Total: 1
Recent GPUs leverage Winograd convolution and structured pruning to significantly accelerate inference. First, Winograd convolution is theoretically 2.25x faster than standard convolution. Second, structured pruning reduces inference time without additional overhead as the pruning ratio increases. However, applying conventional structured pruning alongside Winograd convolution is inefficient. Existing structured pruning methods, which do not account for how GPUs process Winograd convolution, require large pruning unit sizes, leading to significant information loss. In this paper, we propose Winograd Structured Pruning (WINS), the first approach to employ optimized structured pruning for Winograd convolution. WINS is designed based on an in-depth analysis of Winograd convolution's computational characteristics on GPUs. Additionally, we introduce two variants, WINS-B and WINS-AB, which further enhance performance. Experimental results show that WINS-AB achieves up to 2.8x practical speedup in baseline inference on GPUs while preserving the accuracy of ResNet-18 on ImageNet.