Scaling effect of spin-torque nano-oscillators

Chao, Xiaohui; Jamali, Mahdi; Wang, Jian Ping

doi:10.1063/1.4974014

Cited by 9 publications

(4 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The DC amplifiers must deliver a power sufficient to match the threshold current of spintronic oscillators. Projections show that spintronic oscillators can be downscaled to nano-pillars of 20 nm radius, and then have a threshold current density of 10 10 A m −2 [66]. With a resistance area product RA = 10 −12 Ω m 2 , as seen in the literature [61], and radius of 20 nm, the minimum power delivered by each DC amplifier to a spintronic oscillator is 0.1 μW, which means that the minimum power consumption of the system per neuron is 0.1 μW.…”

Section: Benchmarking the Performance Of A Convolutional Neural Netwo...mentioning

confidence: 99%

“…Since spintronic oscillators and resonators can be downscaled to 20 nm [66], we can consider unit cells of area 40 × 40 nm 2 . The area of a traditional unfolded convolution implemented on a crossbar array scales with N in N c × N out N m [14,21] where N in is the number of pixels in the input images, N c is the number of channels, N out is the number of pixels in the output feature maps, and N m is the number of output feature maps.…”

Section: Benchmarking the Performance Of A Convolutional Neural Netwo...mentioning

confidence: 99%

“…Since the threshold current density is typically around 10 11 A m −2 , in the future the threshold current of these oscillators can be decreased to tens of μA with lateral size reduction to tens of nanometers. Moreover, projections show that size reduction could also reduce threshold current density down to 10 10 A m −2 [66].…”

Section: Spintronic Oscillators Simulationsmentioning

confidence: 99%

See 2 more Smart Citations

Convolutional neural networks with radio-frequency spintronic nano-devices

Leroux

Riz

Sanz-Hernández

et al. 2022

Neuromorph. Comput. Eng.

View full text Add to dashboard Cite

Convolutional neural networks [1,2] are state-of-the-art and ubiquitous in modern signal processing and machine vision. Nowadays, hardware solutions based on emerging nanodevices are designed to reduce the power consumption of these networks. This is done either by using devices that implement convolutional filters and sequentially multiply consecutive subsets of the input, or by using different sets of devices to perform the different multiplications in parallel to avoid storing intermediate computational steps in memory. Spintronics devices are promising for information processing because of the various neural and synaptic functionalities they offer. However, due to their low OFF/ON ratio, performing all the multiplications required for convolutions in a single step with a crossbar array of spintronic memories would cause sneak-path currents. Here we present an architecture where synaptic communications are based on a resonance effect. These synaptic communications thus have a frequency selectivity that prevents crosstalk caused by sneak-path currents. We first demonstrate how a chain of spintronic resonators can function as synapses and make convolutions by sequentially rectifying radio-frequency signals encoding consecutive sets of inputs. We show that a parallel implementation is possible with multiple chains of spintronic resonators. We propose two different spatial arrangements for these chains. For each of them, we explain how to tune many artificial synapses simultaneously, exploiting the synaptic weight sharing specific to convolutions. We show how information can be transmitted between convolutional layers by using spintronic oscillators as artificial microwave neurons. Finally, we simulate a network of these radio-frequency resonators and spintronic oscillators to solve the MNIST handwritten digits dataset, and obtain results comparable to software convolutional neural networks. Since it can run convolutional neural networks fully in parallel in a single step with nano devices, the architecture proposed in this paper is promising for embedded applications requiring machine vision, such as autonomous driving.

show abstract

Section: Benchmarking the Performance Of A Convolutional Neural Netwo...mentioning

confidence: 99%

Section: Benchmarking the Performance Of A Convolutional Neural Netwo...mentioning

confidence: 99%

Section: Spintronic Oscillators Simulationsmentioning

confidence: 99%

See 1 more Smart Citation

Convolutional neural networks with radio-frequency spintronic nano-devices

Leroux

Riz

Sanz-Hernández

et al. 2022

Neuromorph. Comput. Eng.

View full text Add to dashboard Cite

show abstract

“…In the context of hard disk drive (HDD) reading head, various advantages of an STO device has been pointed out, including high data-transfer rate 15,18,19 , good scalability 19 and ability to selectively read signal from multilayer media 16 . However, the development of STO readers has been hindered by their potential vulnerability to magnetic noise, which becomes more problematic as the size of the device decreases, leading to smaller free layer volumes and therefore bigger spectrum linewidths 14,20,21 . From the noise perspective, it would be beneficial for STO devices to remain relatively large.…”

Section: Introductionmentioning

confidence: 99%

Simultaneous readout of two adjacent bit tracks with a spin-torque oscillator

Chęciński

Frankowski

Stobiecki

2018

Journal of Magnetism and Magnetic Materials

View full text Add to dashboard Cite

We propose a novel setup for a spin-torque oscillator reader in magnetic hard disk drive technology. Two adjacent bit tracks are to be read simultaneously, leading to high data transfer rate and increased resilience to noise as the lateral size of the oscillator device is allowed to remain larger than the bit width. We perform micromagnetic simulations of an example system and find that the magnetization response has a clear unimodal character, which enables for detection of two bit values at the same time. We analyze the frequency of the device under the influence of two different external fields and conduct a simulation of a successful dynamic readout. We estimate the signal linewidth and signal-to-noise ratios of the setup and show that it may be potentially beneficial for magnetic readout applications.

show abstract