Nodes/cores/storage for CLM simulations

[Aber]

Hi, I apologize for the complete newbie question (I asked this for CESM in another part of the forum but I also needed to ask for CLM run offline):
I've never run CLM before. I am trying to determine ~ how much resources I would need on my local university HPC cluster to do so. What are the typical resources required (in terms of number of nodes/cores) to run an global simulation of CLM at a decent resolution (e.g., 1deg) and for a few decades, within an acceptable amount of time? I'd also be very grateful if you could describe how much space is needed to store model outputs in that case (for a typical number of variables and temporal resolutions).
Thanks so much!

 

[oleson]

You could look here at the timing tables.

CESM2 Timing, Performance & Load Balancing Data

cseg.cgd.ucar.edu

These are for our machine cheyenne that was mothballed early this year, our new machine is a bit faster.
For example, this land-only CLM5 BGC-CROP compset at 1deg (1850_DATM%GSWP3v1_CLM50%BGC-CROP_SICE_SOCN_MOSART_CISM2%NOEVOLVE_SWAV_SIAC_SESP) uses 1836 cores and gets about 230 model years per wall-clock day.
I imagine your results will depend on your system configuration.
Currently, a standard CLM monthly average history file is about 285 MB in size.