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

Re: NetCDF Java Read API



great, thanks for the update.

Greg Rappa wrote:
> John Caron wrote:
>> It doesnt appear that you are correctly chunking. an h5dump -h shows:
>>   DATASET "VIL" {
>>       DATATYPE  H5T_STD_I16LE
>>       DATASPACE  SIMPLE { ( 24, 1, 3520, 5120 ) / ( 24, 1, 3520, 5120 ) }
>>       ATTRIBUTE "DIMENSION_LIST" {
>>          DATATYPE  H5T_VLEN { H5T_REFERENCE}
>>          DATASPACE  SIMPLE { ( 4 ) / ( 4 ) }
>>       }
>>
>> chunking on ( 1, 1, 3520, 5120 ) would probably be a reasonable thing
>> to do, depending on how you plan to access the data.
>>
>> It does seem to be using excessive memory anyway, i am investigating.
>> Only Netcdf-Java 4.0 is reliable with HDF5/Netcdf4 files, and we are
>> still shaking bugs out so this may be one.
>>   
> Hi John,
> 
> I finally had an opportunity to re-work my NetCDF chunking sizes today
> and I managed to implement the chunking you recommended
> (1,1,3520,5120).  My new NetCDF files exhibit an h5dump as follows:
> 
> h5dump --header --properties VILForecast.20081125T224500Z.nc
> ...
>   DATASET "VIL" {
>      DATATYPE  H5T_STD_I16LE
>      DATASPACE  SIMPLE { ( 24, 1, 3520, 5120 ) / ( 24, 1, 3520, 5120 ) }
>      STORAGE_LAYOUT {
>         CHUNKED ( 1, 1, 3520, 5120 )
>         SIZE 45012209 (19.219:1 COMPRESSION)
>       }
>      FILTERS {
>         COMPRESSION DEFLATE { LEVEL 1 }
>      }
> ..
> 
> The new chunking sizes appear to fix the Java read problem that I
> reported back on 14 Nov 2008.
> 
> Thanks for your help (and yours too, Russ),
> Greg.
>