JSSC 2022第8期Memory65nmCIM

STICKER-IM A 65 nm Computing-in-Memory NN Processor Using Block-Wise Sparsity Op

STICKER-IM是一款基于65nm工艺的存内计算神经网络处理器，采用块稀疏架构提升能效。

5.8–158 TOPS/W 平均系统能效

存内计算神经网络处理器块稀疏能效优化65nm CMOS

▸块稀疏架构支持激活和权重稀疏优化

▸自适应核/通道映射与调度策略

▸优化的存内计算宏单元与自适应ADC

Abstract

Computing-in-memory (CIM) is a promising architecture for energy-efﬁcient neural network (NN) proces- sors. Several CIM macros have demonstrated high energy efﬁciency, while CIM-based system-on-a-chip is not well explored. This work presents a CIM NN processor, named STICKER-IM, which is implemented with sophisticated sys- tem integration. Three key innovations are proposed. First, a CIM-friendly block-wise sparsity (BWS) architecture is designed, enabling both activation-sparsity-aware accelera