The following instructions are for ECCC users.

If you are using SCM outside ECCC, please see the instructions for external users.

How to get, compile and run SCM at the CMC.

SCM - Single Column Model

This project requires an X session to display results. It is recommended to either run on EC machines or obtain a node on PPP with -X:

qsub -I -X -lselect=1:ncpus=80:mpiprocs=80:ompthreads=1:mem=100gb -lplace=scatter

Warning: this repository uses submodules. Make sure you follow the instructions below.

Getting scm git repository

Choose one of the following methods:

1. cloning only the necessary components:

git clone git@gitlab.science.gc.ca:MIG/scm.git
cd scm

2. or cloning everything, including rpn-si libraries (rmn, vgrid, rpncomm, tdpack) in one step:

git clone --recursive git@gitlab.science.gc.ca:MIG/scm.git
cd scm

3. or cloning in several steps

git clone git@gitlab.science.gc.ca:MIG/scm.git
cd scm

Update rpn-si libraries and cmake_rpn submodules

git submodule update --init --recursive

Choosing a version

git branch # what is the current branch
git branch -a # list all branches (look at the list of remote branches to choose from)
git tag # list tags (if you want to select a tagged version)
git checkout <hash|branch|tag> # checkout a branch, a tag, or a specific hash. Example: git checkout 5.3
File MANIFEST shows all the components and versions for hash|branch|tag checked out

Or, if you want, create your own branch from the current branch

git checkout -b mybranch

Preparing scm compilation for Intel compiler

. ./.eccc_setup_intel

Before the first build, or if you made important changes (such as updating other submodules, or adding or removing source files):

. ./.initial_setup

And before the first run, create these links to GEM and SCM databases

./scripts/link-dbase.sh

Building and installing SCM

See cado -h (short help) or cado help. For example: cado cmake generates Makefiles to compile scm, modelutils and rpnphy

Configure:

cado cmake

Compile:

cado build -j

Install in working directory

cado work -j

cado work -j can be used to compile and install in the same step.

In development mode, scm is compiled using Intel shared libraries: use the following command to compile with static libraries:

cado cmake-static

See others options with cado -h (short help) or cado help

Running SCM

cd $SCM_WORK
scmcase -get gabls-3
runscm

See examples in README-SCM file

Structure of the working environment

The structure of the build and work directories is different whether the $storage_model environment variable exists:

The following environment variables are created (examples):

• scm_DIR = directory where the git clone was created

• SCM_WORK = work directory

• SCM_ARCH = architecture, for example ubuntu-18.04-amd64-64-intel-2022.1.2

• COMPILER_SUITE = compiler suite, for example Intel

• COMPILER_VERSION = compiler version, for example 2022.1.2

• SCM_MODEL_DFILES = data base to run SCM cases

• ATM_MODEL_DFILES = data base for physics

• SCMDIAG_PLUGIN_PATH = plugin libraries for scmdiag

• SCMDIAG_SCRIPTS_LIBPATH = libraries for scmdiag

• SCM_STORAGE_DIR = where build and work directories are situated

  • Example if $storage_model variable exists:

    • SCM_STORAGE_DIR=/local/storage/scm/ubuntu-18.04-amd64-64-intel-2022.1.2
    • in scm_DIR:
      • build-ubuntu-18.04-amd64-64-intel-2022.1.2 is a link, such as: /local/storage/scm/ubuntu-18.04-amd64-64-intel-2022.1.2/build
      • work-ubuntu-18.04-amd64-64-intel-2022.1.2 is a link, such as: /local/storage/scm/ubuntu-18.04-amd64-64-intel-2022.1.2/work

  • Example if $storage_model variable doesn't exist:

    • SCM_STORAGE_DIR=$HOME/scm/
    • directories situated in scm_DIR:
      • build-ubuntu-18.04-amd64-64-intel-2022.1.2
      • work-ubuntu-18.04-amd64-64-intel-2022.1.2