Fast Convolution Meets Low Precision: Exploring Efficient Quantized Winograd Convolution on Modern CPUs

doi:10.1145/3632956

	Fast Convolution Meets Low Precision: Exploring Efficient Quantized Winograd Convolution on Modern CPUs
	Wang, Xueying 1,5; Li, Guangli 2,3,6,7; Jia, Zhen 4,8; Feng, Xiaobing 2,3,6,7; Wang, Yida 4,8
	2024-03-01
发表期刊	ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION
ISSN	1544-3566
卷号	21 期号:1 页码:26
摘要	Low-precision computation has emerged as one of the most effective techniques for accelerating convolutional neural networks and has garnered widespread support on modern hardware. Despite its effectiveness in accelerating convolutional neural networks, low-precision computation has not been commonly applied to fast convolutions, such as the Winograd algorithm, due to numerical issues. In this article, we propose an effective quantizedWinograd convolution, named LoWino, which employs an in-side quantization method in theWinograd domain to reduce the precision loss caused by transformations. Meanwhile, we present an efficient implementation that integrates well-designed optimization techniques, allowing us to fully exploit the capabilities of low-precision computation on modern CPUs. We evaluate LoWino on two Intel Xeon Scalable Processor platforms with representative convolutional layers and neural network models. The experimental results demonstrate that our approach can achieve an average of 1.84x and 1.91x operator speedups over state-of-the-art implementations in the vendor library while preserving accuracy loss at a reasonable level.
关键词	Deep learning winograd convolution low-precision computation
DOI	10.1145/3632956
收录类别	SCI
语种	英语
资助项目	National Key R&D Program of China[2021ZD0110101] ; National Natural Science Foundation of China[62090024] ; National Natural Science Foundation of China[62232015] ; National Natural Science Foundation of China[62302479] ; China Postdoctoral Science Foundation[2023M733566] ; Innovation Funding of ICT, CAS[E361010]
WOS研究方向	Computer Science
WOS类目	Computer Science, Hardware & Architecture ; Computer Science, Theory & Methods
WOS记录号	WOS:001193465400005
出版者	ASSOC COMPUTING MACHINERY
引用统计
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/38765
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Li, Guangli
作者单位	1.Beijing Univ Posts & Telecommun, Beijing, Peoples R China 2.Chinese Acad Sci, Inst Comp Technol, Beijing, Peoples R China 3.Univ Chinese Acad Sci, Beijing, Peoples R China 4.Amazon Web Serv, Seattle, WA USA 5.Beijing Univ Posts & Telecommun, 10 Xitucheng Rd, Beijing 100876, Peoples R China 6.Chinese Acad Sci, Inst Comp Technol, 6 Kexueyuan South Rd, Beijing 100190, Peoples R China 7.Univ Chinese Acad Sci, 6 Kexueyuan South Rd, Beijing 100190, Peoples R China 8.AmazonWeb Serv, 2795 Augustine Dr, Santa Clara, CA 95054 USA
推荐引用方式 GB/T 7714	Wang, Xueying,Li, Guangli,Jia, Zhen,et al. Fast Convolution Meets Low Precision: Exploring Efficient Quantized Winograd Convolution on Modern CPUs[J]. ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION,2024,21(1):26.
APA	Wang, Xueying,Li, Guangli,Jia, Zhen,Feng, Xiaobing,&Wang, Yida.(2024).Fast Convolution Meets Low Precision: Exploring Efficient Quantized Winograd Convolution on Modern CPUs.ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION,21(1),26.
MLA	Wang, Xueying,et al."Fast Convolution Meets Low Precision: Exploring Efficient Quantized Winograd Convolution on Modern CPUs".ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION 21.1(2024):26.