From Supercomputers to Desktops: an Interactive and Portable System for Particle In Cell Simulation and Visualization using Commodity Hardware

Generating and analyzing visual representations of simulation data in situ (i.e., during its execution) has become increasingly essential for managing the complexity and scale of modern computational models. Traditionally, this approach has been limited to high-performance computing (HPC) environments, where vast quantities of data are generated far faster than they can be stored. However, the need for real-time data analysis and visualization extends beyond these specialized systems.In this paper, we explore how advanced simulation and in situ analysis and visualization frameworks can be employed on commodity hardware and present an experimental system that integrates in situ capabilities with a Particle In Cell simulation (PIConGPU) and different frameworks for streaming visualization (OpenVisus), and virtual reality visualization (Godot). These frameworks are interfaced with the simulation through a computational steering and in situ visualization system (SCIRun) which has been extended to provide handling of large scale simulation data and VR visualization.We conduct performance studies of the experimental system using different commodity hardware configurations, and both containerized and native environments. Our study investigates bottlenecks related to performance constraints imposed by those different environments and explores techniques to efficiently manage and overcome those challenges posed by the large data generated by simulation.