A Coupled CMOS Oscillator Array for 8ns and 55pJ Inference in Convolutional Neural Networks
A Coupled CMOS Oscillator Array for 8ns and 55pJ Inference in Convolutional Neural Networks
复制标题
用于卷积神经网络中 8ns 和 55pJ 推理的耦合 CMOS 振荡器阵列
DOI：
--
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
I. Young
中科院分区：
文献类型：
--
作者：
D. Nikonov;P. Kurahashi;J. Ayers;Hyung;Y. Fan;I. Young
设计和制造的基于 26 个 CMOS 环形振荡器阵列的振荡器神经网络 (ONN)。 ONN 用于推理卷积神经网络所需的图像片段和内核的点积。使用当前 DAC 将输入编码为振荡器的频移。匹配度 (DOM) 由振荡器同步确定。测量结果表明 DOM 和点积具有高度相关性。推理需要8ns的时间和55pJ的能量。
Oscillator neural networks (ONN) based on arrays of 26 CMOS ring oscillators designed and fabricated. ONN are used for inference of dot products with image fragments and kernels necessary for convolutional neural networks. The inputs are encoded as frequency shifts of oscillators using current DACs. Degree of match (DOM) is determined from oscillators synchronization. Measurements demonstrate high correlation of DOM and dot products. Inference requires the time of 8ns and energy of 55pJ.