Design of MTJ-Based nonvolatile logic gates for quantized neural networks

Logic gates using magnetic tunnel junction (MTJ)-based nonvolatile logic-in-memory (NV-LIM) architecture are designed for quantized neural networks (QNNs) for Internet-of-Things applications. The NV-LIM-based implementation reduces data transfer costs between storage and logic gate components, thereby greatly enhancing the energy efficiency of inference operations in QNNs. The impact of the proposed nonvolatile logic gates for binary and ternary neural networks on energy consumption, delay, and area overhead reduction is demonstrated through circuit evaluations based on the parameters of the measured MTJ devices.