[MITgcm-support] Problem with MPI execution

Martin Losch Martin.Losch at awi.de
Thu May 22 09:25:17 EDT 2008


I don't think that the MPI within the MITgcm is the problem. Try to  
generate a non-mpi executable and see if the problem goes away.

The error message is clear: eedata cannot be opened. Is it in the  
correct directory? What about read permissions? Maybe the read  
permission are different for mpirun? This is the code where it happens:

>       OPEN(UNIT=eeDataUnit,FILE='eedata',STATUS='OLD',
>      &     err=1,IOSTAT=errIO)
>       IF ( errIO .GE. 0 ) GOTO 2
>     1 CONTINUE
>        WRITE(msgBuf,'(A)')
>      &  'S/R EESET_PARMS'
>        CALL PRINT_ERROR( msgBuf , 1)
>        WRITE(msgBuf,'(A)')
>      &  'Unable to open execution environment'
>        CALL PRINT_ERROR( msgBuf , 1)
>        WRITE(msgBuf,'(A)')
>      &  'parameter file "eedata"'
>        CALL PRINT_ERROR( msgBuf , 1)
>        CALL EEDATA_EXAMPLE
>        STOP 'ABNORMAL END: S/R EESET_PARMS'

no other ideas on this side of the Atlantic ...

Martin

On 22 May 2008, at 15:06, Dimitris Menemenlis wrote:

> Abbas, you probably have already done this but to be sure:
> did you use option "-mpi" when you generated the Makefile with  
> genmake2 ?
>
> Dimitris Menemenlis <menemenlis at sbcglobal.net>
> 5056 Oakwood Ave, La Canada, CA 91011-2450
> tel/fax: 818-790-6735;  cell: 818-625-6498
>
> On May 22, 2008, at 6:02 AM, Abbas Dorostkar wrote:
>
>> Hi Martin,
>> Thanks for quick reply.
>> It is strange because I have this file in the folder I run  
>> mitgcmuv. I
>> have attached it:
>>
>>> # Example "eedata" file
>>>> # Lines beginning "#" are comments
>>>> # nTx - No. threads per process in X
>>>> # nTy - No. threads per process in Y
>>>> &EEPARMS
>>>> nTx=1,
>>>> nTy=1,
>>>> usingMPI=.TRUE.
>>>> &
>>>> # Note: Some systems use & as the
>>>
>>> # namelist terminator. Other systems
>>>> # use a / character (as shown
>>
>> Do you have any other idea? I have been trying to fix this problem  
>> for a
>> while
>> Abbas
>>
>>
>>
>>
>>
>>
>>
>> -----Original Message-----
>> From: mitgcm-support-bounces at mitgcm.org
>> [mailto:mitgcm-support-bounces at mitgcm.org] On Behalf Of Martin Losch
>> Sent: May 22, 2008 3:42 AM
>> To: mitgcm-support at mitgcm.org
>> Subject: Re: [MITgcm-support] Problem with MPI execution
>>
>> Abbas,
>>
>> you are missing the file "eedata", as the error messgae clearly tells
>> you. I won't tell you how often I made this mistake!!!
>>
>> Martin
>>
>> On 21 May 2008, at 22:46, Abbas Dorostkar wrote:
>>
>>> Dear all,
>>>
>>>
>>>
>>> I have been trying to run the exp1 with MPI execution on my desktop
>>> (ia32_linux equipped with one dual 1.5 processor) before running my
>>> own model on a node with 72  dual-core processors and 570 GB RAM. I
>>> haven't got any error during compiling. However, when I run the
>>> mitgcmuv with command "mpirun -np 2 ./mitgcmuv", I get following
>>> error:
>>>
>>>
>>>
>>> STOP ABNOSTOP ABNORMAL END: S/R EESET_PARMS
>>> RMAL rank 0 in job 1  LEO.CiVil.QueensU.Ca_55941   caused
>>> collective abort of all ranks
>>>  exit status of rank 0: return code 0
>>>
>>> (PID.TID 0000.0001) *** ERROR *** S/R EESET_PARMS
>>> (PID.TID 0000.0001) *** ERROR *** Unable to open execution  
>>> environment
>>> (PID.TID 0000.0001) *** ERROR *** parameter file "eedata"
>>>
>>>
>>>
>>> I run successfully some simple "Hello World"-type MPI programs,
>>> showing my MPI (MPICH2) install is working correctly. I don't know
>>> what I am missing?? Could someone provide some solution?  Here I
>>> have attached size.h, eedata and optfile :
>>>
>>>
>>>      PARAMETER (
>>>
>>>     &           sNx =  60,
>>>
>>>     &           sNy =  60,
>>>
>>>     &           OLx =   2,
>>>
>>>     &           OLy =   2,
>>>
>>>     &           nSx =   1,
>>>
>>>     &           nSy =   1,
>>>
>>>     &           nPx =   2,
>>>
>>>     &           nPy =   1,
>>>
>>>     &           Nx  = sNx*nSx*nPx,
>>>
>>>     &           Ny  = sNy*nSy*nPy,
>>>
>>>     &           Nr  =   4)---------------------------------------
>>> # Example "eedata" file
>>>
>>> # Lines beginning "#" are comments
>>>
>>> # nTx - No. threads per process in X
>>>
>>> # nTy - No. threads per process in Y
>>>
>>> &EEPARMS
>>>
>>> nTx=1,
>>>
>>> nTy=1,
>>>
>>> usingMPI=.TRUE.
>>>
>>> &
>>>
>>> # Note: Some systems use & as the
>>>
>>> # namelist terminator. Other systems
>>>
>>> # use a / character (as shown
>>> here).---------------------------------------
>>> #!/bin/bash
>>>
>>> #
>>>
>>> #  $Header: /u/gcmpack/MITgcm/tools/build_options/linux_ia32_g77
>>> +mpi_cg01,v 1.6 2006/03/24 22:34:43 edhill Exp $
>>>
>>> #  $Name:  $
>>>
>>>
>>>
>>> FC='/usr/local/bin/mpif77'
>>>
>>> CC='/usr/local/bin/mpicc'
>>>
>>> DEFINES='-DALLOW_USE_MPI -DALWAYS_USE_MPI -D_BYTESWAPIO -
>>> DWORDLENGTH=4'
>>>
>>> INCLUDEDIRS='/usr/local/include'
>>>
>>> INCLUDES='-I/usr/local/include'
>>>
>>> CPP='/lib/cpp  -traditional -P'
>>>
>>> NOOPTFLAGS='-O0'
>>>
>>>
>>>
>>> if test "x$IEEE" = x ; then
>>>
>>>    #  No need for IEEE-754
>>>
>>>    FFLAGS='-Wimplicit -Wunused -Wuninitialized'
>>>
>>>    FOPTIM='-O3 -malign-double -funroll-loops'
>>>
>>> else
>>>
>>>    #  Try to follow IEEE-754
>>>
>>>    FFLAGS='-Wimplicit -Wunused -ffloat-store'
>>>
>>>    FOPTIM='-O0 -malign-double'
>>>
>>> fi
>>>
>>>
>>>
>>> # netcdf
>>>
>>> #LIBS="-lnetcdf"
>>> Your help would be much appreciated.
>>>
>>> Thanks a lot
>>>
>>> Abbas
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> MITgcm-support mailing list
>>> MITgcm-support at mitgcm.org
>>> http://mitgcm.org/mailman/listinfo/mitgcm-support
>>
>>
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-support
>>
>> _______________________________________________
>> MITgcm-support mailing list
>> MITgcm-support at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-support
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support




More information about the MITgcm-support mailing list