I have created a Python software package that allows users to take advantage of multi-processing on multiple nodes. This code is developed as a combination of the multi-process API and the MPI4Py API (similar to a MAP). By leveraging the MPI4Py code we can distribute the data across nodes and then generate a pool on each node to process the data concurrently.
The benefits of this code allows users to take full advantage of mapping programming pools to different allocated compute nodes. while the implementations of this are diverse I am currently in the process of writing a paper to list the computational speed up, comparison to other software and analysis of the code.
Feel free to checkout the code by clicking here