accelerating a function that spends a lot of time in ncread

The Matlab profiler shows a function I wrote spends most of its time executing ncread many times to get scalers and array subsets from a single file (actually a NetCDF THREDDS aggregation). My question is: Might I get a significant speed-up using lower level functions, i.e. open the file once with netcdf.open, use netcdf.getVar for reads, and then close it?

[youn]

Hello

did you find a solution for the ncread speed?

I have the same problem here

thanks a lot

Answer 1

[Stephen]

You might.

Have you tried a simple script encasing your proposed netcdt.open and .getVar calls inside the "tic" and "toc"?

It is unclear from the data provided. How many times are you calling ncread? How much time does it take per call? I would compare the profiler results against tic/toc benchmarking of the commands you're thinking of using. If the new functions take a lot less time, it might be worth modifying your code to use those commands.