High‐Performance GPU Techniques for 2D Convolution Operators in Deep Neural Networks
Concurrency and Computation: Practice and Experience · 2026-07 · 38 (14)
DOI: 10.1002/cpe.70835 ↗Kurz gefasst
Fachartikel in der Zeitschrift Concurrency and Computation: Practice and Experience, online erschienen am 19.07.2026. Verfasst von Hui Wei und 8 weiteren Autor:innen. Die Arbeit nennt 23 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
This paper presents an optimized implementation of Winograd non‐fused convolution. Our optimizations include both application‐independent and Winograd‐specific software techniques, such as a specialized interface‐kernel data format (tile‐united CNHW layout) to enhance memory access efficiency; warp specialization and double‐buffered prefetching to effectively exploit computational resources and memory bandwidth; and the use of shuffle instructions to conserve hardware resources. We propose a GPU‐based Multi‐Modal Parallelism Method for 2D Winograd non‐fused convolution and provide a supplementary explanation of Winograd's tile extraction, which reduces memory usage and computation. The proposed techniques were evaluated at the kernel level in two environments (ENV1–GTX 980 GPU, CUDA 9.2, cuDNN 7.6.4; ENV2–GTX 1650Ti GPU, CUDA 10.2, cuDNN 8.2.0) using a wide range of CNN layer benchmark‐compliant parameters. Compared with the state‐of‐the‐art Winograd non‐fused convolution in cuDNN, our implementation achieves speedups of 1.64× and 1.28× for the two environments, respectively.
Zitieren
APA
Wei, H., Liu, E., Zhao, Y., Liu, B., Li, D., Xiong, J., Zhang, C., Wei, Y., & Xu, Z. (2026). High‐Performance GPU Techniques for 2D Convolution Operators in Deep Neural Networks. Concurrency and Computation: Practice and Experience, 38(14). https://doi.org/10.1002/cpe.70835
Vancouver
Wei H, Liu E, Zhao Y, Liu B, Li D, Xiong J, et al. High‐Performance GPU Techniques for 2D Convolution Operators in Deep Neural Networks. Concurrency and Computation: Practice and Experience. 2026;38(14). doi:10.1002/cpe.70835
BibTeX
@article{Wei2026,
author = {Wei, Hui and Liu, Enjie and Zhao, Youbing and Liu, Baiyu and Li, Dayou and Xiong, Jinquan and Zhang, Chuan and Wei, Yongqiang and Xu, Zhaoxing},
title = {High‐Performance GPU Techniques for 2D Convolution Operators in Deep Neural Networks},
journal = {Concurrency and Computation: Practice and Experience},
year = {2026},
volume = {38},
number = {14},
doi = {10.1002/cpe.70835}
}
EUVIMED