Understanding MPI_Gather

[emilio2510]

Hi there,

I’m relatively new to Fortran and to this community. I’m trying to write some code that performs some calculations on a multidimensional array using MPI. And for the life of me I can’t get MPI_Gather to work, and I’m sure it’s something very obvious that I’m missing here.

My code is:

program test

    USE mpi
    USE netcdf

    IMPLICIT NONE

    INTEGER,ALLOCATABLE,DIMENSION(:,:,:) :: array, global_array
    INTEGER :: j_start, j_end, i_start, i_end, J, I, t

    ! MPI
    INTEGER :: mpirank, mpisize, mpierr

    ! Initialize MPI
    call MPI_INIT(mpierr)
    call MPI_COMM_SIZE(MPI_COMM_WORLD, mpisize, mpierr)
    call MPI_COMM_RANK(MPI_COMM_WORLD, mpirank, mpierr)

    ALLOCATE(array(4,4,2))
    ALLOCATE(global_array(4,4,2))
    
    j_start = 1
    j_end   = 4
    i_start = 1 + (mpirank*4/mpisize)
    i_end   = (mpirank + 1)*4/mpisize
    
    DO t=1,2

        DO J=j_start,j_end

            DO I=i_start,i_end

                array(J,I,t) = J+I+t
            
            END DO

        END DO ! J

    END DO ! t

    IF (mpirank==0) WRITE(*,*) 'before MPI_Gather...'
    WRITE(*,*) mpirank, array(1,:,1)
    
    CALL MPI_Barrier(MPI_COMM_WORLD, mpierr)

    CALL MPI_Gather(array(:,i_start:i_end,:), 4*2*2, MPI_INTEGER, & 
                    global_array(:,i_start:i_end,:), 4*2*2, MPI_INTEGER, &
                    0, MPI_COMM_WORLD, mpierr)

    IF (mpirank == 0) THEN
        WRITE(*,*) 'AFTER MPI_Gather...'
        DO J=1,4
            WRITE(*,*) (global_array(J,I,t), I=1,4,1)
        END DO
    END IF
    
    DEALLOCATE(array)
    DEALLOCATE(global_array)

    ! Finalise MPI
    CALL MPI_FINALIZE(mpierr)

end program test 

 And I get these results, which are obviously wrong:

 before MPI_Gather...
           1           0           0           5           6
           0           3           4           0           0
 AFTER MPI_Gather...
           0     5289440           0           0
           0           0           0           0
         225           0           0           0
           0           0           0           0 

I’m using Intel Fortran 2022, and I compile the program with mpiifort

Any help/comment/advice will be truly appreciated!

[AishwaryaCV_Intel]

Hi,

Thank you for posting in Intel Communities.

Could you please provide the command that you used to run your mpi code?

Could you also elaborate on what you are trying to do in the sample code that you provided? 

And What is your Input & expected output? 

Thanks And Regards,

Aishwarya

[Barbara_P_Intel]

I'm moving this over to the HPC Toolkit Forum. MPI questions are answered there.

[AishwaryaCV_Intel]

Hi, 

We haven't heard back from you,Could you please provide us with the requested details? 

Thanks & Regards,

Aishwarya

[AishwaryaCV_Intel]

Hi,

I have not heard back from you. This thread will no longer be monitored by Intel. If you need further assistance, please post a new question.

Thanks and Regards,

Aishwarya