A RISC-V Extended Infrastructure for CNNs Through Pipelined Computing and Data Dependence Optimization

doi:10.1109/TCAD.2025.3565470

	A RISC-V Extended Infrastructure for CNNs Through Pipelined Computing and Data Dependence Optimization
	Luo, Teng 1,2; Xia, Tengfei 1,2; Chen, Jiayuan 3; Fan, Zhihua 1,2; Li, Wenming 1,2; Mu, Yudong 1,2; An, Xuejun 1,2; Ye, Xiaochun 1,2; Fan, Dongrui 1,2
	2025-11-01
发表期刊	IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS
ISSN	0278-0070
卷号	44 期号:11 页码:4141-4154
摘要	With the rapid development of Artificial Intelligence (AI), convolutional neural networks (CNNs) have been widely applied in fields like computer vision and recommendation systems. This growth has intensified the demand for hardware acceleration of CNNs. Existing accelerators are either designed as co-processors or improve performance through extended instructions. While these methods can significantly improve performance, they often result in limited programming and execution flexibility. In this article, we design custom RISC-V instructions specifically for CNNs to maximize data reuse and exploit parallelism. Then, to efficiently execute CNNs instructions, we extend a pipelined vector computing unit (PPVCU). Finally, we incorporate pattern detection logic (PDL) to identify common data dependence patterns in CNNs, enabling the data dependence computing unit (DDCU) to process instructions within each pattern in parallel. Experimental results show that our approach achieves, on average, 9.54x performance improvement and 6.7x energy efficiency improvement compared to our baseline, 8.34x performance improvement, and 3.1x energy efficiency improvement compared to state-of-the-art designs.
关键词	Artificial intelligence Convolution Convolutional neural networks Computer architecture Computational efficiency Pipelines Logic Filters Fans Biological system modeling Convolutional neural networks (CNNs) acceleration dataflow optimization pipelined computing RISC-V extended instructions
DOI	10.1109/TCAD.2025.3565470
收录类别	SCI
语种	英语
资助项目	National Key Research and Development Program of China[2023YFB4503500] ; Institute of Computing Technology, Chinese Academy of Sciences a AT China Mobile Communications Group Company, Ltd. ; Joint Institute, Beijing Nova Program[20220484054] ; Joint Institute, Beijing Nova Program[20230484420] ; Beijing Natural Science Foundation[L234078] ; SKLP Foundation[CLQD202502]
WOS研究方向	Computer Science ; Engineering
WOS类目	Computer Science, Hardware & Architecture ; Computer Science, Interdisciplinary Applications ; Engineering, Electrical & Electronic
WOS记录号	WOS:001600047600021
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
引用统计
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/41611
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Chen, Jiayuan; Fan, Zhihua
作者单位	1.Chinese Acad Sci, Inst Comp Technol, SKLP, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Sch Comp Sci, Beijing 100049, Peoples R China 3.China Mobile Res Inst, Res Dept Network & IT Technol, Beijing 100032, Peoples R China
推荐引用方式 GB/T 7714	Luo, Teng,Xia, Tengfei,Chen, Jiayuan,et al. A RISC-V Extended Infrastructure for CNNs Through Pipelined Computing and Data Dependence Optimization[J]. IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS,2025,44(11):4141-4154.
APA	Luo, Teng.,Xia, Tengfei.,Chen, Jiayuan.,Fan, Zhihua.,Li, Wenming.,...&Fan, Dongrui.(2025).A RISC-V Extended Infrastructure for CNNs Through Pipelined Computing and Data Dependence Optimization.IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS,44(11),4141-4154.
MLA	Luo, Teng,et al."A RISC-V Extended Infrastructure for CNNs Through Pipelined Computing and Data Dependence Optimization".IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS 44.11(2025):4141-4154.