Performance optimization of convolution calculation by blocking and sparsity on GPU.

Convolution neural network (CNN) plays a paramount role in machine learning, which has made significant contributions in medical image classification, natural language processing, recommender system and so on. A successful convolution neural network can achieve excellent performance with fast execution time. The convolution operation dominates the total operation time of convolution neural network. Therefore, in this paper, we propose a novel convolution method on Graphic Processing Units (GPUs), which reduces the convolution operation time and improves the execution speed by approximately 2X than the state of the art convolution algorithm. Our work is based on the observation that the sparsity of the input feature map of convolution operation is relatively large, and the zero value of the feature map is redundancy for convolution result. Therefore, we skip the zero value calculation and improve the speed by compressing the feature map. Besides, the shape of the feature map for the deep network is small, and the number of threads is limited. Therefore, for a limited number of threads, it is necessary to reduce the amount of calculation to increase the calculation speed. Our algorithm has a good effect on the convolution operation for the feature map of the deep network with large sparsity and small size.