Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

  1. Load the same modules that a full E3SM case would load:

    1. E3SM’s CIME provides a script which will give the csh output needed to load the correct modules and other configuration settings:

    2. eval `~/codes/acme/cime/scripts/Tools/get_case_env`

    3. Most users will be using the bash shell, in which case examine the output of ~/codes/acme/cime/scripts/Tools/get_case_env and modify to work with bash

  2. Run cmake from your $wdir

    1. cd $wdir

    2. cmake -C $mach $homme

    3. make -j4 theta-l (optional step to see if code will compile)

    4. For additional CMAKE options, see $homme/README.cmake. Options include number of vertical levels (default 26), type of output, native or lat/lon or disabled, ability to enable/disable energy diagnostics, ability to disable kokkos components (SL transport).

  3. After step 2 is complete, one can build many different targets. The generic targets preqx, theta-l, sweqx use the configurations options specified above. Then there are a host of specific targets such as “theta-l-nlev20-native” that are used by various test cases.

  4. Run DCMIP2012 Test 3-1 (simple gravity wave)

    1. cd $wdir/dcmip_tests/dcmip2012_test3.1_nh_gravity_waves/theta-l

    2. make install (copies test case related files from $homme)

    3. ./build.sh (build the correct executable with correct compile time options for this test)

    4. ls jobscript-sh*.

      1. look for a suitable jobscript for your machine. most test cases include “jobscript-snl.sh” which runs on E3SM institutional clusters, and jobscript-knl.sh for NERSC Cori. Some editing may be neccessary.

      2. The jobscript usually runs included NCL scripts for creating standard test case plots from the output

  5. All the DCMIP2012 and DCMIP2016 test cases can be run similar to the above example for Test 3-1.

  6. Other standalone HOMME test cases of interest. These usually contain a single script that will configure and build the model (assuming step1 above is done)

    1. Held-Suarez: see $homme/test/held_suarez0 and the script “hs.job”

    2. NGGPS and V1 benchmarks: see $homme/test/benchmark and associated scripts.

  7. HOMME doubly periodic test cases