中国物理B ›› 2025, Vol. 34 ›› Issue (1): 18501-018501.doi: 10.1088/1674-1056/ad8b38

• • 上一篇    下一篇

Atmospheric neutron single event effects for multiple convolutional neural networks based on 28-nm and 16-nm SoC

Xu Zhao(赵旭)1, Xuecheng Du(杜雪成)1,†, Chao Ma(马超)1, Zhiliang Hu(胡志良)2,3, Weitao Yang(杨卫涛)4, and Bo Zheng(郑波)1   

  1. 1 School of Nuclear Science and Technology, University of South China, Hengyang 421001, China;
    2 Spallation Neutron Source Science Center, Dongguan 523000, China;
    3 Institute of High Energy Physics, Chinese Academy of Sciences (CAS), Beijing 100049, China;
    4 School of Microelectronics, Xidian University, Xian 710071, China
  • 收稿日期:2024-09-22 修回日期:2024-10-21 接受日期:2024-10-25 出版日期:2025-01-25 发布日期:2025-01-02
  • 通讯作者: Xuecheng Du E-mail:duxuecheng@usc.edu.cn
  • 基金资助:
    Project supported by the National Natural Science Foundation of China (Grant No. 12305303), the Natural Science Foundation of Hunan Province of China (Grant Nos. 2023JJ40520, 2024JJ2044, and 2021JJ40444), the Science and Technology Innovation Program of Hunan Province, China (Grant No. 2020RC3054), the Postgraduate Scientific Research Innovation Project of Hunan Province, China (Grant No. CX20240831), the Natural Science Basic Research Plan in the Shaanxi Province of China (Grant No. 2023-JC-QN-0015), and the Doctoral Research Fund of University of South China (Grant No. 200XQD033).

Atmospheric neutron single event effects for multiple convolutional neural networks based on 28-nm and 16-nm SoC

Xu Zhao(赵旭)1, Xuecheng Du(杜雪成)1,†, Chao Ma(马超)1, Zhiliang Hu(胡志良)2,3, Weitao Yang(杨卫涛)4, and Bo Zheng(郑波)1   

  1. 1 School of Nuclear Science and Technology, University of South China, Hengyang 421001, China;
    2 Spallation Neutron Source Science Center, Dongguan 523000, China;
    3 Institute of High Energy Physics, Chinese Academy of Sciences (CAS), Beijing 100049, China;
    4 School of Microelectronics, Xidian University, Xian 710071, China
  • Received:2024-09-22 Revised:2024-10-21 Accepted:2024-10-25 Online:2025-01-25 Published:2025-01-02
  • Contact: Xuecheng Du E-mail:duxuecheng@usc.edu.cn
  • Supported by:
    Project supported by the National Natural Science Foundation of China (Grant No. 12305303), the Natural Science Foundation of Hunan Province of China (Grant Nos. 2023JJ40520, 2024JJ2044, and 2021JJ40444), the Science and Technology Innovation Program of Hunan Province, China (Grant No. 2020RC3054), the Postgraduate Scientific Research Innovation Project of Hunan Province, China (Grant No. CX20240831), the Natural Science Basic Research Plan in the Shaanxi Province of China (Grant No. 2023-JC-QN-0015), and the Doctoral Research Fund of University of South China (Grant No. 200XQD033).

PDF (PC)

1

赞

摘要： The single event effects (SEEs) evaluations caused by atmospheric neutrons were conducted on three different convolutional neural network (CNN) models (Yolov3, MNIST, and ResNet50) in the atmospheric neutron irradiation spectrometer (ANIS) at the China Spallation Neutron Source (CSNS). The Yolov3 and MNIST models were implemented on the XILINX 28-nm system-on-chip (SoC). Meanwhile, the Yolov3 and ResNet50 models were deployed on the XILINX 16-nm FinFET UltraScale+MPSoC. The atmospheric neutron SEEs on the tested CNN systems were comprehensively evaluated from six aspects, including chip type, network architecture, deployment methods, inference time, datasets, and the position of the anchor boxes. The various types of SEE soft errors, SEE cross-sections, and their distribution were analyzed to explore the radiation sensitivities and rules of 28-nm and 16-nm SoC. The current research can provide the technology support of radiation-resistant design of CNN system for developing and applying high-reliability, long-lifespan domestic artificial intelligence chips.

关键词: single event effects, atmospheric neutron, system on chip, convolutional neural network

Abstract: The single event effects (SEEs) evaluations caused by atmospheric neutrons were conducted on three different convolutional neural network (CNN) models (Yolov3, MNIST, and ResNet50) in the atmospheric neutron irradiation spectrometer (ANIS) at the China Spallation Neutron Source (CSNS). The Yolov3 and MNIST models were implemented on the XILINX 28-nm system-on-chip (SoC). Meanwhile, the Yolov3 and ResNet50 models were deployed on the XILINX 16-nm FinFET UltraScale+MPSoC. The atmospheric neutron SEEs on the tested CNN systems were comprehensively evaluated from six aspects, including chip type, network architecture, deployment methods, inference time, datasets, and the position of the anchor boxes. The various types of SEE soft errors, SEE cross-sections, and their distribution were analyzed to explore the radiation sensitivities and rules of 28-nm and 16-nm SoC. The current research can provide the technology support of radiation-resistant design of CNN system for developing and applying high-reliability, long-lifespan domestic artificial intelligence chips.

Key words: single event effects, atmospheric neutron, system on chip, convolutional neural network

中图分类号:  (Semiconductor-device characterization, design, and modeling)

  • 85.30.De
61.82.Fk (Semiconductors) 85.35.-p (Nanoelectronic devices) 61.80.Hg (Neutron radiation effects)