Thanks for the infos.
In the meantime I found a workaround.
Instead of resized datatypes with holes I use simple vectors
with appropriately calculated offsets in MPI_FILE_WRITE_AT.
Probably not as elegant, but seems to work OK.
On 7/18/07, Robert Latham <robl_at_[hidden]> wrote:
> On Tue, Jul 10, 2007 at 04:36:01PM +0000, jody wrote:
> > I think there is still some problem.
> > I create different datatypes by resizing MPI_SHORT with
> > different negative lower bounds (depending on the rank)
> > and the same extent (only depending on the number of processes).
> > However, I get an error as soon as MPI_File_set_view is called with my
> > datatype:
> > Error: Unsupported datatype passed to ADIOI_Count_contiguous_blocks
> > [aim-nano_02:22229] MPI_ABORT invoked on rank 0 in communicator
> > MPI_COMM_WORLD with errorcode 1
> Hi Jody
> I was wrong about this being a problem with OpenMPI's version of
> ROMIO. The OpenMPI guys have synced up fairly recently with the
> OpenMPI in MPICH2.
> ROMIO, even the very latest in CVS version, doesn't support resized
> types yet.
> Looks like you'll have to take George's alternate idea of MPI_UB and
> We'll let the OpenMPI guys know when resized support is in place.
> Sorry for the confusion.
> Rob Latham
> Mathematics and Computer Science Division A215 0178 EA2D B059 8CDF
> Argonne National Lab, IL USA B29D F333 664A 4280 315B
> users mailing list