Parallel

Computing

and I/O

EduMPI is a novel learning support tool for parallel programming education with MPI. EduMPI makes it possible to easily execute MPI programs on a high-performance cluster via a user interface without having to deal with SSH, Linux commandos or workload management (e.g. through SLURM).

In addition, learners receive immediate feedback by visualizing performance data of the individual processes about the ongoing MPI communication at runtime. This type of performance analysis does not require extensive knowledge (as is the case when using Score-P, Scalasca, TAU, etc.) and additionally shows new types of performance data, such as internal, underlying MPI communications, and algorithms in collective communication.

The EduMPI GUI is made available to students for use as a so-called “AppImage” for Linux. This works without installations or configurations. However, many students do not have a Linux system, which is why it would be necessary to be able to use this GUI on Windows and/or MacOS systems.

The GUI is developed in QT (C++, QML) and is built with CMake. Your task is to find ways of deploying this GUI to Windows or MacOS with as little installation effort as possible. For this purpose, the currently running Linux processes must be adapted so that they work on Windows. In addition, a functioning deployment procedure must be found or developed.

Experience in C++ is an advantage, but not essential, as hardly any changes need to be made
to the source code of the GUI.

• General information on the deployment of Qt applications: https://doc.qt.io/qt-6/deployment.html
• Tools and ideas used to deploy the Linux application: https://github.com/linuxdeploy/linuxdeploy-plugin-qt and https://github.com/linuxdeploy/linuxdeploy-plugin-appimage

Contact: Michael Kuhn and Anna-Lena Roth