Wasn't sure if I should have revived the following thread, but I
*think* I'm having a different issue:
http://hdf-forum.184993.n3.nabble.com/Fwd-H5Fclose-hangs-when-using-parallel-HDF5-td4027003.html
I'm using hdf5-1.8.12 in my code on a PBS cluster. The code works on
two different Mac OSX (2 and 8 core) machines. The code also runs on
the cluster when I use 4 procs (1 node), or 16 procs (2 nodes, 8
procs/node). However, after it runs, and all the *.h5 data files are
written without error, I get a segmentation fault. If I try to run
with other node/proc configurations, like 32 procs (4 nodes and 8
proc/node), the application hangs up when h5fclose_f is called. I have
included a truncated version of the subroutine that contains the call
to h5fclose_f. Oh, and the code runs without error on any node/proc
configuration when I do not use HDF5 output.
Any insight or advice will be much appreciated!
==========================================
integer(HID_T) :: file_id ! File identifier
integer(HID_T) :: memspace ! File identifier
integer(HID_T) :: dset_id ! Dataset identifier
integer(HID_T) :: plist_id ! Property list identifier
integer(HID_T) :: filespace ! Dataspace indentifier in file
integer(HSIZE_T) :: array_dims(4) ! dimensions of the arrays
integer :: error ! Error flag
integer(HSIZE_T), dimension(2) :: chunk_dims
integer(HSIZE_T), dimension(2) :: block
integer(HSIZE_T), dimension(2) :: stride
integer(HSIZE_T), dimension(2) :: count
integer(HSSIZE_T), dimension(2) :: offset
integer :: i,j,k,isc
! initialize FORTRAN interface and open data set and file
call h5open_f(error)
! setup file access property list with parallel I/O access
call h5pcreate_f(H5P_FILE_ACCESS_F,plist_id, error)
call h5pset_fapl_mpio_f(plist_id, comm, MPI_INFO_NULL, error)
! create new h5 file collectively
write(filen,'(i4.4)') nfile
filename ='HDF5-2D/Field_'//filen//'.h5'
nfile = nfile + 1
call h5fcreate_f(filename, H5F_ACC_TRUNC_F, file_id, error,
access_prp = plist_id)
call h5pclose_f(plist_id,error)
! Grid
do i=1,nx
xhdf(i) = xm(i+imin-1)
end do
do j=1,ny
yhdf(j) = ym(j+jmin-1)
end do
array_dims(1) = Nx
call h5ltmake_dataset_float_f(file_id, "X", 1, array_dims,
xhdf(1:Nx), error)
array_dims(1) = Ny
call h5ltmake_dataset_float_f(file_id, "Y", 1, array_dims,
yhdf(1:Ny), error)
! Lagrangian particle data
array_dims(1) = npart
chunk_dims(1) = npart_proc(irank) !npart_
if (irank.eq.1) then
offset(1) = 0
else
offset(1) = sum(npart_proc(1:(irank-1)))
end if
stride(1) = 1
count(1) = 1
block(1) = chunk_dims(1)
! Particle id
call h5screate_simple_f(1, array_dims, filespace, error)
call h5screate_simple_f(1, chunk_dims, memspace, error)
call h5pcreate_f(H5P_DATASET_CREATE_F, plist_id, error)
! call h5pset_chunk_f(plist_id, 1, chunk_dims, error)
call h5dcreate_f(file_id, "part_id", H5T_NATIVE_REAL, filespace, &
dset_id, error, plist_id)
call h5sclose_f(filespace, error)
call h5dget_space_f(dset_id, filespace, error)
call h5sselect_hyperslab_f (filespace, H5S_SELECT_SET_F, offset,
count, error, &
stride, block)
call h5pcreate_f(H5P_DATASET_XFER_F,plist_id,error)
call h5pset_dxpl_mpio_f(plist_id, H5FD_MPIO_INDEPENDENT_F,error)
do i=1,npart_
bufferp(i) = real(part(i)%id,WP)
end do
call h5dwrite_f(dset_id, H5T_NATIVE_REAL, bufferp(:), array_dims, error, &
file_space_id = filespace, mem_space_id =
memspace, xfer_prp = plist_id)
call h5sclose_f(filespace,error)
call h5sclose_f(memspace,error)
call h5dclose_f(dset_id,error)
call h5pclose_f(plist_id,error)
!!!! There are 7 more Lagrangian datasets in the full code
! Eulerian grid data
array_dims(1) = Ny
array_dims(2) = Nx
chunk_dims(1) = ny_
chunk_dims(2) = nx_
stride(1) = 1
stride(2) = 1
count(1) = 1
count(2) = 1
block(1) = chunk_dims(1)
block(2) = chunk_dims(2)
offset(1) = jmin_ - nover - 1
offset(2) = imin_ - nover - 1
! U
call h5screate_simple_f(2, array_dims, filespace, error)
call h5screate_simple_f(2, chunk_dims, memspace, error)
call h5pcreate_f(H5P_DATASET_CREATE_F, plist_id, error)
call h5pset_chunk_f(plist_id, 2, chunk_dims, error)
call h5dcreate_f(file_id, "U", H5T_NATIVE_REAL, filespace, &
dset_id, error, plist_id)
call h5sclose_f(filespace, error)
call h5dget_space_f(dset_id, filespace, error)
call h5sselect_hyperslab_f (filespace, H5S_SELECT_SET_F, offset,
count, error, &
stride, block)
call h5pcreate_f(H5P_DATASET_XFER_F,plist_id,error)
call h5pset_dxpl_mpio_f(plist_id, H5FD_MPIO_INDEPENDENT_F,error)
buffer3(:,:) = transpose(Ui(imin_:imax_,jmin_:jmax_,kmin_))
call h5dwrite_f(dset_id, H5T_NATIVE_REAL, buffer3(:,:),
array_dims, error, &
file_space_id = filespace, mem_space_id =
memspace, xfer_prp = plist_id)
call h5sclose_f(filespace,error)
call h5sclose_f(memspace,error)
call h5dclose_f(dset_id,error)
call h5pclose_f(plist_id,error)
!!!! There are (potentially) 9 more Eulerian datasets in the full code
! close the dataset, the file and the FORTRAN interface
call h5fclose_f(file_id, error)
call h5close_f(error)
==========================================
_______________________________________________
Hdf-forum is for HDF software users discussion.
[email protected]
http://lists.hdfgroup.org/mailman/listinfo/hdf-forum_lists.hdfgroup.org
Twitter: https://twitter.com/hdf5
