skip to main content


This content will become publicly available on June 7, 2025

Title: Deep random forest with ferroelectric analog content addressable memory

Deep random forest (DRF), which combines deep learning and random forest, exhibits comparable accuracy, interpretability, low memory and computational overhead to deep neural networks (DNNs) in edge intelligence tasks. However, efficient DRF accelerator is lagging behind its DNN counterparts. The key to DRF acceleration lies in realizing the branch-split operation at decision nodes. In this work, we propose implementing DRF through associative searches realized with ferroelectric analog content addressable memory (ACAM). Utilizing only two ferroelectric field effect transistors (FeFETs), the ultra-compact ACAM cell performs energy-efficient branch-split operations by storing decision boundaries as analog polarization states in FeFETs. The DRF accelerator architecture and its model mapping to ACAM arrays are presented. The functionality, characteristics, and scalability of the FeFET ACAM DRF and its robustness against FeFET device non-idealities are validated in experiments and simulations. Evaluations show that the FeFET ACAM DRF accelerator achieves ∼106×/10× and ∼106×/2.5× improvements in energy and latency, respectively, compared to other DRF hardware implementations on state-of-the-art CPU/ReRAM.

 
more » « less
Award ID(s):
2312517 2321840 2319198 2127780
NSF-PAR ID:
10531048
Author(s) / Creator(s):
; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ;
Publisher / Repository:
Science Advances
Date Published:
Journal Name:
Science Advances
Volume:
10
Issue:
23
ISSN:
2375-2548
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract

    Ferroelectrics offer a promising material platform to realize energy-efficient non-volatile memory technology with the FeFET-based implementations being one of the most area-efficient ferroelectric memory architectures. However, the FeFET operation entails a fundamental trade-off between the read and the program operations. To overcome this trade-off, we propose in this work, a novel device concept, Mott-FeFET, that aims to replace the Silicon channel of the FeFET with VO2- a material that exhibits an electrically driven insulator–metal phase transition. The Mott-FeFET design, which demonstrates a (ferroelectric) polarization-dependent threshold voltage, enables the read current distinguishability (i.e., the ratio of current sensed when the Mott-FeFET is in state 1 and 0, respectively) to be independent of the program voltage. This enables the device to be programmed at low voltages without affecting the ability to sense/read the state of the device. Our work provides a pathway to realize low-voltage and energy-efficient non-volatile memory solutions.

     
    more » « less
  2. Ferroelectric field‐effect transistors (FeFETs) employing graphene on inorganic perovskite substrates receive considerable attention due to their interesting electronic and memory properties. They are known to exhibit an unusual hysteresis of electronic transport that is not consistent with the ferroelectric polarization hysteresis and is previously shown to be associated with charge trapping at graphene–ferroelectric interface. Here, an electrical measurement scheme that minimizes the effect of charge traps and reveals the polarization‐dependent hysteresis of electronic transport in graphene–Pb(Zr,Ti)O3FeFETs is demonstrated. Observation of the polarization‐dependent conductivity hysteresis is important for the fundamental understanding of the interplay between the ferroelectric polarization and charge carriers in graphene. It is also important for practical memory applications because this hysteresis emulates the operation of nonvolatile memories and reveals the range of ON and OFF currents that can be achieved in long term data storage. It is demonstrated that this measurement scheme can be used to optimize the memory performance of graphene–PZT FeFETs that can exhibit nonvolatile time‐independent ON/OFF ratios of over 5. The described measurement technique can potentially be used in the studies of kinetics of charge trap dissipation, polarization‐dependent properties, and memory performance of FeFET devices comprising other 2D materials and various ferroelectric substrates.

     
    more » « less
  3. This paper presents a Ferroelectric FET (FeFET) based processing-in-memory (PIM) architecture to accelerate inference of deep neural networks (DNNs). We propose a digital in-memory vector-matrix multiplication (VMM) engine design utilizing the FeFET crossbar to enables bit-parallel computation and eliminate analog-to-digital conversion in prior mixed-signal PIM designs. A dedicated hierarchical network-on-chip (H-NoC) is developed for input broadcasting and on-the-fly partial results processing, reducing the data transmission volume and latency. Simulations in 28nm CMOS technology show 115x and 6.3x higher computing efficiency (GOPs/W) over desktop GPU (Nvidia GTX 1080Ti) and ReRAM based design, respectively. 
    more » « less
  4. Field programmable gate array (FPGA) is widely used in the acceleration of deep learning applications because of its reconfigurability, flexibility, and fast time-to-market. However, conventional FPGA suffers from the trade-off between chip area and reconfiguration latency, making efficient FPGA accelerations that require switching between multiple configurations still elusive. Here, we propose a ferroelectric field-effect transistor (FeFET)–based context-switching FPGA supporting dynamic reconfiguration to break this trade-off, enabling loading of arbitrary configuration without interrupting the active configuration execution. Leveraging the intrinsic structure and nonvolatility of FeFETs, compact FPGA primitives are proposed and experimentally verified. The evaluation results show our design shows a 63.0%/74.7% reduction in a look-up table (LUT)/connection block (CB) area and 82.7%/53.6% reduction in CB/switch box power consumption with a minimal penalty in the critical path delay (9.6%). Besides, our design yields significant time savings by 78.7 and 20.3% on average for context-switching and dynamic reconfiguration applications, respectively.

     
    more » « less
  5. This study investigates the electrical characteristics observed in n-channel and p-channel ferroelectric field effect transistor (FeFET) devices fabricated through a similar process flow with 10 nm of ferroelectric hafnium zirconium oxide (HZO) as the gate dielectric. The n-FeFETs demonstrate a faster complete polarization switching compared to the p-channel counterparts. Detailed and systematic investigations using TCAD simulations reveal the role of fixed charges and interface traps at the HZO-interfacial layer (HZO/IL) interface in modulating the subthreshold characteristics of the devices. A characteristic crossover point observed in the transfer characteristics of n-channel devices is attributed with the temporary switching between ferroelectric-based operation to charge-based operation, caused by the pinning effect due to the presence of different traps. This experimental study helps understand the role of charge trapping effects in switching characteristics of n- and p-channel ferroelectric FETs.

     
    more » « less