Automatic Generation of Multi-Precision Multi-Arithmetic CNN Accelerators for FPGAs
Authors:
- 1 December 2019
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 45-53
Abstract
Modern deep Convolutional Neural Networks (CNNs) are computationally demanding, yet real applications often require high throughput and low latency. To help tackle these problems, we propose Tomato, a framework designed to automate the process of generating efficient CNN accelerators. The generated design is pipelined and each convolution layer uses different arithmetics at various precisions. Using Tomato, we showcase state-of-the-art multi-precision multi-arithmetic networks, including MobileNet-V1, running on FPGAs. To our knowledge, this is the first multi-precision multi-arithmetic autogeneration framework for CNNs. In software, Tomato fine-tunes pretrained networks to use a mixture of short powers-of-2 and fixed-point weights with a minimal loss in classification accuracy. The fine-tuned parameters are combined with the templated hardware designs to automatically produce efficient inference circuits in FPGAs. We demonstrate how our approach significantly reduces model sizes and computation complexities, and permits us to pack a complete ImageNet network onto a single FPGA without accessing off-chip memories for the first time. Furthermore, we show how Tomato produces implementations of networks with various sizes running on single or multiple FPGAs. To the best of our knowledge, our automatically generated accelerators outperform closest FPGA-based competitors by at least 2-4× for lantency and throughput; the generated accelerator runs ImageNet classification at a rate of more than 3000 frames per second.Keywords
All Related Versions
This publication has 25 references indexed in Scilit:
- ImageNet classification with deep convolutional neural networksCommunications of the ACM, 2017
- Fused-layer CNN accelerators2016
- Angel-Eye: A Complete Design Flow for Mapping CNN onto Customized Hardware2016
- fpgaConvNet: A Framework for Mapping Convolutional Neural Networks on FPGAs2016
- Energy-Efficient CNN Implementation on a Deeply Pipelined FPGA Cluster2016
- Going deeper in facial expression recognition using deep neural networks2016
- Going Deeper with Embedded FPGA Platform for Convolutional Neural Network2016
- Throughput-Optimized OpenCL-based FPGA Accelerator for Large-Scale Convolutional Neural Networks2016
- Fixed-point feedforward deep neural network design using weights +1, 0, and −12014
- Learning sets of filters using back-propagationComputer Speech & Language, 1987