TrackGNN: A Highly Parallelized and Self-Adaptive GNN Accelerator for Track Reconstruction on FPGAs

Shuyang Li; Hanqing Zhang; Ruiqi Chen; Bruno Da Silva; Giorgian Borca-Tasciuc; Dantong Yu; Cong Hao

doi:10.1109/FCCM62733.2025.00029

TrackGNN: A Highly Parallelized and Self-Adaptive GNN Accelerator for Track Reconstruction on FPGAs

Shuyang Li
, Hanqing Zhang
, Ruiqi Chen
, Bruno Da Silva
, Giorgian Borca-Tasciuc
, Dantong Yu
, Cong Hao

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

1 Scopus citations

Abstract

Real-time track reconstruction in high energy physics imposes stringent latency constraints, hindering the deployment of graph neural networks (GNNs) on general-purpose platforms. We present TrackGNN¹¹https//github.com/silvenachen/TrackGNN, an open-sourced GNN accelerator for track reconstruction. Using a dataflow architecture with multiple parallelism and a self-adaptive renaming mechanism, TrackGNN shows 27.6× speedup over CPUs, up to 101.1× over GPUs, and 5.7× over an FPGA overlay. Compared with FlowGNN, the renaming mechanism also reduces end-to-end latency by 1.12-1.16× with negligible resource overhead.

Original language	English (US)
Title of host publication	Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	269
Number of pages	1
ISBN (Electronic)	9798331502812
DOIs	https://doi.org/10.1109/FCCM62733.2025.00029
State	Published - 2025
Event	33rd IEEE Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025 - Fayetteville, United States Duration: May 4 2025 → May 7 2025

Publication series

Name	Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025

Conference

Conference	33rd IEEE Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025
Country/Territory	United States
City	Fayetteville
Period	5/4/25 → 5/7/25

All Science Journal Classification (ASJC) codes

Artificial Intelligence
Computer Science Applications
Control and Optimization
Software

Access to Document

10.1109/FCCM62733.2025.00029

Cite this

Li, S., Zhang, H., Chen, R., Da Silva, B., Borca-Tasciuc, G., Yu, D., & Hao, C. (2025). TrackGNN: A Highly Parallelized and Self-Adaptive GNN Accelerator for Track Reconstruction on FPGAs. In Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025 (pp. 269). (Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/FCCM62733.2025.00029

Li, Shuyang ; Zhang, Hanqing ; Chen, Ruiqi et al. / TrackGNN : A Highly Parallelized and Self-Adaptive GNN Accelerator for Track Reconstruction on FPGAs. Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025. Institute of Electrical and Electronics Engineers Inc., 2025. pp. 269 (Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025).

@inproceedings{cdd66f6a7e9945d191bdcdfdb545f1ca,

title = "TrackGNN: A Highly Parallelized and Self-Adaptive GNN Accelerator for Track Reconstruction on FPGAs",

abstract = "Real-time track reconstruction in high energy physics imposes stringent latency constraints, hindering the deployment of graph neural networks (GNNs) on general-purpose platforms. We present TrackGNN11https//github.com/silvenachen/TrackGNN, an open-sourced GNN accelerator for track reconstruction. Using a dataflow architecture with multiple parallelism and a self-adaptive renaming mechanism, TrackGNN shows 27.6× speedup over CPUs, up to 101.1× over GPUs, and 5.7× over an FPGA overlay. Compared with FlowGNN, the renaming mechanism also reduces end-to-end latency by 1.12-1.16× with negligible resource overhead.",

author = "Shuyang Li and Hanqing Zhang and Ruiqi Chen and \{Da Silva\}, Bruno and Giorgian Borca-Tasciuc and Dantong Yu and Cong Hao",

note = "Publisher Copyright: {\textcopyright} 2025 IEEE.; 33rd IEEE Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025 ; Conference date: 04-05-2025 Through 07-05-2025",

year = "2025",

doi = "10.1109/FCCM62733.2025.00029",

language = "English (US)",

series = "Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "269",

booktitle = "Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025",

address = "United States",

}

Li, S, Zhang, H, Chen, R, Da Silva, B, Borca-Tasciuc, G, Yu, D & Hao, C 2025, TrackGNN: A Highly Parallelized and Self-Adaptive GNN Accelerator for Track Reconstruction on FPGAs. in Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025. Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025, Institute of Electrical and Electronics Engineers Inc., pp. 269, 33rd IEEE Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025, Fayetteville, United States, 5/4/25. https://doi.org/10.1109/FCCM62733.2025.00029

TrackGNN: A Highly Parallelized and Self-Adaptive GNN Accelerator for Track Reconstruction on FPGAs. / Li, Shuyang; Zhang, Hanqing; Chen, Ruiqi et al.
Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025. Institute of Electrical and Electronics Engineers Inc., 2025. p. 269 (Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - TrackGNN

T2 - 33rd IEEE Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025

AU - Li, Shuyang

AU - Zhang, Hanqing

AU - Chen, Ruiqi

AU - Da Silva, Bruno

AU - Borca-Tasciuc, Giorgian

AU - Yu, Dantong

AU - Hao, Cong

PY - 2025

Y1 - 2025

N2 - Real-time track reconstruction in high energy physics imposes stringent latency constraints, hindering the deployment of graph neural networks (GNNs) on general-purpose platforms. We present TrackGNN11https//github.com/silvenachen/TrackGNN, an open-sourced GNN accelerator for track reconstruction. Using a dataflow architecture with multiple parallelism and a self-adaptive renaming mechanism, TrackGNN shows 27.6× speedup over CPUs, up to 101.1× over GPUs, and 5.7× over an FPGA overlay. Compared with FlowGNN, the renaming mechanism also reduces end-to-end latency by 1.12-1.16× with negligible resource overhead.

AB - Real-time track reconstruction in high energy physics imposes stringent latency constraints, hindering the deployment of graph neural networks (GNNs) on general-purpose platforms. We present TrackGNN11https//github.com/silvenachen/TrackGNN, an open-sourced GNN accelerator for track reconstruction. Using a dataflow architecture with multiple parallelism and a self-adaptive renaming mechanism, TrackGNN shows 27.6× speedup over CPUs, up to 101.1× over GPUs, and 5.7× over an FPGA overlay. Compared with FlowGNN, the renaming mechanism also reduces end-to-end latency by 1.12-1.16× with negligible resource overhead.

UR - https://www.scopus.com/pages/publications/105007988884

UR - https://www.scopus.com/pages/publications/105007988884#tab=citedBy

U2 - 10.1109/FCCM62733.2025.00029

DO - 10.1109/FCCM62733.2025.00029

M3 - Conference contribution

AN - SCOPUS:105007988884

T3 - Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025

SP - 269

BT - Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 4 May 2025 through 7 May 2025

ER -

Li S, Zhang H, Chen R, Da Silva B, Borca-Tasciuc G, Yu D et al. TrackGNN: A Highly Parallelized and Self-Adaptive GNN Accelerator for Track Reconstruction on FPGAs. In Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025. Institute of Electrical and Electronics Engineers Inc. 2025. p. 269. (Proceedings - 2025 IEEE 33rd Annual International Symposium on Field-Programmable Custom Computing Machines, FCCM 2025). doi: 10.1109/FCCM62733.2025.00029

TrackGNN: A Highly Parallelized and Self-Adaptive GNN Accelerator for Track Reconstruction on FPGAs

Abstract

Publication series

Conference

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this