Running HADDOCK3 with MPI
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:
#!/bin/bash
#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
Then:
sbatch docking-protein-protein-mpi.job
Please report any errors as an issue.