SPRITE: Sparsity-Aware Neural Processing Unit with Constant Probability of Index-Matching
Sungju Ryua, Youngtaek Ohb, Taesu Kimc, Daehyun Ahnd and Jae-Joon Kime
Pohang University of Science and Technology (POSTECH), Pohang, Republic of Korea
aMarcello.Traiola@lirmm.fr@postech.ac.kr
bArnaud.Virazel@lirmm.fr@postech.ac.kr
cPatrick.Girard@lirmm.fr@postech.ac.kr
dPatrick.Girard@lirmm.fr@postech.ac.kr
ePatrick.Girard@lirmm.fr@postech.ac.kr
ABSTRACT
Sparse neural networks are widely used for memory savings. However, irregular indices of non-zero input activations and weights tend to degrade the overall system performance. This paper presents a scheme to maintain constant probability of indexmatching for weight and input over a wide range of sparsity overcoming a critical limitation in previous works. A sparsityaware neural processing unit based on the proposed scheme improves the system performance up to 6.1⨯ compared to previous sparse convolutional neural network hardware accelerators.
Keywords: Sparse Neural Network, Convolutional Neural Network, Hardware Accelerator, Index-Matching.
