收藏切换
Cross-Layer Optimization for Fault-Tolerant Deep Learning Accelerators
收藏切换
Qing Zhang1, 2, Cheng Liu1, 2, *, Bo Liu3, Haitong Huang1, 2, Ying Wang1, 2, Huawei Li1, 2, Xiaowei Li1, 2
Journal of Computer Research and Development | 2024, 61(6) : 1370 - 1387
Less
收藏切换
Journal of Computer Research and Development | 2024, 61(6): 1370-1387
• Special lssue on Cross-layer Optimization of Processor Chips •
Cross-Layer Optimization for Fault-Tolerant Deep Learning Accelerators
Full
Qing Zhang1, 2, Cheng Liu1, 2, *, Bo Liu3, Haitong Huang1, 2, Ying Wang1, 2, Huawei Li1, 2, Xiaowei Li1, 2
Affiliations
  • 1State Key Lab of Processors (Institute of Computing Technology, Chinese Academy of Sciences), Beijing 100190
  • 2University of Chinese Academy of Sciences, Beijing 100049
  • 3Institute of Beijing Control Engineering, Beijing 100094
  • born in 1998. Master candidate. His main research interests include deep learning hardware acceleration and fault-tolerant chip design

    born in 1984. PhD, associate professor. His main research interests include domain-specific accelerator design, fault-tolerant computing, and near-data computing

    born in 1977. PhD, professor. His main research interests include fault-tolerant computing and onboard computing

    born in 1999. Master candidate. His main research interests include deep learning algorithm and accelerator design, and fault-tolerant deep learning

    born in 1985. PhD, professor. His main research interests include approximate/error-tolerant computing, VLSI design, and energy-efficient accelerators

    born in 1974. PhD, professor. Her main research interests include EDA, approximate computing, fault-tolerant computing, and VLSI verification and test

    born in 1964. PhD, professor. Member of IEEE. His main research interests include VLSI test, fault-tolerant computing, and hardware security

About Author:

born in 1998. Master candidate. His main research interests include deep learning hardware acceleration and fault-tolerant chip design

born in 1984. PhD, associate professor. His main research interests include domain-specific accelerator design, fault-tolerant computing, and near-data computing

born in 1977. PhD, professor. His main research interests include fault-tolerant computing and onboard computing

born in 1999. Master candidate. His main research interests include deep learning algorithm and accelerator design, and fault-tolerant deep learning

born in 1985. PhD, professor. His main research interests include approximate/error-tolerant computing, VLSI design, and energy-efficient accelerators

born in 1974. PhD, professor. Her main research interests include EDA, approximate computing, fault-tolerant computing, and VLSI verification and test

born in 1964. PhD, professor. Member of IEEE. His main research interests include VLSI test, fault-tolerant computing, and hardware security

Published: 2024-06-01 doi: 10.7544/issn1000-1239.202331005
Outline
收藏切换

Fault-tolerant deep learning accelerator is the basis for highly reliable deep learning processing, and is also critical to deploy deep learning in safety-critical applications such as avionics and robotics. Since deep learning is known to be both computing-intensive and memory-intensive, traditional fault-tolerant approaches based on redundant computing will incur substantial overhead including power consumption and chip area. To this end, we propose to characterize deep learning vulnerability difference across both neurons and bits of each neuron, and leverage the vulnerability difference to enable selective protection of the deep learning processing components from the perspective of architecture layer and circuit layer respectively for the sake of lower fault-tolerant design overhead. At the same time, we observe the correlation between model quantization and bit protection overhead of the underlying processing elements of deep learning accelerators, and propose to reduce the bit protection overhead by adding additional quantization constrain without compromising the model accuracy. Finally, we employ Bayesian optimization strategy to co-optimize the correlated cross-layer design parameters at algorithm layer, architecture layer, and circuit layer to minimize the hardware resource consumption while fulfilling multiple user constraints including reliability, accuracy, and performance of the deep learning processing at the same time.

cross-layer optimization  /  fault-tolerant deep learning accelerator  /  vulnerability factor  /  hybrid architecture  /  selective redundancy
张青, 刘成, 刘波, 黄海同, 王颖, 李华伟, 李晓维. 容错深度学习加速器跨层优化. 计算机研究与发展, 2024 , 61 (6) : 1370 -1387 . DOI: 10.7544/issn1000-1239.202331005
Qing Zhang, Cheng Liu, Bo Liu, Haitong Huang, Ying Wang, Huawei Li, Xiaowei Li. Cross-Layer Optimization for Fault-Tolerant Deep Learning Accelerators[J]. Journal of Computer Research and Development, 2024 , 61 (6) : 1370 -1387 . DOI: 10.7544/issn1000-1239.202331005
Year 2024 volume 61 Issue 6
68
32
Cite this Article
BibTeX
Article Info
doi: 10.7544/issn1000-1239.202331005
  • Receive Date:2024-02-01
  • Online Date:2026-11-17
  • Published:2024-06-01
Article Data
Affiliations
History
  • Received:2024-02-01
  • Revised:2024-03-14
Affiliations
    1State Key Lab of Processors (Institute of Computing Technology, Chinese Academy of Sciences), Beijing 100190
    2University of Chinese Academy of Sciences, Beijing 100049
    3Institute of Beijing Control Engineering, Beijing 100094
References
Share
https://castjournals.cast.org.cn/joweb/jsjyjyfz/EN/10.7544/issn1000-1239.202331005
Share to
QR

Scan QR to access full text

Cite this article
BibTeX
Citations
表12种不同金属材料的力学参数
科
Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)
属
Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏
  • BibTeX
  • EndNote
  • RefWorks
  • TxT