A 2.9TOPS/W Deep Convolutional Neural Network SoC in FD-SOI 28nm for Intelligent Embedded Systems

📋 论文概要

该论文介绍了一款基于28nm FD-SOI工艺的深度卷积神经网络SoC，用于智能嵌入式系统，实现了2.9TOPS/W的高能效。解决了在嵌入式平台上高效运行DCNN的功耗和性能瓶颈问题。

💡 主要创新点

• 提出了一种针对DCNN优化的SoC架构，通过数据流和计算单元的高效协同设计实现高能效。
• 利用FD-SOI工艺的背偏置技术，动态调整阈值电压以降低功耗。
• 集成了专用加速器和可编程处理器，支持多种网络结构并提升计算密度。

🏷 关键词

📄 原文摘要

Elio Guidetti1, Fabio De Ambroggi4, Tommaso Majo1, Paolo Zambotti4, Manuj Ayodhyawasi2, Harvinder Singh2, Nalin Aggarwal2 STMicroelectronics, Cornaredo, Italy STMicroelectronics, Greater Noida, India 3 STMicroelectronics, Geneva, Switzerland 4 STMicroelectronics, Agrate Brianza, Italy 1 2 A booming number of computer vision, speech recognition, and signal processing applications, are increasingly benefiting from the use of deep convolutional neural networks (DCNN) stemming from the seminal work of Y. LeCun et al. [1] and others that led to winning the 2012 ImageNet Large Scale Visual Recognition Challenge with AlexNet [2], a DCNN significantly outperforming classical approaches for the first time. In order to deploy these technologies in mobile and wearable devices, hardware acceleration plays a critical role for real-time operation with very limited power consumption and with embedded memory overcoming

👥 作者与机构

Giuseppe Desoli1, Nitin Chawla2, Thomas Boesch3, Surinder-pal Singh2,