NeuC-CIM: A 1.3pJ/SOP Neuromorphic Charge-Domain Compute-in-Memory Macro for Spiking Neural Network

This work presents NeuC-CIM, an energy efficient neuromorphic compute-in-memory macro. The proposed multi-bit charge-domain synapse and the event-triggered spiking neuron enables efficient, DC-current free operation. Moreover, the capacity extension module and neuron cyclic calibration scheme are proposed to reduce the nonlinearity and mismatch-induced accuracy loss in analog synapse and neuron circuits. The effectiveness of NeuC-CIM is verified in silicon with various neuromorphic tasks, showing a state-of-the-art <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$1.3 \text{pJ}/\text{SOP}$</tex> energy efficiency with comparable accuracy.