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

[netCDF #CBQ-986702]: Why is nc_put_varm unbearable slow?

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 Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated.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: [netCDF #CBQ-986702]: Why is nc_put_varm unbearable slow?
  • Date: Sun, 17 Mar 2013 15:18:42 -0600
Florian-

nc_put_varm (and nc_get_varm) are inherently
slow in their general form because of the nature
of the algorithm required. Bottom line
is that nc_put_varm has a loop that writes
a single value at a time. Thus the total
time is roughly |nc_put_var1| * N
where |nc_put_var1| is the time to write one
element of the array and N is the total number
of writes, which in your program is 640*412.
The nc_put_vara is significantly faster because
its operation is optimized into the netcdf library.
These optimizations cannot be used when the imapp argument
to nc_put_varm is used or the stride is something other than
all ones. If the imapp argument is NULL,
and the stride is NULL or all ones,
then nc_put_varm is equivalent
to and is converted to a call to nc_put_vara.

=Dennis Heimbigner
  Unidata


Ticket Details
===================
Ticket ID: CBQ-986702
Department: Support netCDF
Priority: Normal
Status: Closed