Investigation of heterogeneous computing platforms for real-time data analysis in the CBM experiment

• Published in: Computer physics communications 2020-08, Vol.253, p.107190, Article 107190
• Main Authors: Singhal, V., Chattopadhyay, S., Friese, V.
• Format: Article
• Language: English
• Subjects: CBM; CUDA; GPU; Heterogeneous computing; OpenCL
• ISSN: 0010-4655; 1879-2944
• DOI: 10.1016/j.cpc.2020.107190

Future experiments in high-energy physics will pose stringent requirements to computing, in particular to real-time data processing. As an example, the CBM experiment at FAIR Germany intends to perform online data selection exclusively in software, without using any hardware trigger, at extreme interaction rates of up to 10 MHz. In this article, we describe how heterogeneous computing platforms, Graphical Processing Units (GPUs) and CPUs, can be used to solve the associated computing problems on the example of the first-level event selection process sensitive to J/ψ decays using muon detectors. We investigate and compare pure parallel computing paradigms (Posix Thread, OpenMP, MPI) and heterogeneous parallel computing paradigms (CUDA, OpenCL) on both CPU and GPU architectures and demonstrate that the problem under consideration can be accommodated with a moderate deployment of hardware resources, provided their compute power is made optimal use of. In addition, we compare OpenCL and pure parallel computing paradigms on CPUs and show that OpenCL can be considered as a single parallel paradigm for all hardware resources. •Systematic study and development of an event selection algorithm for the CBM-MUCH.•The FLES process suppresses the archival data rate by almost two orders of magnitude.•Process satisfy the CBM requirements for high-rate data taking at 107 events per second.•Almost a million events per second can be processed using a single NVIDIA Tesla GPU.•Comparison performed between OpenCL, pthread, OpenMP and MPI as open-source concurrency paradigms.