I'm a little confused by specs for the installation of the netCDF library. Per the WRF user's guide:
"If netCDF-4 is used, be sure that it is installed without activating parallel I/O based on HDF5. The WRF modeling system can use either the classic data model from netCDF-3 or the compression options supported in netCDF-4. Beginning with V4.4, the ability to write compressed NetCDF-4 files in parallel is available. With this option performance is slower than with pnetcdf but can be notably faster than the use of regular NetCDF on parallel file systems. Compression provides files significantly smaller than pnetcdf generates. It is expected that files sizes will differ with compression."
At first this appears to indicate that if using netCDF-4, I cannot use a parallel version of the netCDF library. However, the red text seems to indicate that is no longer true beginning in WRF V4.4. Can someone explain this contradiction? I'm on an HPC environment that does not allow for loading a serial version of netCDF when loading intel mpi. So if I can just use the parallel version of netCDF after all that would make life simpler.
"If netCDF-4 is used, be sure that it is installed without activating parallel I/O based on HDF5. The WRF modeling system can use either the classic data model from netCDF-3 or the compression options supported in netCDF-4. Beginning with V4.4, the ability to write compressed NetCDF-4 files in parallel is available. With this option performance is slower than with pnetcdf but can be notably faster than the use of regular NetCDF on parallel file systems. Compression provides files significantly smaller than pnetcdf generates. It is expected that files sizes will differ with compression."
At first this appears to indicate that if using netCDF-4, I cannot use a parallel version of the netCDF library. However, the red text seems to indicate that is no longer true beginning in WRF V4.4. Can someone explain this contradiction? I'm on an HPC environment that does not allow for loading a serial version of netCDF when loading intel mpi. So if I can just use the parallel version of netCDF after all that would make life simpler.