Monday, 23 February 2026, 15:00 CET
Conventional MPI trace visualizations become unwieldy as the number of processes increases. Identifying global patterns is challenging, and both the underlying machine topology and parallel program structure are often obscured by the rigid two-dimensional rank-time graphs. To address these limitations, we propose a three-dimensional video visualization of MPI programs that adapts its spatial layout with both the machine topology and the parallel decomposition of the numerical problem while mapping process time evolution to actual video playback time.
Using Blender, we developed a framework that automatically translates MPI trace data into 3D visual scenes. In this webinar, we explore various display choices, including color-time gradients and transparency, to enhance interpretability. Our approach provides a new, intuitive exploration of both large-scale and local behaviors and patterns. We showcase the utility of this method with idle wave and process desynchronization phenomena. In ongoing work, we use the experience gained with the Blender-based workflow to develop an interactive, browser-based tool that can also be used in VR environments.
About the Presenter
Jean-Yves Verhaeghe graduated at FAU in Computational Engineering as part of the EUMaster4HPC program. He is now working at NHR@FAU on “ParaViz3D,” a visualization tool that displays MPI traces as animated 3D scenes. The tool aims at making those traces and MPI performance analysis and optimization more intuitive.
In his spare time he likes photography, movies, and running. He is also actively learning German. He considers pursuing a PhD degree in the next years.