2020/01/27 by M. Rovere, Rovere, Marco, Z. Chen +7 · 1 citation
Computer Science · #Advanced Data Storage Technologies #Caching and Content Delivery #Distributed and Parallel Computing Systems #FOS: Physical sciences #High Energy Physics - Experiment (hep-ex) #Instrumentation and Detectors (physics.ins-det)
paper · pdf · doi:10.48550/arxiv.2001.09761
openalex publication_date 2020/01/27 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
One of the challenges of high granularity calorimeters, such as that to be\nbuilt to cover the endcap region in the CMS Phase-2 Upgrade for HL-LHC, is that\nthe large number of channels causes a surge in the computing load when\nclustering numerous digitised energy deposits (hits) in the reconstruction\nstage. In this article, we propose a fast and fully-parallelizable\ndensity-based clustering algorithm, optimized for high occupancy scenarios,\nwhere the number of clusters is much larger than the average number of hits in\na cluster. The algorithm uses a grid spatial index for fast querying of\nneighbours and its timing scales linearly with the number of hits within the\nrange considered. We also show a comparison of the performance on CPU and GPU\nimplementations, demonstrating the power of algorithmic parallelization in the\ncoming era of heterogeneous computing in high energy physics.\n