An 82nW 0.53pJ/SOP Clock-Free Spiking Neural Network with 40µs Latency for AIoT Wake-Up Functions Using Ultimate-Event-Driven Bionic Architecture and Computing-in-Memory Technique

📋 论文概要

该论文提出了一种基于终极事件驱动（UED）的无时钟脉冲神经网络（SNN）芯片，用于AIoT设备的唤醒功能。通过异步脉冲传播和处理，实现了82nW的超低功耗、0.53pJ/SOP的高能效和40µs的低延迟，解决了传统AIoT设备在随机稀疏事件下的高功耗和延迟问题。

💡 主要创新点

• 提出终极事件驱动（UED）架构，模仿人脑异步脉冲处理，实现无时钟操作，大幅降低动态功耗。
• 实现82nW功耗和40µs延迟的SNN芯片，适用于AIoT设备的事件驱动唤醒场景。
• 每个脉冲操作仅消耗0.53pJ，能效显著优于传统时钟同步方案。

🏷 关键词

📄 原文摘要

Peiyu Chen1, Meng Wu1, Hao Zhang1, Peng Zhou3, Jinguang Liu3, Guangyu Sun1, Jiayoon Ru1, Le Ye1,2, Ru Huang1 Peking University, Beijing, China Advanced Institute of Information Technology of Peking University, Hangzhou, China, 3 Nano Core Chip Electronic Technology, Hangzhou, China 1 2 *Equally-Credited Authors (ECAs) Human brain is a natural ultimate-event-driven (UED) system with low power and realtime response-ability, thanks to the asynchronous propagation and processing of spikes. Power dissipation and latency are major concerns in AIoT devices, usually operating in random-sparse-event (RSE) scenarios (Fig. 22.7.1, top). Being event-driven on the system level, an always-on wake-up system (WUS) detects the valid RSEs energyefficiently and intelligently, and upon detection turns on the power-hungry high-performance system (HPS). Being event-driven on the module level, a prior WUS

👥 作者与机构

Ying Liu*1, Zhixuan Wang*1,2, Wei He1, Linxiao Shen1, Yihan Zhang1,