Tri-state spinning electronic device, storage unit, storage array and read-write circuit

Abstract

The invention provides a tri-state spinning electronic device, a storage unit, a storage array and a read-write circuit, and the tri-state spinning electronic device comprises a bottom electrode, a magnetic tunnel junction and a top electrode from bottom to top. The magnetic tunnel junction comprises a spin-orbit coupling layer, a ferromagnetic free layer, a barrier tunneling layer, a ferromagnetic reference layer, three local magnetic domain wall pinning centers and a magnetic domain wall nucleation center. For modulating an antisymmetric exchange effect, the magnetic domain wall pinning center is embedded into a heavy metal and ferromagnetic free layer interface; the magnetic domain wall nucleation centers are arranged at the two ends of the ferromagnetic free layer; current pulse flows through the spin-orbit coupling layer to generate spin current to be injected into the ferromagnetic free layer, a spin-orbit torque effective field drives a domain wall to move and displace under the regulation and control of a full electric field, the displacement can be modulated through the pulse number, pulse width and direction of the current, and the CMOS process compatibility and high reliability are achieved. The invention also provides a tri-state read-write circuit and a tri-valued network computing application scheme thereof, and high-performance GXNOR operation of a three-valued spinning electronic device is realized.

Description

technical field [0001] The present disclosure relates to the field of integrated circuits, in particular to a tri-state spintronic device, a memory unit, a memory array, and a read-write circuit. Background technique [0002] Due to its unique advantages in image recognition, semantic recognition, and classification tasks, the neural network architecture has set off a research boom in academia and industry. However, the training and recognition process of the traditional convolutional neural network requires a large number of floating-point and double-precision convolution operations. The multiplication and addition operation used in the product operation puts forward high requirements on the energy consumption and duration of the operation, making the training of the neural network take several days or even weeks. [0003] In order to optimize the problems of the traditional convolutional neural network, simplify the operation steps and ensure the accuracy of training and ...

AI Technical Summary

Problems solved by technology

Compared with BNN, TNN has higher information capacity and does not increase the complexity of operation, so it

Images