Abstract:Artificial intelligence has made remarkable advancements in recent years, leading to the development of algorithms and models capable of handling ever-increasing amounts of data. The computational demands of these algorithms necessitate circuit and architecture designs that go beyond the von-Neumann paradigm. Content addressable memories (CAMs), which implement parallel associative search functionality within memory blocks to overcome the memory wall bottleneck, have proven to be effective for data-intensive tasks. While current CAM designs have achieved higher storage density and energy efficiency than their CMOS-based counterparts by leveraging emerging non-volatile memories (NVM), most of these implementations are limited to binary storage cells. In this work, we propose SEE-MCAM, scalable and compact multi-bit CAM (MCAM) designs that utilize the three-terminal ferroelectric FET (FeFET) as the proxy. By exploiting the multi-level-cell characteristics of FeFETs, our proposed SEE-MCAM designs enable multi-bit associative search functions and achieve better energy efficiency and performance than existing FeFET-based CAM designs. We validated the functionality of our proposed designs by achieving 3 bits per cell CAM functionality, resulting in 3x improvement in storage density. The area per bit of the proposed SEE-MCAM cell is 8% of the conventional CMOS CAM. We thoroughly investigated the scalability and robustness of the proposed design. Evaluation results suggest that the proposed 2FeFET-1T SEE-MCAM achieves 9.8x more energy efficiency and 1.6x less search latency compared to the CMOS CAM, respectively. When compared to existing MCAM designs, the proposed SEE-MCAM can achieve 8.7x and 4.9x more energy efficiency than ReRAM-based and FeFET-based MCAMs, respectively. Benchmarking results show that our approach provides up to 3 orders of magnitude improvement in speedup and energy efficiency over a GPU implementation in accelerating a novel quantized hyperdimensional computing (HDC) application.

An innovative sensing architecture for multilevel Flash memory

Analog Content Addressable Memory using Ferroelectric: A Case Study of Search-in-Memory

Adaptive Energy-Aware Design Of A Multi-Bank Flash-Memory Storage System

Implementation of Data Search in Multi-Level NAND Flash Memory by Complementary Storage Scheme

Novel Vertical Channel Double Gate Structures for High Density and Low Power Flash Memory Applications.

Novel SONOS and Floating-Gate Flash Memory Cell Structures for High Density and High Reliability Applications

A 1.35-V 16-Mb Twin-Bit-Cell Virtual-Ground-Architecture Embedded Flash Memory with A Sensing Current Protection Technique

Novel Silicon-Based Flash Cell Structures for Low Power and High Density Memory Applications

A 512 Kb SRAM in 65nm CMOS with Divided Bitline and Novel Two-Stage Sensing Technique

SEE-MCAM: Scalable Multi-bit FeFET Content Addressable Memories for Energy Efficient Associative Search

A High Speed Low Power Negative Sensing Architecture for 3d Nand Flash Memory

Adaptive Circuit Approaches to Low-Power Multi-Level/Cell FeFET Memory

PVSensing: A Process-Variation-Aware Space Allocation Strategy for 3D NAND Flash Memory

A 280-Kbytes Twin-Bit-Cell Embedded NOR Flash Memory with a Novel Sensing Current Protection Enhanced Technique and High-Voltage Generating Systems

A novel flash memory cell and design optimization for high density and low power application

Multilevel Storage Implementation Based on Multiple Stable Resistance States for High Density Phase Change Memory Application

Lowering Latency of Embedded Memory by Exploiting In-Cell Victim Cache Hierarchy Based on Emerging Multi-Level Memory Devices

A Novel Group-Division-Multiplexing Complementary Symmetric Reference Sensing Scheme for MRAM

Design of Reading Circuit for High-Reliability 55-Nm Split-Gate SuperFlash Technology

TCAD Simulation Studies on Ultra-Low-Power Non-Volatile Memory

A Scalable Integration Process for Ultrafast Two-Dimensional Flash Memory