Bias-scalable analog computing is attractive for implementing machine learning (ML) processors with distinct power-performance specifications. For instance, ML implementations for server workloads are focused on higher computational throughput for faster training, whereas ML implementations for edge devices are focused on energy-efficient inference. In this paper, we demonstrate the implementation of bias-scalable approximate analog computing circuits using the generalization of the margin-propagation principle called shape-based analog computing (S-AC). The resulting S-AC core integrates several near-memory compute elements, which include: (a) non-linear activation functions; (b) inner-product compute circuits; and (c) a mixed-signal compressive memory, all of which can be scaled for performance or power while preserving its functionality. Using measured results from prototypes fabricated in a 180nm CMOS process, we demonstrate that the performance of computing modules remains robust to transistor biasing and variations in temperature. In this paper, we also demonstrate the effect of bias-scalability and computational accuracy on a simple ML regression task.
A set of highly-optimized custom macro extensions is developed for a 7nm CMOS cell library for implementing Temporal Neural Networks (TNNs) that can mimic brain-like sensory processing with extreme energy efficiency. A TNN prototype (13,750 neurons and 315,000 synapses) for MNIST requires only 1.56mm2 die area and consumes only 1.69mW.
In this paper, we present an energy-efficient SNN architecture, which can seamlessly run deep spiking neural networks (SNNs) with improved accuracy. First, we propose a conversion aware training (CAT) to reduce ANN-to-SNN conversion loss without hardware implementation overhead. In the proposed CAT, the activation function developed for simulating SNN during ANN training, is efficiently exploited to reduce the data representation error after conversion. Based on the CAT technique, we also present a time-to-first-spike coding that allows lightweight logarithmic computation by utilizing spike ti…
University of South Carolina（南卡罗来纳大学）；Lehigh University（理海大学）
In this paper, we develop an in-memory analog computing (IMAC) architecture realizing both synaptic behavior and activation functions within non-volatile memory arrays. Spin-orbit torque magnetoresistive random-access memory (SOT-MRAM) devices are leveraged to realize sigmoidal neurons as well as binarized synapses. First, it is shown the proposed IMAC architecture can be utilized to realize a multilayer perceptron (MLP) classifier achieving orders of magnitude performance improvement compared to previous mixed-signal and digital implementations. Next, a heterogeneous mixed-signal and mixed-pr…
TEXEL: A neuromorphic processor with on-chip learning for beyond-CMOS device integration
University of Groningen（格罗宁根大学）；Technische Universität Ilmenau（伊尔默瑙工业大学）；Forschungszentrum Jülich（于利希研究中心）；University of Zürich（苏黎世大学）；ETH Zürich（苏黎世联邦理工学院）
Artificial neural networks inspired by brain operations can improve the possibilities of solving complex problems more efficiently. Today's computing hardware, on the other hand, is mainly based on von Neumann architecture and CMOS technology, which is inefficient at implementing neural networks. For the first time, we propose an ultrahigh speed, spiking neuromorphic processor architecture built upon single flux quantum (SFQ) based artificial neurons (JJ-Neuron). Proposed architecture has the potential to provide higher performance and power efficiency over the state of the art including CMOS,…
IBM Thomas J. Watson Research Center（IBM托马斯·J·沃森研究中心）；Korea University（高丽大学）
Recently we have shown that an architecture based on resistive processing unit (RPU) devices has potential to achieve significant acceleration in deep neural network (DNN) training compared to today's software-based DNN implementations running on CPU/GPU. However, currently available device candidates based on non-volatile memory technologies do not satisfy all the requirements to realize the RPU concept. Here, we propose an analog CMOS-based RPU design (CMOS RPU) which can store and process data locally and can be operated in a massively parallel manner. We analyze various properties of the C…
This paper addresses a novel five-transistor (5T) CMOS SRAM design with high performance and reliability in 65nm CMOS, and illustrates how it reduces the dynamic power consumption in comparison with the conventional and low-power 6T SRAM counterparts. This design can be used as cache memory in processors and low-power portable devices. The proposed SRAM cell features ~13% area reduction compared to a conventional 6T cell, and features a unique bit-line and negative supply voltage biasing methodology and ground control architecture to enhance performance, and suppress standby leakage power.
University of California, Los Angeles（加州大学洛杉矶分校）；University of California, Irvine（加州大学欧文分校）；National Chiao Tung University（国立交通大学）；Kneron, Inc（Kneron公司）
An analog neural network computing engine based on CMOS-compatible charge-trap transistor (CTT) is proposed in this paper. CTT devices are used as analog multipliers. Compared to digital multipliers, CTT-based analog multiplier shows significant area and power reduction. The proposed computing engine is composed of a scalable CTT multiplier array and energy efficient analog-digital interfaces. Through implementing the sequential analog fabric (SAF), the engine mixed-signal interfaces are simplified and hardware overhead remains constant regardless of the size of the array. A proof-of-concept 7…
Shifting computing architectures from von Neumann to event-based spiking neural networks (SNNs) uncovers new opportunities for low-power processing of sensory data in applications such as vision or sensorimotor control. Exploring roads toward cognitive SNNs requires the design of compact, low-power and versatile experimentation platforms with the key requirement of online learning in order to adapt and learn new features in uncontrolled environments. However, embedding online learning in SNNs is currently hindered by high incurred complexity and area overheads. In this work, we present ODIN, a…
University of Innsbruck（因斯布鲁克大学）；University of Erlangen-Nuremberg（埃尔朗根-纽伦堡大学）
A concept for a novel CMOS image sensor suited for analog image pre-processing is presented in this paper. As an example, an image restoration algorithm for reducing image noise is applied as image pre-processing in the analog domain. To supply low-latency data input for analog image preprocessing, the proposed concept for a CMOS image sensor offers a new sensor signal acquisition method in 2D. In comparison to image pre-processing in the digital domain, the proposed analog image pre-processing promises an improved image quality. Furthermore, the image noise at the stage of analog sensor signa…
This paper proposes four quadrant analog multiplier using CMOS-memristor circuit. Currently, there are plenty of analog multipliers using resistors and CMOS transistors. They can attain perfect multiplication but have several disadvantages such as lower processing speed, higher power consumption and larger chip areas. Memristor based circuits are introduced to resolve the mentioned drawbacks. In this paper current mode four quadrant multiplier based on squaring circuits is taken as a framework, and CMOS transistors are replaced with memristors. The circuit design is simulated with SPICE, and v…
We present an integrated circuit fabricated in a process co-integrating CMOS and hafnium-oxide memristor technology, which provides a prototyping platform for projects involving memristors. Our circuit includes the periphery circuitry for using memristors within digital circuits, as well as an analog mode with direct access to memristors. The platform allows optimizing the conditions for reading and writing memristors, as well as developing and testing innovative memristor-based neuromorphic concepts.
Minor embedding heuristics have become an indispensable tool for compiling problems in quadratically unconstrained binary optimization (QUBO) into the hardware graphs of quantum and CMOS annealing processors. While recent embedding heuristics have been developed for annealers of moderate size (about 2000 nodes) the size of the latest CMOS annealing processor (with 102,400 nodes) poses entirely new demands on the embedding heuristic. This raises the question, if recent embedding heuristics can maintain meaningful embedding performance on hardware graphs of increasing size. Here, we develop an i…
Emerging nano-scale programmable Resistive-RAM (RRAM) has been identified as a promising technology for implementing brain-inspired computing hardware. Several neural network architectures, that essentially involve computation of scalar products between input data vectors and stored network weights can be efficiently implemented using high density cross-bar arrays of RRAM integrated with CMOS. In such a design, the CMOS interface may be responsible for providing input excitations and for processing the RRAM output. In order to achieve high energy efficiency along with high integration density …
Neuromorphic vision sensors (NVS) can enable energy savings due to their event-driven that exploits the temporal redundancy in video streams from a stationary camera. However, noise-driven events lead to the false triggering of the object recognition processor. Image denoise operations require memoryintensive processing leading to a bottleneck in energy and latency. In this paper, we present in-memory filtering (IMF), a 6TSRAM in-memory computing based image denoising for eventbased binary image (EBBI) frame from an NVS. We propose a non-overlap median filter (NOMF) for image denoising. An inm…
With the evolution of heterogeneous computing system, such as network-on-chip, high-performance distributed computing, accelerator-rich architectures and cluster computing, high-speed, energy-efficient and low-latency interfaces among memory-to-processor and processor-to-processor become the key technology to enable those technologies. Simultaneously, the scaling of CMOS makes the switching speed of the transistor up to sub-THz. Radio-frequency or even millimeter-wave modulated signaling interconnect has unique features in ultra-low power operation, dynamic allocation of bandwidth and low late…
Hierarchical Temporal Memory (HTM) is a neuromorphic algorithm that emulates sparsity, hierarchy and modularity resembling the working principles of neocortex. Feature encoding is an important step to create sparse binary patterns. This sparsity is introduced by the binary weights and random weight assignment in the initialization stage of the HTM. We propose the alternative deterministic method for the HTM initialization stage, which connects the HTM weights to the input data and preserves natural sparsity of the input information. Further, we introduce the hardware implementation of the dete…
CMOS-transistors circuits have been used as a conventional approach for designing an analog multiplier in modern era of industrial electronics. However, previous studies have shown, that based on the working region of transistors, such as saturation or weak inversion regions, the circuit may face issues with output ranges and accuracy. One possible solution to that problem could be choosing CMOS-memristors as a basis for the circuit. Although memristor research is still a growing and promising field, one could argue that its implementation could bring many benefits such as increased circuit de…
Thermodynamic RAM (kT-RAM) is a neuromemristive co-processor design based on the theory of AHaH Computing and implemented via CMOS and memristors. The co-processor is a 2-D array of differential memristor pairs (synapses) that can be selectively coupled together (neurons) via the digital bit addressing of the underlying CMOS RAM circuitry. The chip is designed to plug into existing digital computers and be interacted with via a simple instruction set. Anti-Hebbian and Hebbian (AHaH) computing forms the theoretical framework from which a nature-inspired type of computing architecture is built w…
In this paper, we propose an analog pixel differencing circuit for differentiating pixels between frames directly from CMOS pixels. The analog information processing at sensor is a topic of growing appeal to develop edge AI devices. The proposed circuit is integrated into a pixel-parallel and pixel-column architectures. The proposed system is design using TSMC $180nm$ CMOS technology. The power dissipation of the proposed circuit is $96.64mW$, and on-chip ares is $531.66 μm^2$. The architectures are tested for moving object detection application.
An analog synapse circuit based on ferroelectric-metal field-effect transistors is proposed, that offers 6-bit weight precision. The circuit is comprised of volatile least significant bits (LSBs) used solely during training, and non-volatile most significant bits (MSBs) used for both training and inference. The design works at a 1.8V logic-compatible voltage, provides 10^10 endurance cycles, and requires only 250ps update pulses. A variant of LeNet trained with the proposed synapse achieves 98.2% accuracy on MNIST, which is only 0.4% lower than an ideal implementation of the same network with …
Generative Adversarial Network (GAN) is a well known computationally complex algorithm requiring signficiant computational resources in software implementations including large amount of data to be trained. This makes its implementation in edge devices with conventional microprocessor hardware a slow and difficult task. In this paper, we propose to accelerate the computationally intensive GAN using memristive neural networks in analog domain. We present a fully analog hardware design of Deep Convolutional GAN (DCGAN) based on CMOS-memristive convolutional and deconvolutional networks simulated…
Neuromorphic computing promises revolutionary improvements over conventional systems for applications that process unstructured information. To fully realize this potential, neuromorphic systems should exploit the biomimetic behavior of emerging nanodevices. In particular, exceptional opportunities are provided by the non-volatility and analog capabilities of spintronic devices. While spintronic devices have previously been proposed that emulate neurons and synapses, complementary metal-oxide-semiconductor (CMOS) devices are required to implement multilayer spintronic perceptron crossbars. Thi…
SpiNNaker is an ARM-based processor platform optimized for the simulation of spiking neural networks. This brief describes the roadmap in going from the current SPINNaker1 system, a 1 Million core machine in 130nm CMOS, to SpiNNaker2, a 10 Million core machine in 22nm FDSOI. Apart from pure scaling, we will take advantage of specific technology features, such as runtime adaptive body biasing, to deliver cutting-edge power consumption. Power management of the cores allows a wide range of workload adaptivity, i.e. processor power scales with the complexity and activity of the spiking network. Ad…
Conceptual memristors have recently gathered wider interest due to their diverse application in non-von Neumann computing, machine learning, neuromorphic computing, and chaotic circuits. We introduce a compact CMOS circuit that emulates idealized memristor characteristics and can bridge the gap between concepts to chip-scale realization by transcending device challenges. The CMOS memristor circuit embodies a two-terminal variable resistor whose resistance is controlled by the voltage applied across its terminals. The memristor 'state' is held in a capacitor that controls the resistor value. Th…
We experimentally demonstrate classification of 4x4 binary images into 4 classes, using a 3-layer mixed-signal neuromorphic network ("MLP perceptron"), based on two passive 20x20 memristive crossbar arrays, board-integrated with discrete CMOS components. The network features 10 hidden-layer and 4 output-layer analog CMOS neurons and 428 metal-oxide memristors, i.e. is almost an order of magnitude more complex than any previously reported functional memristor circuit. Moreover, the inference operation of this classifier is performed entirely in the integrated hardware. To deal with larger cross…
Analog programming of CMOS-compatible Al$_2$O$_3$/TiO$_\textrm{2-x}$ memristor at 4.2 K after metal-insulator transition suppression by cryogenic reforming
AI applications have emerged in current world. Among AI applications, computer-vision (CV) related applications have attracted high interest. Hardware implementation of CV processors necessitates a high performance but low-power image detector. The key to energy-efficiency work lies in analog-digital converting, where output of imaging detectors is transferred to digital domain and CV algorithms can be performed on data. In this paper, analog-digital converter architectures are compared, and an example ADC design is proposed which achieves both good performance and low power consumption.
Emerging resistive-crossbar memory (RCM) technology can be promising for computationally-expensive analog pattern-matching tasks. However, the use of CMOS analog-circuits with RCM would result in large power-consumption and poor scalability, thereby eschewing the benefits of RCM-based computation. We propose the use of low-voltage, fast-switching, magneto-metallic spin-neurons for ultra low-power non-Boolean computing with RCM. We present the design of analog associative memory for face recognition using RCM, where, substituting conventional analog circuits with spin-neurons can achieve ~100x …
Deep learning hardware designs have been bottlenecked by conventional memories such as SRAM due to density, leakage and parallel computing challenges. Resistive devices can address the density and volatility issues, but have been limited by peripheral circuit integration. In this work, we demonstrate a scalable RRAM based in-memory computing design, termed XNOR-RRAM, which is fabricated in a 90nm CMOS technology with monolithic integration of RRAM devices between metal 1 and 2. We integrated a 128x64 RRAM array with CMOS peripheral circuits including row/column decoders and flash analog-to-dig…
A time-domain analog-weighted-sum calculation model based on a pulse-width modulation (PWM) approach is proposed. The proposed calculation model can be applied to any types of network structure including multi-layer feedforward networks. We also propose very large-scale integrated (VLSI) circuits to implement the proposed model. Unlike the conventional analog voltage or current mode circuits used in computing-in-memory circuits, our time-domain analog circuits use transient operation in charging/discharging processes to capacitors. Since the circuits can be designed without operational amplifi…
Modern analog computing for solving differential and matrix equations
Institute for Artificial Intelligence, School of Integrated Circuits, Peking University, Beijing Advanced Innovation Center for Integrated Circuits（人工智能研究院，集成电路学院，北京大学，北京集成电路先进创新中心）；Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano（电子、信息与生物工程系，米兰理工大学）；IBM Research Europe（IBM欧洲研究院）
Ferroelectric field effect transistors (FeFETs) are being actively investigated with the potential for in-memory computing (IMC) over other non-volatile memories (NVMs). Content Addressable Memories (CAMs) are a form of IMC that performs parallel searches for matched entries over a memory array for a given input query. CAMs are widely used for data-centric applications that involve pattern matching and search functionality. To accommodate the ever expanding data, it is attractive to resort to analog CAM for memory density improvement. However, the digital CAM design nowadays based on standard …
Neuromorphic computing architectures enable the dense co-location of memory and processing elements within a single circuit. This co-location removes the communication bottleneck of transferring data between separate memory and computing units as in standard von Neuman architectures for data-critical applications including machine learning. The essential building blocks of neuromorphic systems are non-volatile synaptic elements such as memristors. Key memristor properties include a suitable non-volatile resistance range, continuous linear resistance modulation and symmetric switching. In this …
Although Analog-to-digital converters (ADCs) are critical components in mixed-signal integrated circuits (IC), their performance has not been improved significantly over the last decade. To achieve a radical improvement (compact, low power and reliable ADCs), spintronics can be considered as a proper candidate due to its compatibility with CMOS and wide applications in storage, neuromorphic computing, and so on. In this paper, a proof-of-concept of a 3-bit spin-CMOS Flash ADC using in-plane-anisotropy magnetic tunnel junctions (i-MTJs) with spin-orbit torque (SOT) switching mechanism is design…
We present a low barrier magnet based compact hardware unit for analog stochastic neurons and demonstrate its use as a building-block for neuromorphic hardware. By coupling circular magnetic tunnel junctions (MTJs) with a CMOS based analog buffer, we show that these units can act as leaky-integrate-and fire (LIF) neurons, a model of biological neural networks particularly suited for temporal inferencing and pattern recognition. We demonstrate examples of temporal sequence learning, processing, and prediction tasks in real time, as a proof of concept demonstration of scalable and adaptive signa…
Brain-inspired computing and neuromorphic hardware are promising approaches that offer great potential to overcome limitations faced by current computing paradigms based on traditional von-Neumann architecture. In this regard, interest in developing memristor crossbar arrays has increased due to their ability to natively perform in-memory computing and fundamental synaptic operations required for neural network implementation. For optimal efficiency, crossbar-based circuits need to be compatible with fabrication processes and materials of industrial CMOS technologies. Herein, we report a compl…
Due to the massive parallel computing capability and outstanding image and signal processing performance, cellular neural network (CNN) is one promising type of non-Boolean computing system that can outperform the traditional digital logic computation and mitigate the physical scaling limit of the conventional CMOS technology. The CNN was originally implemented by VLSI analog technologies with operational amplifiers and operational transconductance amplifiers as neurons and synapses, respectively, which are power and area consuming. In this paper, we propose a hybrid structure to implement the…
