Probabilistic-Bits based on Ferroelectric Field-Effect Transistors for Stochastic Computing

Abstract

A probabilistic-bit (p-bit) is the fundamental building block in the circuit network of a stochastic computing, and it could produce a continuous random bit-stream with tunable probability. Utilizing the stochasticity in few-domain ferroelectric material(FE), we propose for the first time, the p-bits based on ferroelectric FET. The stochasticity of the FE p-bits stems from the thermal noise-induced lattice vibration, which renders dipole fluctuations and is tunable by an external electric field. The impact of several key FE parameters on p-bits' stochasticity is evaluated, where the domain properties are revealed to play crucial roles. Furthermore, the integer factorization based on FE p-bits circuit network is performed to verify its functionality, and the accuracy is found to depend on FE p-bits' stochasticity. The proposed FE p-bits possess the advantages of both extremely low hardware coast and the compatibility with CMOS-technology, rendering it a promising candidate for stochastic computing applications.

23 pages, 7 figures and supplementary materials with 3 notes

References in corpus (5)

• Massively Parallel Probabilistic Computing with Sparse Ising Machines
• Nonlinear Dynamics of Domain Wall Propagation in Epitaxial Ferroelectric Thin Films
• A full-stack view of probabilistic computing with p-bits: devices, architectures and algorithms
• Hardware-aware $insitu$ Boltzmann machine learning using stochastic magnetic tunnel junctions
• Electron-Phonon Scattering in Planar MOSFETs: NEGF and Monte Carlo Methods

Cited by in corpus (7)

• CMOS + stochastic nanomagnets: heterogeneous computers for probabilistic inference and learning
• pc-COP: An Efficient and Configurable 2048-p-Bit Fully-Connected Probabilistic Computing Accelerator for Combinatorial Optimization
• The First Hardware Demonstration of a Universal Programmable RRAM-based Probabilistic Computer for Molecular Docking
• Metrics for spin-based computing
• Ground-State Probabilistic Logic with the Simplest Binary Energy Landscape for Probabilistic Computing
• Sub-universal variational circuits for combinatorial optimization problems
• ThRIve: Thermally Robust CNN Inference via Low-Rank Adaptation in Heterogeneous PIM Architectures