Abstract: Computing-in-memory (CIM) has been proven to achieve high energy efficiency and significant acceleration effects on neural networks with high computational parallelism. Based on typical ...