Abstract
AbstractThis paper presents the first study of Graphcore’s Intelligence Processing Unit (IPU) in the context of particle physics applications. The IPU is a new type of processor optimised for machine learning. Comparisons are made for neural-network-based event simulation, multiple-scattering correction, and flavour tagging, implemented on IPUs, GPUs and CPUs, using a variety of neural network architectures and hyperparameters. Additionally, a Kálmán filter for track reconstruction is implemented on IPUs and GPUs. The results indicate that IPUs hold considerable promise in addressing the rapidly increasing compute needs in particle physics.
Funder
Science and Technology Facilities Council
Publisher
Springer Science and Business Media LLC
Subject
Nuclear and High Energy Physics,Computer Science (miscellaneous),Software
Reference70 articles.
1. Aaij R et al (2018) Physics case for an LHCb Upgrade II—opportunities in flavour physics, and beyond, in the HL-LHC era
2. Abi B et al (2020) Deep Underground Neutrino Experiment (DUNE), far detector technical design report, volume III DUNE far detector technical coordination
3. Leggett C, Shapoval I (2018) Simulating HEP workflows on heterogeneous architectures. In: 14th International Conference on e-Science, p 343. https://doi.org/10.1109/eScience.2018.00087
4. Yeo B, Lee M, Kuno Y (2019) GPU-accelerated event reconstruction for the COMET phase-I experiment
5. Cenci R, Di Luca A, Lazzari F, Morello MJ, Punzi G (2020) Real-time reconstruction of long-lived particles at LHCb using FPGAs. J Phys 1525(1):012101. https://doi.org/10.1088/1742-6596/1525/1/012101
Cited by
1 articles.
订阅此论文施引文献
订阅此论文施引文献，注册后可以免费订阅5篇论文的施引文献，订阅后可以查看论文全部施引文献
1. Reducing Memory Requirements for the IPU using Butterfly Factorizations;Proceedings of the SC '23 Workshops of The International Conference on High Performance Computing, Network, Storage, and Analysis;2023-11-12