[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[netCDF #FIU-947657]: Re: NetCDF-4 patches

Hi Karen,

On the phone, you told me grep showed no occurrence of nf90_mpiio in any of
the netCDF 4.2 sources, which I have verified.

However, after I did a "make install", I noticed these definitions in
/usr/local/include:

  grep -i -nH -e mpiio *
  netcdf.h:127:#define NC_MPIIO       0x2000 
  netcdf.inc:1430:      integer nf_mpiio      
  netcdf.inc:1431:      parameter (nf_mpiio = 8192)
  netcdf.mod:649:674 'nf90_mpiio' 'netcdf' 'nf90_mpiio' 1 ((PARAMETER 
UNKNOWN-INTENT
  netcdf.mod:5141:'nf90_mpiio' 0 674 'nf90_mpiposix' 0 675 'nf90_netcdf4' 0 676

I don't know how to interpret what's in the binary F90 .mod module files,
but that would be where parameter constants get defined.  Are you doing
what's needed to include the netcdf module in your F90 code?

--Russ

> > There used to be flags to control how the parallel open was performed
> > (nf90_mpiio and nf90_mpiposix).  These appear to be gone altogether.  What
> > is the strategy now.  I had been using nf90_mpiio and need to figure out a
> > way to maintain builds against both source bases.
> 
> Is this the change that caused these not to work?
> 
> https://www.unidata.ucar.edu/jira/browse/NCF-135
> 
> If so, maybe this example at the bottom of that bug report shows how to do it?
> 
> Note Fortran test code shows how to call from Fortran with a communicator that
> will work:
> 
> nf90_create(path="/tmp/test.nc", cmode=IOR(NF90_NETCDF4, NF90_MPIIO), 
> ncid=ncid, &
> comm=MPI_COMM_WORLD, info=MPI_INFO_NULL)
> 
> --Russ
> 
> Russ Rew                                         UCAR Unidata Program
> address@hidden                      http://www.unidata.ucar.edu
> 
> 

Russ Rew                                         UCAR Unidata Program
address@hidden                      http://www.unidata.ucar.edu



Ticket Details
===================
Ticket ID: FIU-947657
Department: Support netCDF
Priority: Normal
Status: Closed