[MITgcm-support] netcdf output files
jklymak at uvic.ca
Thu Apr 11 16:10:24 EDT 2019
First, KL10 outputs to the diagnostics package as well as raw mdsio, so you may want to start using th diagnostics package. Note that using the diagnostics package, even with mdsio, also allows you to have one file per timestep and `globalFiles=.TRUE.` so maybe that is good enough for you?
If you use MNC you will have to glue the netcdf files together.
You may also be interested in my PR for global netcdf files. I’m not sure what the status of it is in the project, but I use it all the time now. https://github.com/MITgcm/MITgcm/pull/31 <https://github.com/MITgcm/MITgcm/pull/31>
> On Apr 11, 2019, at 12:57 PM, Yangxin He <y67he at uwaterloo.ca> wrote:
> Hi there,
> I have been using mdsio output files for a while, and globalFiles=.TRUE., so for each variable I have 2 files *.data and *.meta.
> Now I wanna reduce my file creation rate so I am trying the MNC package. I see the netcdf packs the velocities and temperature files into one netcdf file, so I can reduce file creation by about 8. But at the same time for each cpu I get one netcdf file, currently I use 38 cpus so thats 38 netcdf files for each dumpFreq. Plus I am also using KL10 package and the viscosity files with KL10 do not seem to get compressed into the netcdf files.
> My question is, before I explore more about the MNC, if my goal is to reduce my file creation rate, should I stick to mdsio outputs?
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org <mailto:MITgcm-support at mitgcm.org>
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support <http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the MITgcm-support