Low Latency Edge Classification GNN for Particle Trajectory Tracking on FPGAs

Shi Yu Huang, Yun Chen Yang, Yu Ru Su, Bo Cheng Lai, Javier Duarte, Scott Hauck, Shih Chieh Hsu, Jin Xuan Hu, Mark S. Neubauer

研究成果: Conference contribution同行評審

摘要

In-time particle trajectory reconstruction in the Large Hadron Collider is challenging due to the high collision rate and numerous particle hits. Using GNN (Graph Neural Network) on FPGA has enabled superior accuracy with flexible trajectory classification. However, existing GNN architectures have inefficient resource usage and insufficient parallelism for edge classification. This paper introduces a resource-efficient GNN architecture on FPGAs for low latency particle tracking. The modular architecture facilitates design scalability to support large graphs. Leveraging the geometric properties of hit detectors further reduces graph complexity and resource usage. Our results on Xilinx UltraScale+ VU9P demonstrate 1625x and 1574x performance improvement over CPU and GPU respectively.

原文English
主出版物標題Proceedings - 2023 33rd International Conference on Field-Programmable Logic and Applications, FPL 2023
編輯Nele Mentens, Nele Mentens, Leonel Sousa, Pedro Trancoso, Nikela Papadopoulou, Ioannis Sourdis
發行者Institute of Electrical and Electronics Engineers Inc.
頁面294-298
頁數5
ISBN(電子)9798350341515
DOIs
出版狀態Published - 2023
事件33rd International Conference on Field-Programmable Logic and Applications, FPL 2023 - Gothenburg, Sweden
持續時間: 4 9月 20238 9月 2023

出版系列

名字Proceedings - 2023 33rd International Conference on Field-Programmable Logic and Applications, FPL 2023

Conference

Conference33rd International Conference on Field-Programmable Logic and Applications, FPL 2023
國家/地區Sweden
城市Gothenburg
期間4/09/238/09/23

指紋

深入研究「Low Latency Edge Classification GNN for Particle Trajectory Tracking on FPGAs」主題。共同形成了獨特的指紋。

引用此