Complex particle reconstruction software used by High Energy Physics experiments already pushes the edges of computing resources with demanding requirements for speed and memory throughput, but the future experiments pose an even greater challenge. Although many supercomputers have already reached petascale capacities using many-core architectures and accelerators, numerous scientific applications still need to be adapted to make use of these new resources. To ensure a smooth transition to a platform-agnostic code base, we developed a prototype of a portability and parallelization framework named vecpar. In this paper, we introduce the technical concepts, the main features and we demonstrate the framework's potential by comparing the runtimes of the single-source vecpar implementation (compiled for different architectures) with native serial and parallel implementations, which reveal significant speedup over the former and competitive speedup versus the latter. Further optimizations and extended portability options are currently investigated and are therefore the focus of future work.

@inproceedings{VAFFAMSKSY23,
	author	 = {Georgiana Mania and Nicholas Styles and Michael Kuhn and Andreas Salzburger and Beomki Yeo and Thomas Ludwig},
	title	 = {{Vecpar -- A Framework for Portability and Parallelization}},
	year	 = {2023},
	booktitle	 = {{Computational Science -- ICCS 2023}},
	editor	 = {Jiri Mikyska and Clelia de Mulatier and Maciej Paszynski and Valeria V. Krzhizhanovskaya and Jack J. Dongarra and Peter M.A. Sloot},
	publisher	 = {Springer Nature Switzerland},
	address	 = {Cham},
	pages	 = {253--267},
	conference	 = {ICCS 2023},
	location	 = {Prague},
	doi	 = {https://doi.org/10.1007/978-3-031-35995-8_18},
	abstract	 = {Complex particle reconstruction software used by High Energy Physics experiments already pushes the edges of computing resources with demanding requirements for speed and memory throughput, but the future experiments pose an even greater challenge. Although many supercomputers have already reached petascale capacities using many-core architectures and accelerators, numerous scientific applications still need to be adapted to make use of these new resources. To ensure a smooth transition to a platform-agnostic code base, we developed a prototype of a portability and parallelization framework named vecpar. In this paper, we introduce the technical concepts, the main features and we demonstrate the framework's potential by comparing the runtimes of the single-source vecpar implementation (compiled for different architectures) with native serial and parallel implementations, which reveal significant speedup over the former and competitive speedup versus the latter. Further optimizations and extended portability options are currently investigated and are therefore the focus of future work.},
	url	 = {https://doi.org/10.1007/978-3-031-35995-8_18},
}