You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am working on an asyncronous implementation of the parallelio library in cesm. I would like to prototype parallelio moving data from cpu memory to gpu memory, doing some data compression and/or data postprocessing on the gpu and then writing directly from the gpu to disk.
Can I assume the parallel write part of your use case is the typical concurrent writes from multiple processes to a single shared file? e.g. using collective I/O mode.
Just a reminder, PnetCDF (and the classic NetCDF file formats) does not support data compression.
Has anyone tried this?
The text was updated successfully, but these errors were encountered: