In-Memory Computing Paradigms by Exploiting the Intrinsic Interactions Between STT and SOT

In-memory computing (IMC) based on magnetic random access memory (MRAM) can effectively overcome the bottlenecks of the Von-Neumann architecture, thanks to its advantages of nonvolatility, high speed, and low power consumption. Especially, the toggle-spin torque (TST)-MRAM with the joint effect of spin-transfer torque (STT) and spin-orbit torque (SOT) is extremely suitable for the IMC paradigms due to the operation flexibility of a pair of currents. In this article, we propose several novel IMC paradigms based on the TST-MRAM using the intrinsic interactions between STT and SOT. The logic operation could be implemented in a compact method. Moreover, the proposed reconfigurable logic shows great efficiency in performing XOR operation. Afterward, we develop a theoretical model for the proposed IMC paradigms to facilitate the performance evaluation. Meanwhile, good agreement between analytical solutions and simulation results is demonstrated. Finally, the functionality of the proposed IMC paradigms is validated by transient simulation. Our proposal provides a novel route for designing MRAM-based IMC paradigms.