[CP2K:4499] building hybrid MPI+OpenMP

[Iain Bethune]

You also need to ensure the -openmp flag gets passed to the linker, that doesn't seem to be the case in your arch file (although it was in the earlier one).  I'm afraid I can't think of anything else obviously wrong.  We used an arch file (below) recently on another intel system and it worked OK.  The major differences are we static linked everything, and used the libfftw3_omp rather than _threads.  You could try and see if that helps any?

Cheers

- Iain

CC       = cc
CPP      =
FC       = mpiifort -openmp
LD       = mpiifort -openmp
AR       = ar -r
 
FFTW_DIR = /users/fiona/fftw3_intel_threaded
LIBXC_DIR = /users/fiona/lib/libxc-intel
LIBINT_DIR = /users/fiona/lib/libint-intel
LIBSMM_DIR = /users/fiona/lib/cp2klibs-intel
LIBGRID_DIR = /users/fiona/lib/cp2klibs-intel
 
CPPFLAGS =
DFLAGS   = -D__INTEL -D__FFTSG -D__FFTW3 -D__LIBINT -D__LIBXC2 -D__parallel \
       -D__BLACS -D__SCALAPACK -D__HAS_smm_dnn -D__HAS_LIBGRID
CFLAGS   = $(DFLAGS)
# Version 13.1.0 of compiler
MKLROOT = /apps/dommic/intel/composer_xe_2013.2.146/mkl/
FCFLAGS  = $(DFLAGS) -O2 -g -traceback -fpp -free \
       -I$(FFTW_DIR)/include
LDFLAGS  = $(FCFLAGS) -static-intel
LIBS     = $(FFTW_DIR)/lib/libfftw3.a $(FFTW_DIR)/lib/libfftw3_omp.a \
       $(MKLROOT)/lib/intel64/libmkl_scalapack_lp64.a \
       -Wl,--start-group  $(MKLROOT)/lib/intel64/libmkl_intel_lp64.a \
       $(MKLROOT)/lib/intel64/libmkl_sequential.a \
       $(MKLROOT)/lib/intel64/libmkl_core.a \
       $(MKLROOT)/lib/intel64/libmkl_blacs_intelmpi_lp64.a -Wl,--end-group \
       -lpthread -lm \
       $(LIBINT_DIR)/lib/libderiv.a $(LIBINT_DIR)/lib/libint.a -lstdc++ \
       -L$(LIBXC_DIR)/lib -lxc \
       $(LIBSMM_DIR)/libsmm_dnn.a $(LIBGRID_DIR)/libgrid.a
 
OBJECTS_ARCHITECTURE = machine_intel.o
 
--

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, Mayfield Road, Edinburgh, EH9 3JZ








On 18 Jul 2013, at 14:45, Steve Schmerler wrote:

> On Jul 15 09:45 +0100, Iain Bethune wrote:
>> Hi Steve,
> 
> [Sorry for the delayed response.]
> 
>> Try rebuilding without using the -heap-arrays 64 flag, as I have found problems with this in combination with OpenMP before (but not yet got to the root cause).  Let me know if it helps or not!
> 
> No, unfortunately not. It only prevents the output to hang in the one
> case mentioned and instead shows the same MKL error, just like all other
> cases. Even with a very reduced arch file (no FFTW, no fancy compiler
> flags), I get the same errors:
> 
> ------------------------------------------------------------------
> CC       = cc
> CPP      = 
> FC       = mpiifort
> LD       = mpiifort
> AR       = ar -r
> DFLAGS   = -D__INTEL -D__parallel -D__BLACS -D__SCALAPACK -D__FFTSG
> CPPFLAGS =
> FCFLAGS  = $(DFLAGS) -O2 -free -fpp -openmp
> FCFLAGS2 = $(DFLAGS)  -O1 -free
> LIBS     = -lmkl_blacs_intelmpi_lp64 -lmkl_scalapack_lp64 \
>           -lmkl_intel_lp64 -lmkl_core -lmkl_sequential -liomp5 
> ------------------------------------------------------------------
> 
> I'm out of ideas right now. Luckily, the MPI-only version is fast enough
> for our systems at the moment on the number of cores where it scales. 
> Thanks.
> 
> best,
> Steve
> 
> -- 
> Steve Schmerler
> Institut für Theoretische Physik
> TU Freiberg, Germany
> 
> -- 
> 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/groups/opt_out.
> 
> 
> 


-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.