@article{ART003280265},
author={Wonho Lee and Jong Wook Kwak},
title={Channel-Wise Thread Indexing for Performance Improvement of Decomposable Winograd Convolution},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2025},
volume={30},
number={12},
pages={15-24}
TY - JOUR
AU - Wonho Lee
AU - Jong Wook Kwak
TI - Channel-Wise Thread Indexing for Performance Improvement of Decomposable Winograd Convolution
JO - Journal of The Korea Society of Computer and Information
PY - 2025
VL - 30
IS - 12
PB - The Korean Society Of Computer And Information
SP - 15
EP - 24
SN - 1598-849X
AB - Convolutional neural networks (CNNs) often require large receptive fields, making acceleration algorithms and GPU kernel configurations key factors in optimizing inference performance. In decomposable Winograd convolution algorithms, previous thread indexing methods lead to thread divergence, where threads within a warp are serialized due to differences in filter sizes. In this paper, we introduce a channel-wise thread indexing that eliminates thread divergence by mapping each convolutional filter channel to the same warp. This approach ensures uniform filter sizes across threads within a warp, significantly enhancing performance. Experiments show that the proposed method removes all potential thread divergence across diverse convolution configurations, including various filter sizes and input/output channel counts, reducing execution time up to 15% on state-of-the-art CNN models. These results demonstrate the potential for improving CNN computational efficiency on SIMT architectures.
KW - GPU;GPGPU(General Purpose computing on Graphics Processing Units);;CUDA(Compute Unified Device Architecture);CNN(Convolutional Neural Network);;thread divergence
DO -
UR -
ER -
Wonho Lee and Jong Wook Kwak. (2025). Channel-Wise Thread Indexing for Performance Improvement of Decomposable Winograd Convolution. Journal of The Korea Society of Computer and Information, 30(12), 15-24.
Wonho Lee and Jong Wook Kwak. 2025, "Channel-Wise Thread Indexing for Performance Improvement of Decomposable Winograd Convolution", Journal of The Korea Society of Computer and Information, vol.30, no.12 pp.15-24.
Wonho Lee, Jong Wook Kwak "Channel-Wise Thread Indexing for Performance Improvement of Decomposable Winograd Convolution" Journal of The Korea Society of Computer and Information 30.12 pp.15-24 (2025) : 15.
Wonho Lee, Jong Wook Kwak. Channel-Wise Thread Indexing for Performance Improvement of Decomposable Winograd Convolution. 2025; 30(12), 15-24.
Wonho Lee and Jong Wook Kwak. "Channel-Wise Thread Indexing for Performance Improvement of Decomposable Winograd Convolution" Journal of The Korea Society of Computer and Information 30, no.12 (2025) : 15-24.
Wonho Lee; Jong Wook Kwak. Channel-Wise Thread Indexing for Performance Improvement of Decomposable Winograd Convolution. Journal of The Korea Society of Computer and Information, 30(12), 15-24.
Wonho Lee; Jong Wook Kwak. Channel-Wise Thread Indexing for Performance Improvement of Decomposable Winograd Convolution. Journal of The Korea Society of Computer and Information. 2025; 30(12) 15-24.
Wonho Lee, Jong Wook Kwak. Channel-Wise Thread Indexing for Performance Improvement of Decomposable Winograd Convolution. 2025; 30(12), 15-24.
Wonho Lee and Jong Wook Kwak. "Channel-Wise Thread Indexing for Performance Improvement of Decomposable Winograd Convolution" Journal of The Korea Society of Computer and Information 30, no.12 (2025) : 15-24.
