MPI parallel error

Tyler Gubb tag... at gmail.com
Wed Aug 5 13:57:56 UTC 2015


Good morning all,

I'm running into an error executing CP2K 2.6.1 with mpi.  The error is the 
following:

 ...

 mepos           0           0 numpe           1           1

 ncol_global          39

 nrow_global          39

 ncol_locals           0

 nrow_locals           0


 ****************************************************************************

 *** 18:46:53 ERRORL2 in cp_fm_struct:cp_fm_struct_create processor 0  :: 
***

 *** err=-300 sum of local cols not equal global colscp_fm_struct.F line  
***

 *** 270                                                                  
***

 ****************************************************************************



 ===== Routine Calling Stack ===== 


            5 mpools_rebuild_fm_pools

            4 qs_env_setup

            3 qs_init_subsys

            2 quickstep_create_force_env

            1 CP2K

 CP2K| sum of local cols not equal global colscp_fm_struct.F line 270

 CP2K| Abnormal program termination, stopped by process number 0

 application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0


 Further, although I'm not sure if it's relevant, CP2K is recognizing 
multiple MPI processes but not memory:

 GLOBAL| Total number of message passing processes                         
    2

 GLOBAL| Number of threads for this process                                
    1

 GLOBAL| This output is from process                                       
    0


 MEMORY| system memory details [Kb]

 MEMORY|                        rank 0           min           max       
average

 MEMORY| MemTotal                    0             0             0         
    0

 MEMORY| MemFree                     0             0             0         
    0

 MEMORY| Buffers                     0             0             0         
    0

 MEMORY| Cached                      0             0             0         
    0

 MEMORY| Slab                        0             0             0         
    0

 MEMORY| SReclaimable                0             0             0         
    0

 MEMORY| MemLikelyFree               0             0             0         
    0


Here is my arch file:


INTEL_MKL = /opt/intel/mkl

INTEL_FFT = /opt/intel/mkl/interfaces/fftw3xf

INTEL_INC = /opt/intel/include

INTEL_LIB = $(INTEL_MKL)/lib

FFTW_INC  = /opt/intel/mkl/include/fftw


CC       = mpicc -O2 -m64

CPP     = cpp

FC       = mpif90 -mkl -m64

LD       = mpif90 -mkl -m64

AR       = xiar -r



DFLAGS   = -D__INTEL -D__FFTMKL -D__FFTW3 -D__FFTSG -D__parallel -D__BLACS 
-D__SCALAPACK 

CPPFLAGS = -C -traditional $(DFLAGS) -I$(INTEL_INC) -I$(INTEL_FFT)

FCFLAGS  = $(DFLAGS) -I$(INTEL_INC) -I$(INTEL_FFT) -FR -O2 -xhost -fpp 
-free -g -m64 -heap-arrays 64 -funroll-loops

LDFLAGS  = $(FCFLAGS) -I$(INTEL_INC) -I$(INTEL_FFT) -I$(FFTW_INC) \

           -L/opt/intel/mkl/include/fftw -L$(FFTW_INC)

LIBS     = 
/Users/tylergubb/Build/scalapack-2.0.2/libscalapack-intelmpif90-3.1.4.a \

           $(INTEL_FFT)/libfftw3xf_gnu.a 



OBJECTS_ARCHITECTURE = machine_intel.o


I am using Intel ifort for the fortran compiler, MPICH 3.1.4 for the MPI 
wrapper, and gcc for the C compiler.  I had read in another thread here 
that -D__parallel needs to be lowercase, though both produced the same 
error.  I compiled MPICH with the following:


HYDRA build details:

    Version:                                 3.1.4

    Release Date:                            Fri Feb 20 15:02:56 CST 2015

    CC:                              gcc    

    CXX:                             g++    

    F77:                             ifort   

    F90:                             ifort -O3 -xHost  

    Configure options:                       '--disable-option-checking' 
'--prefix=/Users/tylerarthurgubb/mpich-3.1.4' '--enable-fast=O3' 'FC=ifort' 
'FCFLAGS=-O3 -xHost -O3' 'AR=xiar' '--cache-file=/dev/null' '--srcdir=.' 
'CC=gcc' 'CFLAGS= -O3' 'LDFLAGS= ' 'LIBS=-lpthread ' 'CPPFLAGS= 
-I/Users/tylerarthurgubb/Build/mpich-3.1.4/src/mpl/include 
-I/Users/tylerarthurgubb/Build/mpich-3.1.4/src/mpl/include 
-I/Users/tylerarthurgubb/Build/mpich-3.1.4/src/openpa/src 
-I/Users/tylerarthurgubb/Build/mpich-3.1.4/src/openpa/src -D_REENTRANT 
-I/Users/tylerarthurgubb/Build/mpich-3.1.4/src/mpi/romio/include'

    Process Manager:                         pmi

    Launchers available:                     ssh rsh fork slurm ll lsf sge 
manual persist

    Topology libraries available:            hwloc

    Resource management kernels available:   user slurm ll lsf sge pbs 
cobalt

    Checkpointing libraries available:


Oddly, the arch-file etc I'm currently using is identical to many previous 
builds I had successfully compiled and run, though I am on OSX 10.10.4.  I 
would greatly appreciate any assistance, or suggestions for hunting down 
the real culprit.


Thank you all so much and I look forward to using CP2K again!

Tyler Gubb.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150805/bae527d9/attachment.htm>


More information about the CP2K-user mailing list