Hi,
I have encountered a situation where nc_get_vara and nc_get_vars operations run
much slower on NetCDF-4 files than they do on NetCDF-3 files.
Here is an example CDL for one of my files (with irrelevant stuff removed):
netcdf ecmwf_0p5deg_ocf_t2m_mars_opn {
dimensions:
basetime = UNLIMITED ; // (100 currently)
forecast = 32 ;
bounds = 2 ;
latitude = 121 ;
longitude = 101 ;
variables:
float temperature_2m(basetime, forecast, latitude, longitude) ;
temperature_2m:long_name = "Air temperature 2m above the
surface" ;
temperature_2m:units = "K" ;
temperature_2m:_FillValue = 9.96921e+36f ;
temperature_2m:missing_value = 9.96921e+36f ;
temperature_2m:valid_min = 180.f ;
temperature_2m:valid_max = 330.f ;
temperature_2m:standard_name = "air_temperature" ;
temperature_2m:cell_methods = "area: mean" ;
temperature_2m:coordinates = "level" ;
}
I can create a NetCDF-3 or NetCDF-4 file using this CDL. The problem is that
when I try slicing out a single record (on the unlimited dimension) using
nc_get_vara or nc_get_vars functions, it can take an order of magnitude
(really) longer on a NetCDF-4 file than with a NetCDF-3 file. This affects
NetCDF-4 files with or without compression. I've experimented with various
NetCDF-4 chunking options, and that doesn't help either.
Has anyone else encountered such slowness issues with NetCDF-4? My solution is
to keep the data files in NetCDF-3, but I would really like to migrate to
NetCDF-4.
Tim Hume
Centre for Australian Weather and Climate Research
Australian Bureau of Meteorology
