A 28-nm 25.1 TOPS/W Sparsity-Aware CNN-GCN Deep Learning SoC for Mobile Augmented Reality

Huang, Wen CongWen CongHuangLin, I. TingI. TingLinChen, Wen ChingWen ChingChenLin, Liang YiLiang YiLinChang, Nian ShyangNian ShyangChangCHUN-PIN LINChen, Chi ShiChi ShiChenCHIA-HSIANG YANG2023-05-222023-05-222022-01-01978166549772507431562https://scholars.lib.ntu.edu.tw/handle/123456789/631209This work presents the first CNN-GCN SoC for diverse AI vision computations on mobile augmented reality (AR). A CNN engine utilizes the channel-wise feature sparsity with a specialized processing element to achieve an up to 8× higher throughput and 6.1× energy efficiency. A GCN engine is implemented for graph-based action recognition. The computational complexity and memory usage are minimized by lever-aging matrix and graph properties. The proposed SoC achieves 25.1 TOPS/W energy efficiency for CNN inference, outperforming prior designs by 2.0×. It delivers 72 action/s on action recognition, exceeding prior art by 18× in latency.[SDGs]SDG7A 28-nm 25.1 TOPS/W Sparsity-Aware CNN-GCN Deep Learning SoC for Mobile Augmented Realityconference paper10.1109/VLSITechnologyandCir46769.2022.98302612-s2.0-85135216426https://api.elsevier.com/content/abstract/scopus_id/85135216426