MOFA Mini App

A containerized example application of MOF generation code found here.

How to run:

0. Create your GitHub personal access token, see here.

Docker

CPU

1. Pull image from Github Registry

echo <personal_access_token> | docker login ghcr.io -u <github_username> --password-stdin
docker pull ghcr.io/diaspora-project/mofa-mini-app:main
docker tag ghcr.io/diaspora-project/mofa-mini-app:main mofa

2. Run container to execute application

docker run --rm mofa

GPU

1. Pull image from Github Registry

echo <personal_access_token> | docker login ghcr.io -u <github_username> --password-stdin
docker pull ghcr.io/diaspora-project/mofa-mini-app:gpu
docker tag ghcr.io/diaspora-project/mofa-mini-app:gpu mofa-gpu

2. Run container to execute application

docker run --rm --gpus all mofa-gpu

Apptainer on Polaris (GPU)

To run on Polaris, it is recommended to follow instructions for running Apptainer on Polaris provided here.

1. Load Apptainer modules

module use /soft/spack/gcc/0.6.1/install/modulefiles/Core
module load apptainer

2. Build Apptainer image from Github Registry

echo <personal_access_token> | apptainer remote login -u <github_username> --password-stdin docker://ghcr.io
export APPTAINER_TMPDIR=/local/scratch # have more space available for building image
apptainer build --fakeroot --force mofa docker://ghcr.io/diaspora-project/mofa-mini-app:gpu

3. Clone and cd to the MOFa code

git clone https://github.com/globus-labs/mof-generation-at-scale.git
cd mof-generation-at-scale

4. Generate required input files in apptainer environment (optional, if files have already been generated)

cd input-files/zn-paddle-pillar
apptainer exec mofa /bin/bash
source activate mofa
python assemble_inputs.py
exit
cd ../../

5. Run container from mof-generation-at-scale root dir to execute application

apptainer run --nv ../mofa