Running HADDOCK3 with MPIΒΆ

To run this example you must have mpi4py installed in the haddock3 python environment, and OpenMPI in the host system.

$ pip install mpi4py
# or
$ conda install -c conda-forge mpi4py

Do not run it as: mpirun -np haddock3 run.cfg, this logic is handled internally.

Edit the .cfg according to the .job parameters you have set

In the .job header:

#SBATCH --nodes=5
#SBATCH --tasks-per-node=96

In the .cfg params

ncores = 480

Then prepare the rest of the .job file according to your cluster, a SLURM example is provided at haddock3/examples/docking-protein-protein/docking-protein-protein-mpi.job and below:

#SBATCH --nodes=5
#SBATCH --tasks-per-node=96
#SBATCH -J haddock3mpi

# make sure anaconda is activated
source $HOME/software/miniconda3/bin/activate
conda activate haddock3

# go to the example directory
cd $HOME/repos/haddock3/examples/docking-protein-protein

# remove any old runs
rm -rf run-mpi

# execute
haddock3 docking-protein-protein-mpi-test.cfg


sbatch docking-protein-protein-mpi.job

Please report any errors as an issue.