Running CUPiD via CESM Workflow

Running CUPiD via CESM Workflow#

CUPiD can be run either independently or via the CESM workflow. If you want to install CUPiD and run directly, we recommend looking at the Run Standalone pages. If you already have CESM installed and want to run CUPiD after the short term archiver is run as part of the CESM workflow, you have come to the right place!

Setup#

Install cupid analysis and infrastructure environments per the usual setup instructions.

Information on what is being run when running via CESM workflow#

  • Look at helper_scripts/cesm_postprocessing.sh for information on how CUPiD is run as part of the CESM workflow

  • Look at cime_config/config_tool.xml for more detailed env_postprocessing xml variables

Adjust CUPiD configuration within CESM#

XML changes are a simple way to change the CUPiD configuration when you are running from CESM. See variables that you may want to adjust in your case directory with this query:

./xmlquery -p CUPID

Adjust any variables that you would like (eg, the base case to compare against, how long to run for, etc) with the following command:

./xmlchange <VARIABLE>=<VALUE>

Descriptions of XML variables and default values are in CUPiD/cime_config/config_tool.xml.

Adjust wallclock time if needed#

You can adjust the wallclock time for CUPiD specifically by running the following command:

./xmlchange --subgroup case.cupid JOB_WALLCLOCK_TIME={new time}

You might want to do this if you are running an example with lots of computationally intensive diagnostic notebooks (for instance, a wallclock time of 6hr may be necessary for ILAMB).

Adjust memory if needed#

Change the memory required for running CUPiD as needed by running a command like the following:

./xmlchange CUPID_MEM_PER_TASK={new memory [in GB]}

For instance, you may want to use 35 GB per task when running ILAMB. Note that the default is 10GB per task.

How to run CUPiD#

If you want to run CUPiD automatically after the short-term archiver finishes, you can run the following command from your case directory in order to turn on postprocessing:

./xmlchange RUN_POSTPROCESSING=TRUE

Alternatively, in your case directory, you can also run CUPiD independently with this command after case.st_archive has completed:

./case.submit --only-job case.cupid
Contents