[paper] STT-MTJ Device Model

Haoyan Liu and Takashi Ohsawa
General-Purpose STT-MTJ Device Model Based on the Fokker-Planck Equation
IEEE Transactions On Nanotechnology, VOL. 22, 2023 659 A
DOI: 10.1109/TNANO.2023.3322468.

Graduate School of Information, Production and Systems, Waseda University (J)

Abstract: A thermally agitated device model of spin-transfer torque magnetic tunnel junction (STT-MTJ) based on the Fokker-Planck equation is proposed which is implemented into HSPICE by using Verilog-A. We compared different techniques of finite difference method (FDM) and analyzed the impact of the solvers on computational efficiency and accuracy. A framework is proposed which traces dynamics of a particular STT-MTJ’s angle between the magnetic moments of the free and the pinned layers and makes the model applicable to a wide range of circuits. The model was applied to the 4T2MTJ memory cell array and a leaky integrate and-fire (LIF) neuron circuit to validate the stochastic switching characteristic and the angle prediction function. In the memory array simulations, the CPU time consumption for this model is 1/30 of the model which is based on the stochastic Landau-Lifshitz Gilbert-Slonczewski equation.

Fig: (a) Structure of 1T1MTJ synapse. (b) Binary weights in 10 neurons and an input digit ‘9’ of spiking neural network (surrounded by the dotted square) used for the experiment shown. Each digit is a 28×28 matrix. Each figure shows two output spikes fired in the neurons representing ‘0-9’. The total spike numbers of the neurons which represent 0-9 are 2, 3, 4, 3, 4, 4, 4, 4, 4 and 9. 

Acknowledgement: This work was supported in part by Synopsys Corporation, in part by JSPS KAKENHI under Grant JP20K04626, in part by VLSI Design and Education Center (VDEC), University of Tokyo with collaboration with Cadence Corporation, and in part by the cooperation of organization between Kioxia Corporation and Waseda University.

[paper] OpenSpike: An OpenRAM SNN Accelerator

Farhad Modaresi¹, Matthew Guthaus², and Jason K. Eshraghian³
OpenSpike: An OpenRAM SNN Accelerator
arXiv:2302.01015v1 [cs.AR] 2 Feb 2023

1) Dept. of Electrical Engineering Allameh Mohaddes Nouri University Nur, Mazandaran, Iran
2) Dept. of Computer Science and Engineering, UC Santa Cruz Santa Cruz, CA, United States
3) Dept. of Electrical and Computer Engineering, UC Santa Cruz Santa Cruz, CA, United States

Abstract: This paper presents a spiking neural network (SNN) accelerator made using fully open-source EDA tools, process design kit (PDK), and memory macros synthesized using Open- RAM. The chip is taped out in the 130 nm SkyWater process and integrates over 1 million synaptic weights, and offers a reprogrammable architecture. It operates at a clock speed of 40 MHz, a supply of 1.8 V, uses a PicoRV32 core for control, and occupies an area of 33.3 mm2. The throughput of the accelerator is 48,262 images per second with a wallclock time of 20.72 μs, at 56.8 GOPS/W. The spiking neurons use hysteresis to provide an adaptive threshold (i.e., a Schmitt trigger) which can reduce state instability. This results in high performing SNNs across a range of benchmarks that remain competitive with state-of-the-art, full precision SNNs.

The design is open sourced and available online: https://github.com/sfmth/OpenSpike

Fig: OpenSpike core - system architecture and data flow

 

 

[paper] TCAD modeling of neuromorphic systems based on ferroelectric tunnel junctions

Yu He, Wei-Choon Ng and Lee Smith

TCAD modeling of neuromorphic systems based on ferroelectric tunnel junctions

J Comput Electron (2020)

DOI: 10.1007/s10825-020-01544-z

Abstract: A new compact model for HfO2-based ferroelectric tunnel junction (FTJ) memristors is constructed based on detailed physical modeling using calibrated TCAD simulations. A multi-domain configuration of the ferroelectric material is demonstrated to produce quasi-continuous conductance of the FTJ. This behavior is demonstrated to enable a robust spike-timing-dependent plasticity-type learning capability, making FTJs suitable for use as synaptic memristors in a spiking neural network. Using both TCAD–SPICE mixed-mode and pure SPICE compact model approaches, we apply the newly developed model to a crossbar array configuration in a handwritten digit recognition neuromorphic system and demonstrate an 80% successful recognition rate. The applied methodology demonstrates the use of TCAD to help develop and calibrate SPICE models in the study of neuromorphic systems.

Fig: Electric field–polarization relationship. Solid line: multi-domain simulation; dashed line: single-domain simulation; dot: measurement