Nonvolatile Spintronic Memory Cells for Neural Networks

Abstract

A new spintronic nonvolatile memory cell analogous to 1T DRAM with non-destructive READ is proposed. The cells can be used as neural computing units. A dual-circuit neural network architecture is proposed to leverage these devices against the complex operations involved in convolutional networks. Simulations based on HSPICE and MATLAB were performed to study the performance of this architecture when classifying images as well as the effect of varying the size and stability of the nanomagnets. The spintronic cells outperform a purely charge-based implementation of the same network, consuming ≈ 100-pJ total energy per image processed.

SHE-MTJ Circuits for Convolutional Neural Networks

“…Each MAAP set operation takes on the order of 1 ns, saving a great deal of time if we assume identical delays in the ADC and digital processing peripherals between the two networks. Finally, comparing the number of MAAP operations N M = 3904 to the number of CeNN operations in an equivalent network in [2], [18] N C ∼ 56000 shows a significant reduction in complexity, even with redundancy. We also note that condensing the neuromorphic layer operations reduces the number of ADC and memory operations that are required to store and access intermediate data compared to a system which explicitly computes each operation, especially those which compute the operations via multiple sub-steps which themselves may require digital processing.…”

Spin Boltzmann machine

“…In [10] nanoscale spintronic oscillators were used to implement the basic block of ANN combining nonlinearity and memory, a successful recognition of spoken digits is reported. The inverse Rashba-Edelstein magnetoelectric neuron was proposed in [11], its effectiveness was demonstrated on handwritten images recognition. 784×200×10 deep belief network consisting of p-bit-based neurons was created using spin-orbit-torque magnetic random access memory in [12].…”