parallel netcdf - first experiences

[Jianwei Li]

Right.
But the > 4G filesize is not really solved neither in netCDF3.5 nor
in PnetCDF0.8.6, because this drawback is implanted in the netCDF
file format itself, which makes use of byte-based offsets & sizes
in the file header.

Jianwei

On Mon, 28 Jul 2003, Rob Ross wrote:

> Ok, if that's true then the first call should be hvector() as well.
> Although given the MPI_BYTE type it won't make a lot of difference except
> in cases where the offset is bigger than an int (and MPI_Aint is bigger
> than an int).
>
> Rob
>
> On Mon, 28 Jul 2003, Jianwei Li wrote:
>
> >
> > Yes. That should be MPI_Type_create_hvector in the case of MPI-2.
> > I believe it's a typo, which is confirmed with the right function
> > name right below that part of src code in "mpinetcdf.c" line 1487.
> >
> > Jianwei
> >
> > On Mon, 28 Jul 2003, Rob Ross wrote:
> >
> > > I think that Rajeev has identified the problem :).
> > >
> > > Rob
> > >
> > > On Mon, 28 Jul 2003, Rajeev Thakur wrote:
> > >
> > > > > > When trying to create the test cases I recognized that
> > > > > mpinetcdf.c tries
> > > > > > to call  MPI_Type_create_vector which I guess is not part of the MPI2
> > > > > > standard. Are there any plans to to add this function to the MPI2 in
> > > > > > future? I also could not find this function in the mpich-1.2.5 release.
> > > > > > Comments are welcome.
> > > > >
> > > > > I'm going to defer to someone more familiar with the differnce between
> > > > > MPI-1 and MPI-2.. We *should* be using the MPI_VERSION macro to call the
> > > > > right type function, like this:
> > > > >
> > > > > #if (MPI_VERSION < 2)
> > > > >             MPI_Type_vector(subcount[0], varp->xsz, ncp->recsize,
> > > > >                             MPI_BYTE, &filetype);
> > > > > #else
> > > > >             MPI_Type_create_vector(subcount[0], varp->xsz, ncp->recsize,
> > > > >                                     MPI_BYTE, &filetype);
> > > > > #endif
> > > >
> > > > There isn't a function called MPI_Type_create_vector in MPI. There is a
> > > > type_create_hvector and a type_vector.
> > > >
> > > > Rajeev
> > > >
> > > >
> > >
> >
> >
>