Python

MPI in Python

This is a Python example based on one from this MPI with Python presentation. This example creates a number of processes that communicate over MPI and pass a simple message around the processes in a ring. See the presentation for more details.

To run it you'll first need to install Python and the mpi4py library.

Installing Conda

For most Python users we recommend installing Conda and using that to manage your environment. To install Conda:

wget https://github.com/conda-forge/miniforge/releases/download/24.7.1-0/Miniforge-pypy3-24.7.1-0-Linux-x86_64.sh

bash Miniforge-pypy3-24.7.1-0-Linux-x86_64.sh  -b -p $HOME/miniforge
eval "$(${HOME}/miniforge/bin/conda shell.bash hook)"
conda init

In order to make Conda available automatically when you log into the cluster you will also need to add the following to your ~/.bash_profile

if [ -e ${HOME}/.bashrc ]
then
    source ${HOME}/.bashrc
fi

Here is some information on the difference between bashrc and bash_profile

After making these changes log out and log back in.

Install additional packages

You can now use the conda command to install additional packages you'll need. To install mpi4py:

conda install mpi4py

It's worth reading through the Conda users guide. Some useful commands are

• conda list lists all installed packages
• conda search finds available packages that match the provided name, for example conda search torch will find all avaialable versions of torch, pytorch etc
• conda update updates packages

Running the sample program

After installing you can submit the program to the cluster with

sbatch ring.sh

You can then check the status of your job with

squeue -u netid

(replace "netid" with your SU netID). The job should move from the PD (pending) state to the R (run) state and then complete, although this may happen too fast to notice. If squeue reports that you have no jobs then it has completed. Check the output with

cat output/ring.out