B1850_docker

Docker container for CESM compset B1850 resolution f09_g17

CESM docker container for F1850 compset and resolution f09_g17 using bioconda cesm docker as a base image.

• Input dataset is stored and available in zenodo.

Compiling with intel compilers

You need to pass parallel_studio_xe_2018_update1_cluster_edition.tgz for being able to compile with intel compilers.

We are using a trial license for this test (see silent.cfg). However, you should update silent.cfg to pass a proper license for running on your platform.

• Intel compilers:

wget http://registrationcenter-download.intel.com/akdlm/irc_nas/tec/12374/parallel_studio_xe_2018_update1_cluster_edition.tgz

Build docker container

• First clone this repository:

git clone https://github.com/NordicESMhub/B1850_docker_intel.git

• Then make sure you move parallel_studio_xe_2018_update1_cluster_edition.tgz in B1850_docker_intel folder as it will be used for building the container.

• Then build the container:

docker build -t nordicesmhub/cesm_b1850:intel .

Running B1850 with docker

Make sure inputdata is available (it won't download it as we suppose it is already on disk).

• The location of the inputdata is /opt/uio/inputdata

mkdir /opt/uio
wget https://zenodo.org/record/3526181/files/inputdata_B1850.tar.gz
tar zxf inputdata_B1850.tar.gz
mv inputdata_container inputdata

• Model outputs are stored in /opt/uio/archive along with the case folder (it can be interesting to check timing).

docker run -i -v /opt/uio/inputdata:/home/cesm/inputdata -v /opt/uio/archive:/home/cesm/archive  -t nordicesmhub/cesm_b1850:intel

Once you start your container, use run_b1850 or any available sub-cases (run_b1850case1 to run_b1850_case6), depending on the number of processors you wish to use.

Update CESM_PES in run_b1850 to change the number of processors per node. We ran this test on [PiZ-Dain] on GPU partitions (12 processors per node).