp that is present in the arch dir of the release 2.6.1. In that case the compilation procedure finishes with the following lines:
>  
> ##################################
> ** cp_parser_ilist_methods   === End of Compilation 1 ===
> ** ai_coulomb   === End of Compilation 1 ===
> /opt/ibmcmp/xlf/bg/14.1/bin/.orig/bgxlf90_r: 1501-224 (S) fatal error in /opt/ibmcmp/xlf/bg/14.1/exe/ipa: signal 9 received
> make[3]: *** [grid_fast.o] Error 251
> make[3]: *** Waiting for unfinished jobs....
> 1501-510  Compilation successful for file cp_parser_ilist_methods.f90.
> 1501-510  Compilation successful for file memory_utilities.f90.
> 1501-510  Compilation successful for file tmc_move_types.f90.
> 1501-510  Compilation successful for file ai_onecenter.f90.
> 1501-510  Compilation successful for file optimize_basis_types.f90.
> 1501-510  Compilation successful for file ai_coulomb.f90.
> 1501-510  Compilation successful for file hfx_compression_core_methods.f90.
> make[2]: *** [all] Error 2
> make[1]: *** [psmp] Error 2
> make: *** [all] Error 2
> ################################
>  
> Could you please help me with these issues?
>  
> Thank you in advance and best regards
>  
> Luca
>  
>  
>  
>  
>  
>  
> ARCH file first attempt.
> #############
> LAPACK_LIB          =/cineca/prod/libraries/lapack/3.4.1/bgq-xl--1.0/lib
> SCALAPACK_LIB   =/cineca/prod/libraries/scalapack/2.0.2/bgq-xl--1.0/lib
> LIBINT_INCL     =/cineca/prod/libraries/libint/2.0/bgq-xl--1.0/include
> LIBINT_LIB      =/cineca/prod/libraries/libint/2.0/bgq-xl--1.0/lib
> FFT_LIB         =/cineca/prod/libraries/fftw/3.3.3/bgq-xl--1.0/lib
> FFT_INC         =/cineca/prod/libraries/fftw/3.3.3/bgq-xl--1.0/include
> ESSL_LIB        =/opt/ibmmath/essl/5.1/lib64
>  
> CC       = bgxlc_r
> CPP      = cpp -traditional -E  -I$(FFT_INC) -I$(LIBINT_INCL) -I$(LIBINT_INCL)/libint -I$(LIBINT_INCL)/libderiv
>  
> FC       = mpixlf95_r
> LD       = mpixlf95_r
> AR       = ar -r
>  
> DFLAGS   =-D__AIX -D__ESSL -D__parallel -D__FFTW3 -D__BLACS -D__SCALAPACK -D__LIBINT -D__LIBXC2
>  
> CPPFLAGS = -C $(DFLAGS) -P
>  
> FCFLAGS  = -qnosave -O3 -qfullpath -qstrict -qarch=qp -qtune=qp -I$(FFT_INC) -tchbI -I$(LIBINT_INCL) -I$(LIBINT_INCL)/libint -I$(LIBINT_INCL)/libderiv
>  
> FCFLAGS_NOOMP  = -qnosave -O3 -qfullpath -qstrict -qarch=qp  -qtune=qp  -I$(FFT_INC) -tchbI
>  
> LDFLAGS  = $(FCFLAGS) -O3 -qstrict -qfullpath  -qarch=qp -qtune=qp  -qnosave -I$(FFT_INC) -tchbI -I$(LIBINT_INCL) -I$(LIBINT_INCL)/libint -I$(LIBINT_INCL)/libderiv  -L$(ESSL_LIB) -lesslsmpbg
>  
> LIBS      = -L$(SCALAPACK_LIB) -lscalapack  -L$(LAPACK_LIB) -llapack -L$(ESSL_LIB) -lesslsmpbg -lxlsmp -L$(FFT_LIB) -lfftw3 -lfftw3_omp -tchbI -L$(LIBINT_LIB) -lderiv -lint -L$(LIBXC_LIB) -lxc
>  
> OBJECTS_ARCHITECTURE = machine_aix.o
>  
>  
>  
>  
> ARCH FILE second attempt:
> #######################
> CC       = mpixlc
> CPP      = cpp -E
> FC       = mpixlf95_r
> LD       = mpixlf95_r
> AR       = ar -r
>  
> DFLAGS   = -D__AIX -D__FFTW3 -D__FFTSG -D__parallel -D__BLACS -D__SCALAPACK
> CPPFLAGS = -C $(DFLAGS) -P -traditional -I/cineca/prod/libraries/fftw/3.3.3/bgq-xl--1.0/include
> FCFLAGS  = -O3 -qhot -g -qsmp=omp -qstrict -qarch=qp -qtune=qp -qmaxmem=-1 \
>            -I/opt/ibmmath/essl/5.1/include \
>            -I/cineca/prod/libraries/fftw/3.3.3/bgq-xl--1.0/include
>  
> LDFLAGS  = $(FCFLAGS) -Wl,--allow-multiple-definition
>  
> LIBS    = -L/cineca/prod/libraries/fftw/3.3.3/bgq-xl--1.0/lib -lfftw3_mpi -lfftw3_omp -lfftw3 -lfftw3f_mpi -lfftw3f_omp -lfftw3f \
>           /cineca/prod/libraries/scalapack/2.0.2/bgq-xl--1.0/lib/libscalapack.a /cineca/prod/libraries/blacs/1.1/bgq-gnu--4.4.6/lib/libblacsF77init.a \
>           /cineca/prod/libraries/blacs/1.1/bgq-gnu--4.4.6/lib/libblacsCinit.a /cineca/prod/libraries/blacs/1.1/bgq-gnu--4.4.6/lib/libblacs.a \
>           /shared/project/data/userexternal /CP2K/lapack-3.5.0/liblapack.a \
>           /opt/ibmmath/essl/5.1/lib64/libesslbg.a \
>           /opt/ibmcmp/xlmass/bg/7.3/bglib64/libmass.a
>  
> OBJECTS_ARCHITECTURE = machine_aix.o
> 
> -- 
> You received this message because you are subscribed to the Google Groups "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns... at googlegroups.com.
> To post to this group, send email to cp... at googlegroups.com.
> Visit this group at http://groups.google.com/group/cp2k.
> For more options, visit https://groups.google.com/d/optout.


-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
Previous message (by thread): issues in compiling cp2k on IBM-BG.Q
Next message (by thread): issues in compiling cp2k on IBM-BG.Q
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the CP2K-user mailing list