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

950807: netCDF on CRAY

This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.

  • Subject: 950807: netCDF on CRAY
  • Date: Mon, 07 Aug 95 15:25:52 -0600
Doug,

> From: Doug Cripe <address@hidden>
> Organization: CSU
> Keywords: 199508071638.AA16186 netCDF CRAY

In the above message you wrote:
 
> I have written a program to read ARM data in NetCDF format, and
> when I run it on Shavano, a curious thing happens.  Only half of the  
> double-precision 'time-offset' values in the file are returned
> using a call to ncvgt.
> 
> The files have only 1 dimension (time), which is size 24
> (observations taken every half hour, 12 hours per file).  I have no  
> problem with the real (float) defined variables being fully returned  
> (i.e. all 24 data points) by ncvgt.  Moreover, the program runs
> fine on our HP work stations.  But, with Shavano, the 'time-offset'  
> values returned correspond to odd-numbered indicies.  For example,  
> using ncdump, the first 6 time-offset values are:
> 
> 1800, 3600, 5400, 7200, 9000, 10800, ...
> 
> What gets returned with ncvgt is 12 (odd-numbered) values:
> 
> 1800, 5400, 9000, ...
> 
> with the remaining 12 values being junk.
> 
> Obviously, Shavano handles double-precision in a way dissimilar to  
> other computers.  What do I do to compensate for this?

I think there must be a bug in your program -- possibly in the way you're
using the netCDF API.  The reason I think this is because I just created
a CDL description of a netCDF dataset:

    19-shavano% cat a.cdl
    netcdf a {
        dimensions:
            time = 24;

        variables:
            double time_off(time);

        data:
            time_off = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12
                       13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24;
    }

and used it to create the binary netCDF dataset:

    24-shavano% ncgen -b a.cdl

As you can see from the following ncdump(1), it worked just fine:

    25-shavano% ncdump a.nc
    netcdf a {
    dimensions:
            time = 24 ;

    variables:
            double time_off(time) ;

    data:

     time_off = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 
        19, 20, 21, 22, 23, 24 ;
    }

--------
Steve Emmerson   <address@hidden>