NetCDF: One or more variable sizes violate format constraints
Norman, Matthew R.
normanmr at ornl.gov
Wed Aug 1 12:37:19 CDT 2018
Thanks for the quick reply. Yeah, sorry, the code I sent was mangled. It should've been nfmpi, not nf90mpi. Anyway, as you checked, the nf90mpi interface works, but the nfmpi interface appears not to. I'm not sure why that is, but I'm glad to have something working now. I switched everything over to nf90mpi, and it's good to go.
Thanks,
-Matt
________________________________________
From: Wei-keng Liao <wkliao at eecs.northwestern.edu>
Sent: Wednesday, August 01, 2018 11:14 AM
To: Norman, Matthew R.
Cc: parallel-netcdf at lists.mcs.anl.gov
Subject: Re: NetCDF: One or more variable sizes violate format constraints
Strange. I was able to run the codes with 1.6.1 and below is from command ncdump -h.
I am using gfortran based MPI compiler but having some minor problem when
compiling your codes, particularly when calling nf90mpi_def_var.
You do not need to specify the length of dimension ID array. Fortran 90
checks that automatically.
My code is attached. Please give it a try and let me know.
% mpifort put_var.f90 -I$HOME/PnetCDF/1.6.1/include -o put_var -L$HOME/PnetCDF/1.6.1/lib -lpnetcdf
% ./put_var
% ncdump -h testfile.nc
netcdf testfile {
dimensions:
t = UNLIMITED ; // (0 currently)
x = 400 ;
y = 400 ;
z = 200 ;
variables:
double t(t) ;
double x(x) ;
double y(y) ;
double z(z) ;
double hy_dens(z) ;
double hy_dens_theta(z) ;
double dens(t, z, y, x) ;
double uwnd(t, z, y, x) ;
double vwnd(t, z, y, x) ;
double wwnd(t, z, y, x) ;
double theta(t, z, y, x) ;
double windmag(t, z, y, x) ;
double ke(t, z, y, x) ;
double vort_xy(t, z, y, x) ;
double vort_xz(t, z, y, x) ;
double tracer1(t, z, y, x) ;
double tracer2(t, z, y, x) ;
double tracer3(t, z, y, x) ;
}
Wei-keng
More information about the parallel-netcdf
mailing list