A 112-765 GOPS/W FPGA-based CNN Accelerator using Importance Map Guided Adaptive Activation Sparsification for Pix2pix Applications
Autor: | Huazhong Yang, Wenyu Sun, Chen Tang, Zhuqing Yuan, Zhe Yuan, Yongpan Liu |
---|---|
Rok vydání: | 2020 |
Předmět: |
business.industry
Computer science Importance map 02 engineering and technology 010501 environmental sciences 01 natural sciences Power (physics) Acceleration External data Transmission (telecommunications) Power consumption 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing business Field-programmable gate array Computer hardware 0105 earth and related environmental sciences Efficient energy use |
Zdroj: | A-SSCC |
Popis: | This paper proposes an algorithm and hardware co-design methodology to accelerate CNNs for pix2pix tasks. An importance map is introduced to train an activation-sparse CNN model, which can effectively reduce the computing cost and external data transmission. Moreover, the model also supports sparse controlling by means of the importance map, making it adaptive for applications with different precision/power requirements. An FPGA-based accelerator with adaptive sparse controlling is designed to support such importance map guided activation sparsity, and demonstrated for super-resolution (SR) application as an example. The accelerator shows advantages in both model accuracy and power consumption. It achieves up to 765 GOPS/W energy efficiency, which is 5.28× batter than the previous FPGA-based SR accelerator. |
Databáze: | OpenAIRE |
Externí odkaz: |