A High Performance Multi-Bit-Width Booth Vector Systolic Accelerator for NAS Optimized Deep Learning Neural Networks
- 10 June 2022
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 69 (9) , 3619-3631
Abstract
Multi-bit-width convolutional neural network (CNN) maintains the balance between network accuracy and hardware efficiency, thus enlightening a promising method for accurate yet energy-efficient edge computing. In this work, we develop state-of-the-art multi-bit-width accelerator for NAS Optimized deep learning neural networks. To efficiently process the multi-bit-width network inferencing, multi-level optimizations have been proposed. Firstly, differential Neural Architecture Search (NAS) method is adopted for the high accuracy multi-bit-width network generation. Secondly, hybrid Booth based multi-bit-width multiply-add-accumulation (MAC) unit is developed for data processing. Thirdly, vector systolic array is proposed for effectively accelerating the matrix multiplications. With vector-style systolic dataflow, both the processing time and logic resources consumption can be reduced when compared with the classical systolic array. Finally, The proposed multi-bit-width CNN acceleration scheme has been practically deployed on FPGA platform of Xilinx ZCU102. Average performance on accelerating the full NAS optimized VGG16 network is 784.2 GOPS, and peek performance of the convolutional layer can reach as high as 871.26 GOPS for INT8, 1676.96 GOPS for INT4, and 2863.29 GOPS for INT2 respectively, which is among the best results in previous CNN accelerator benchmarks.Keywords
Funding Information
- National Natural Science Foundation of China
- Innovative Team Program of Education Department of Guangdong Province (2018KCXTD028)
- Key-Area Research and Development Program of Guangdong Province (2019B010142001)
- Shenzhen Science and Technology Program (KQTD20200820113051096, JCYJ20200109115210307)
This publication has 37 references indexed in Scilit:
- Accelerating Binarized Convolutional Neural Networks with Software-Programmable FPGAs2017
- Optimizing Loop Operation and Dataflow in FPGA Acceleration of Deep Convolutional Neural Networks2017
- Energy-Efficient CNN Implementation on a Deeply Pipelined FPGA Cluster2016
- Fully Convolutional Networks for Semantic SegmentationIEEE Transactions on Pattern Analysis and Machine Intelligence, 2016
- ImageNet Large Scale Visual Recognition ChallengeInternational Journal of Computer Vision, 2015
- Optimizing FPGA-based Accelerator Design for Deep Convolutional Neural Networks2015
- DaDianNao: A Machine-Learning Supercomputer2014
- Deep Neural Networks for Acoustic Modeling in Speech Recognition: The Shared Views of Four Research GroupsIEEE Signal Processing Magazine, 2012
- A serial-parallel architecture for two-dimensional discrete cosine and inverse discrete cosine transformsIEEE Transactions on Computers, 2000
- Systolic matrix and vector multiplication methods for signal processingIEE Proceedings F Communications, Radar and Signal Processing, 1984