Fortran MPI programming...

[tejattf]

Hi ..

Im new to Fortran and MPI...

Matrix addition Programme...

After compiling.. i got error like
forrtl: severe (24): end-of-file during read, unit -4, file stdin


include 'mpif.h'

Integer rank, size, ierr, status(MPI_STATUS_SIZE)
Integer npoc, init, final

real a(200), b(200), c(200)

call MPI_INIT(ierr)
call MPI_COMM_RANK(MPI_COMM_WORLD, rank, ierr)
call MPI_COMM_SIZE(MPI_COMM_WORLD, nproc, ierr)

print *, 'Enter Size of array:'

read *, size

if (rank .eq. 0 ) then

print *, 'Enter values to A:'

read *,(a(i), i=1,size)


print *, 'Enter values to B:'
read *,(b(i), i=1,size)

end if


blocksize= size/nproc

if ( rank .EQ. 0) then

do i=0, (nproc-1)

call MPI_Send(a(i*blocksize+1), blocksize, MPI_REAL, i, 1,
& MPI_COMM_WORLD, ierr)

call MPI_Send(b, blocksize, MPI_REAL, i, 1,
& MPI_COMM_WORLD, ierr)

end do

else


call MPI_Recv(a((rank-1)*blocksize+1),blocksize,MPI_REAL,0,1,
& MPI_COMM_WORLD,status,ierr)

call MPI_Recv(b, size, MPI_REAL, 0, 1,
& MPI_COMM_WORLD, status, ierr)


end if

if(rank .eq. 0) then

init= blocksize * (nproc - 1)+1;
final = size;

else

init= blocksize * (rank - 1)+1;
final= blocksize * rank;

end if



c(i)=0.

do i= 1, size

c(i)=a(i)+b(i)
end do



if ( rank .eq. 0) then

do i=1, size

call MPI_Recv(c,size,MPI_REAL,i,1,MPI_COMM_WORLD,status,ierr)
end do

do i=1, size
print *, c(i)
end do


else

call MPI_Send(c((rank-1)*blocksize),size,MPI_REAL,0,1,
& MPI_COMM_WORLD,status,ierr)

end if

call MPI_Finalize(ierr)

stop

end
thanks in advance...

 

[douchebag]

I suspect that stdin is closed by mpiexec on all ranks but rank 0. You have a read command that all ranks will execute.

 

[tejattf]

I changed prog like...


Ty for replying...
include 'mpif.h'

Integer rank, size, ierr, status(MPI_STATUS_SIZE)
Integer nproc, init, final
parameter(size=20)

real a(size), b(size), c(size)

call MPI_INIT(ierr)
call MPI_COMM_RANK(MPI_COMM_WORLD, rank, ierr)
call MPI_COMM_SIZE(MPI_COMM_WORLD, nproc, ierr)


if (rank .eq. 0 ) then

open(unit=7, file='par_vect.exe', status='unknown' )

print *, 'Enter Values for A:'

do i=1,size

read (7,*) a(i)

end do

do
print *, 'Enter values for B:'

read (7,*) b(i)

end do

end if


blocksize= size/nproc

if ( rank .EQ. 0) then

do i=0, (nproc-1)

call MPI_Send(a(i*blocksize+1), blocksize, MPI_REAL, i, 1,
& MPI_COMM_WORLD, ierr)

call MPI_Send(b, size, MPI_REAL, i, 1,
& MPI_COMM_WORLD, ierr)

end do

else


call MPI_Recv(a((rank-1)*blocksize+1),blocksize,MPI_REAL,0,1,

& MPI_COMM_WORLD,status,ierr)

call MPI_Recv(b, size, MPI_REAL, 0, 1,
& MPI_COMM_WORLD, status, ierr)


end if

if(rank .eq. 0) then

init= blocksize * (nproc - 1)+1;
final = size;

else

init= blocksize * (rank - 1)+1;
final= blocksize * rank;

end if



c(i)=0.

do i= 1, size

c(i)=a(i)+b(i)
end do



if ( rank .eq. 0) then

do i=1, size

call MPI_Recv(c,size,MPI_REAL,i,1,MPI_COMM_WORLD,status,ierr)
end do

! open(7, file='par_vect.exe')

do i=1, size

write (7,*) c(i)
end do

else

call MPI_Send(c((rank-1)*blocksize),size,MPI_REAL,0,1,
& MPI_COMM_WORLD,status,ierr)

end if

call MPI_Finalize(ierr)

close(7)

end


after running elap time is more than i expect...

in shell script i gave 1 node 2 proc... and outfile is *.exe file....