收藏切换
A Strided Matrix Memory Access Conflict Determination Mechanism for Deep Learning Processors
收藏切换
Zhenxing Zhang1, 2, 3, Yongwei Zhao2, Yuanbo Wen2, Qi Guo2, *, Yunji Chen2, 4, 5
Journal of Computer Research and Development | 2026, 63(7) : 1803 - 1815
Less
收藏切换
Journal of Computer Research and Development | 2026, 63(7): 1803-1815
• Computer Architecture •
A Strided Matrix Memory Access Conflict Determination Mechanism for Deep Learning Processors
Full
Zhenxing Zhang1, 2, 3, Yongwei Zhao2, Yuanbo Wen2, Qi Guo2, *, Yunji Chen2, 4, 5
Affiliations
  • 1School of Computer Science and Technology, University of Science and Technology of China, Hefei 230026
  • 2State Key Lab of Processors (Institute of Computing Technology, Chinese Academy of Sciences), Beijing 100190
  • 3Cambricon Technologies, Beijing 100191
  • 4University of Chinese Academy of Sciences, Beijing 101408
  • 5Institute of AI for Industries, Chinese Academy of Sciences, Nanjing 211135
  • born in 1996. PhD. His main research interests include computer architecture and parallel computing

    born in 1993. PhD, associate professor. Member of CCF. His main research interests include intelligent processors and computer architecture

    born in 1997. PhD, associate professor. Member of CCF. His main research interests include computer architecture and deep learning compiler

    born in 1985. PhD, professor. Senior member of CCF. His main research interests include computer architecture, machine learning, and programming models

    born in 1983. PhD, professor. Fellow of CCF. His main research interests include parallel computing, microarchitecture, hardware verification, and computational intelligence

About Author:

born in 1996. PhD. His main research interests include computer architecture and parallel computing

born in 1993. PhD, associate professor. Member of CCF. His main research interests include intelligent processors and computer architecture

born in 1997. PhD, associate professor. Member of CCF. His main research interests include computer architecture and deep learning compiler

born in 1985. PhD, professor. Senior member of CCF. His main research interests include computer architecture, machine learning, and programming models

born in 1983. PhD, professor. Fellow of CCF. His main research interests include parallel computing, microarchitecture, hardware verification, and computational intelligence

Published: 2026-07-01 doi: 10.7544/issn1000-1239.202550124
Outline
收藏切换

With the rapid development of deep learning applications, deep learning processors (DLPs) such as TensorCore GPU, TPU, MLU have risen rapidly and demonstrated excellent performance in matrix operations. To efficiently handle complex matrix data, these processors typically partition sub-matrices and load them into on-chip memory with fixed stride, establishing a distinct pattern of strided memory access. However, devices like NVIDIA H100, Huawei Ascend, and Cambricon MLU currently depend on complex explicit conflict management, lacking dedicated hardware controls for strided memory access conflicts, thus making it challenging to ensure the atomicity and consistency of instruction execution. Traditional memory conflict detection methods focus on scalar operations and have difficulty capturing dependency conflicts among matrix data. To address this issue, we propose a strided matrix memory access conflict determination method that transforms conflict detection into the problem of solving a binary linear Diophantine equation, thereby accurately determining the dependency relationships between memory access instructions. Simulation experiments demonstrate that this approach markedly enhances the memory performance of typical operators in large language models: memory bandwidth utilization can peak at 94% (LPDDR) and 91% (HBM), while the average determination overhead is reduced to 4.68% and 0.31% when compared with the Power test and set intersection methods, respectively. In addition, hardware evaluation based on a 12 nm process shows that the determination design occupies only 0.02333 mm2 and the power consumption is 4.1194 mW. Overall, this mechanism transitions from complex explicit management to efficient hardware-based determination, effectively improving the memory management of DLPs.

AI computing systems  /  deep learning processor  /  strided memory access  /  memory access conflict  /  linear Diophantine equation
张振兴, 赵永威, 文渊博, 郭崎, 陈云霁. 深度学习处理器矩阵跨步访存冲突判定方法. 计算机研究与发展, 2026 , 63 (7) : 1803 -1815 . DOI: 10.7544/issn1000-1239.202550124
Zhenxing Zhang, Yongwei Zhao, Yuanbo Wen, Qi Guo, Yunji Chen. A Strided Matrix Memory Access Conflict Determination Mechanism for Deep Learning Processors[J]. Journal of Computer Research and Development, 2026 , 63 (7) : 1803 -1815 . DOI: 10.7544/issn1000-1239.202550124
Year 2026 volume 63 Issue 7
71
37
Cite this Article
BibTeX
Article Info
doi: 10.7544/issn1000-1239.202550124
  • Receive Date:2025-03-01
  • Online Date:2026-11-17
  • Published:2026-07-01
Article Data
Affiliations
History
  • Received:2025-03-01
  • Revised:2026-03-04
Affiliations
    1School of Computer Science and Technology, University of Science and Technology of China, Hefei 230026
    2State Key Lab of Processors (Institute of Computing Technology, Chinese Academy of Sciences), Beijing 100190
    3Cambricon Technologies, Beijing 100191
    4University of Chinese Academy of Sciences, Beijing 101408
    5Institute of AI for Industries, Chinese Academy of Sciences, Nanjing 211135
References
Share
https://castjournals.cast.org.cn/joweb/jsjyjyfz/EN/10.7544/issn1000-1239.202550124
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