Specializing CGRAs for Light-Weight Convolutional Neural Networks
Authors:
- 26 October 2021
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
- Vol. 41 (10) , 3387-3399
Abstract
Deep neural network (DNN) processing units, or DPUs, are one of the most energy-efficient platforms for DNN applications. However, designing new DPUs for every DNN model is very costly and time consuming. In this article, we propose an alternative approach: to specialize coarse-grained reconfigurable architectures (CGRAs), which are already quite capable of delivering high performance and high energy efficiency for compute-intensive kernels. We identify a small set of architectural features on a baseline CGRA to enable high-performance mapping of depthwise convolution (DWC) and pointwise convolution (PWC) kernels, which are the most important building block in recent light-weight DNN models. Our experimental results using MobileNets demonstrate that our proposed CGRA enhancement can deliver $8\sim 18\times $ improvement in area-delay product (ADP) depending on layer type, over a baseline CGRA with a state-of-the-art CGRA compiler. Moreover, our proposed CGRA architecture can also speed up 3-D convolution with similar efficiency as previous work, demonstrating the effectiveness of our architectural features beyond depthwise separable convolution (DSC) layers.Keywords
Funding Information
- IITP through the Artificial Intelligence Graduate School Program (2020-0-01336)
- Neuromorphic Computing Software Platform for Artificial Intelligence Systems (1711080972)
- ITRC support program (IITP-2021-0-02052)
- NRF (2020R1A2C2015066)
- MSIT of Korea
- Free Innovative Research Fund of UNIST (1.170067.01)
- IC Design Education Center (IDEC), South Korea
This publication has 33 references indexed in Scilit:
- Eyeriss: An Energy-Efficient Reconfigurable Accelerator for Deep Convolutional Neural NetworksIEEE Journal of Solid-State Circuits, 2016
- A CGRA-Based Approach for Accelerating Convolutional Neural Networks2015
- Optimizing stream program performance on CGRA-based systems2015
- Flattening-based mapping of imperfect loop nests for CGRAs2014
- Graph Minor Approach for Application Mapping on CGRAsACM Transactions on Reconfigurable Technology and Systems, 2014
- Evaluator-executor transformation for efficient pipelining of loops with conditionalsACM Transactions on Architecture and Code Optimization, 2013
- Improving performance of nested loops on reconfigurable array processorsACM Transactions on Architecture and Code Optimization, 2012
- Memory access optimization in compilation for coarse-grained reconfigurable architecturesACM Transactions on Design Automation of Electronic Systems, 2011
- ADRES & DRESC: Architecture and Compiler for Coarse-Grain Reconfigurable Processors2007
- Exploiting loop-level parallelism on coarse-grained reconfigurable architectures using modulo schedulingIEE Proceedings - Computers and Digital Techniques, 2003