Exploring Efficient Acceleration Architecture for Winograd-Transformed Transposed Convolution of GANs on FPGAs

Author:

Di XinkaiORCID,Yang Hai-GangORCID,Jia Yiping,Huang Zhihong,Mao Ning

Abstract

The acceleration architecture of transposed convolution layers is essential since transposed convolution operations, as critical components in the generative model of generative adversarial networks, are computationally intensive inherently. In addition, the pre-processing of inserting and padding with zeros for input feature maps causes many ineffective operations. Most of the already known FPGA (Field Programmable Gate Array) based architectures for convolution layers cannot tackle these issues. In this paper, we firstly propose a novel dataflow exploration through splitting the filters and its corresponding input feature maps into four sets and then applying the Winograd algorithm for fast processing with a high efficiency. Secondly, we present an underlying FPGA-based accelerator architecture that features owning processing units, with embedded parallel, pipelined, and buffered processing flow. At last, a parallelism-aware memory partition technique and the hardware-based design space are explored coordinating, respectively, for the required parallel operations and optimal design parameters. Experiments of several state-of-the-art GANs by our methods achieve an average performance of 639.2 GOPS on Xilinx ZCU102 and 162.5 GOPS on Xilinx VC706. In reference to a conventional optimized accelerator baseline, this work demonstrates an 8.6× (up to 11.7×) increase in processing performance, compared to below 2.2× improvement by the prior studies in the literature.

Funder

National Natural Science Foundation of China

Publisher

MDPI AG

Subject

Electrical and Electronic Engineering,Computer Networks and Communications,Hardware and Architecture,Signal Processing,Control and Systems Engineering

Reference29 articles.

1. Generative Adversarial Networks;Goodfellow;Adv. Neural Inf. Process. Syst.,2014

2. Conditional Generative Adversarial Nets;Mirza;arXiv,2014

3. Unsupervised representation learning with deep convolutional generative adversarial networks;Radford;arXiv,2015

Cited by 15 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. Energy-Efficient Implementation of YOLOv8, Instance Segmentation, and Pose Detection on RISC-V SoC;IEEE Access;2024

2. A Reconfigurable GC-CycleGAN Based on Convolutional Optimization;2023 6th International Conference on Artificial Intelligence and Pattern Recognition (AIPR);2023-09-22

3. FPGA Design of Transposed Convolutions for Deep Learning Using High-Level Synthesis;Journal of Signal Processing Systems;2023-08-04

4. FlexCNN: An End-to-end Framework for Composing CNN Accelerators on FPGA;ACM Transactions on Reconfigurable Technology and Systems;2023-03-11

5. A Review of the Optimal Design of Neural Networks Based on FPGA;Applied Sciences;2022-10-24

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3