BoGEMMS-HPC: A Geant4 simulation framework tailored for high performance computing applications
The rising complexity of Geant4 simulations to support design, calibration, and data analysis of ever-advancing technologies for X-ray and gamma-ray observations is increasing the demand for computing resources and storage. Modern applications must rely on customisable, ready-to-use simulation frameworks with native support for high-performance computing (HPC) systems, enabling faster simulations. BoGEMMS-HPC is an open-source Geant4-based simulation framework optimised for HPC applications. It builds on the Bologna Geant4 Multi-Mission Simulator (BoGEMMS) software architecture to deliver scalable and customisable simulations. BoGEMMS-HPC enables efficient large-scale simulations by integrating multi-threading and MPI-based parallelism, support for diverse input/output formats (FITS, SQLite, MongoDB), and runtime-configurable geometry and physics models. Its modular design supports containerised deployment and facilitates portability and maintenance of the code. We describe the framework software architecture focusing on the HPC-enabling features, its novel solutions for geometry modelling, and the input/output interface. The multi-threading and multi-tasking parallelisation schemes are benchmarked both on a local machine and on Leonardo, the pre-exascale Tier-0 supercomputer hosted by CINECA, for the three output interfaces. A factor of 40 speed-up is reached for MPI-based scaling on the 56 CPU cores of a Leonardo node, outperforming the Geant4 built-in multi-threading features. The public repository includes two example applications, verified against laboratory measurements.
Authors
- V. Fioretti (ORCID: https://orcid.org/0000-0002-6082-5384)
- N. Parmiggiani (ORCID: https://orcid.org/0000-0002-4535-5329)
- Alex Ciabattoni
- Andrea Bulgarelli
Institutions
- Cineca (IT)
- National Institute for Astrophysics (IT)
- University of Bologna (IT)
Publication Details
- Journal
- Astronomy and Computing
- Published
- 2026-03-07
- DOI
- https://doi.org/10.1016/j.ascom.2026.101105
- Primary Topic
- Distributed and Parallel Computing Systems
- Type
- article
- Field-Weighted Citation Impact
- 0.00