NOTICE: This version of the NSF Unidata web site (archive.unidata.ucar.edu) is no longer being updated.
Current content can be found at unidata.ucar.edu.
To learn about what's going on, see About the Archive Site.
Jennifer Adams <jma@xxxxxxxxxxxxx> writes: > Dear Experts, > I am working with the 4.1 snapshot of 20091102. I have been using the > nc_inq_libvers() routine to discover the version of the library that is > linked with my program (GrADS). Now I am testing various configure > options, specifically "--with-chunk-cache-size=n" and > "--with-chunk-cache-nelems=n".  Is there any way to discover these cache > parameters from within my program -- something like nc_inq_cache()? Proper > settings for cacheing and chunking are critical for optimizing performance > when working with compressed netcdf files. It would be good if the users > could know these settings before setting chunk sizes for creating > compressed netcdf files. > --Jennifer > -- > Jennifer M. Adams > IGES/COLA > 4041 Powder Mill Road, Suite 302 > Calverton, MD 20705 > [1]jma@xxxxxxxxxxxxx > Howdy Jennifer! Thanks for asking a question for which there is an easy answer. Some of the recent support issues have been real stumpers! There are functions to both set and learn about the chunk cache. Changing the cache settings only applies to future opens/creates, it does not affect already-open files. These don't tell you what the file was opened/created with, just the current settings. So if you change them, and want to know what they are for each file you open/create, then you must remember what they were when the file was opened/created. /* Set the cache size, nelems, and preemption policy. */ EXTERNL int nc_set_chunk_cache(size_t size, size_t nelems, float preemption); /* Get the cache size, nelems, and preemption policy. */ EXTERNL int nc_get_chunk_cache(size_t *sizep, size_t *nelemsp, float *preemptionp); Your question makes me note that these functions are not documented. I am adding them now, and they will be in the snapshot documentation tomorrow. Thanks, Ed -- Ed Hartnett -- ed@xxxxxxxxxxxxxxxx
netcdfgroup
archives: