A near-threshold memristive computing-in-memory engine for edge intelligence

作者：Linfang Wang, Weizeng Li, Zhidao Zhou, Junjie An, Wang Ye, Zhi Li, Hanghang Gao, Hongyang Hu, Jing Liu, Xiaoming Chen, Ling Li, Qi Liu, Mingoo Seok, Chunmeng Dou, Ming Liu · 发表于：Nature Communications · 年份：2025 · DOI：10.1038/s41467-025-61025-4 · 被引用次数：22 · 研究领域：Advanced Memory and Neural Computing、Ferroelectric and Negative Capacitance Devices、Neuroscience and Neural Engineering

Memristive computing-in-memory and near-threshold computing are two unconventional computing paradigms that can potentially enhance the energy efficiency and real-time performance of edge devices. However, their scalability faces challenges, primarily due to process variation. Here, we report a 1-Mb, 16-macro near-threshold memristive computing-in-memory engine. The two-transistor-one-resistor cells provide strong cell current modulation capability with more than 120-times amplified resistance ratio. To mitigate variation issues, we compensate for transistor mismatches by leveraging the intrinsic variations in memristors. Additionally, we propose a charge stacking technique between multiple analog-to-digital converters to perform analog weight-and-combine operations with small energy and area overhead. Moreover, we introduce an inter-macro hybrid control scheme to reduce the task-level inference power. The fabricated chip can perform highly parallel analog computing over 256 input channels with a 2.4% relative standard deviation. It achieves a throughput up to 10.49 tera-operations per second and an energy efficiency up to 88.51 tera-operations per second per watt. Memristive and near-threshold computing hold potential for the next-generation edge AI hardware, but their scalability remains a bottleneck. Wang et al. combine the two paradigms to achieve parallel Mb-level computing, whereby the variation in memristors cancels out the intrinsic transistor mismatch.