In this article, we review the existing analog resistive switching memory (RSM) devices and their hardware technologies for in-memory learning, as well as their challenges and prospects. Since the characteristics of the devices are different for in-memory learning and digital memory applications, it is important to have an in-depth understanding across different layers from devices and circuits to architectures and algorithms. First, based on a top-down view from architecture to devices for analog computing, we define the main figures of merit (FoMs) and perform a comprehensive analysis of analog RSM hardware including the basic device characteristics, hardware algorithms, and the corresponding mapping methods for device arrays, as well as the architecture and circuit design considerations for neural networks. Second, we classify the FoMs of analog RSM devices into two levels. Level 1 FoMs are essential for achieving the functionality of a system (e.g., linearity, symmetry, dynamic range, level numbers, fluctuation, variability, and yield). Level 2 FoMs are those that make a functional system more efficient and reliable (e.g., area, operational voltage, energy consumption, speed, endurance, retention, and compatibility with back-end-of-line processing). By constructing a device-to-application simulation framework, we perform an in-depth analysis of how these FoMs influence in-memory learning and give a target list of the device requirements. Lastly, we evaluate the main FoMs of most existing devices with analog characteristics and review optimization methods from programming schemes to materials and device structures. The key challenges and prospects from the device to system level for analog RSM devices are discussed.

In-memory Learning with Analog Resistive Switching Memory: A Review and Perspective

Racetrack memory (RTM) is a novel spintronic memory-storage technology that has the potential to overcome fundamental constraints of existing memory and storage devices. It is unique in that its core differentiating feature is the movement of data, which is composed of magnetic domain walls (DWs), by short current pulses. This enables more data to be stored per unit area compared to any other current technologies. On the one hand, RTM has the potential for mass data storage with unlimited endurance using considerably less energy than today’s technologies. On the other hand, RTM promises an ultrafast nonvolatile memory competitive with static random access memory (SRAM) but with a much smaller footprint. During the last decade, the discovery of novel physical mechanisms to operate RTM has led to a major enhancement in the efficiency with which nanoscopic, chiral DWs can be manipulated. New materials and artificially atomically engineered thin-film structures have been found to increase the speed and lower the threshold current with which the data bits can be manipulated. With these recent developments, RTM has attracted the attention of the computer architecture community that has evaluated the use of RTM at various levels in the memory stack. Recent studies advocate RTM as a promising compromise between, on the one hand, power-hungry, volatile memories and, on the other hand, slow, nonvolatile storage. By optimizing the memory subsystem, significant performance improvements can be achieved, enabling a new era of cache, graphical processing units, and high capacity memory devices. In this article, we provide an overview of the major developments of RTM technology from both the physics and computer architecture perspectives over the past decade. We identify the remaining challenges and give an outlook on its future.

/pdf/magnetic-racetrack-memory-from-physics-to-the-cusp-of-5brh05qnrb.pdf

Magnetic Racetrack Memory: From Physics to the Cusp of Applications Within a Decade

In this paper, we propose MRIMA, as a novel magnetic RAM (MRAM)-based in-memory accelerator for nonvolatile, flexible, and efficient in-memory computing. MRIMA transforms current spin transfer torque magnetic random access memory (STT-MRAM) arrays to massively parallel computational units capable of working as both nonvolatile memory and in-memory logic. Instead of integrating complex logic units in cost-sensitive memory, MRIMA exploits hardware-friendly bit-line computing methods to implement complete Boolean logic functions between operands within a memory array in a single clock cycle, overcoming the multicycle logic issue in contemporary processing-in-memory (PIM) platforms. We present practical case studies to demonstrate MRIMA’s acceleration for binary-weight and low bit-width convolutional neural networks (CNNs) as well as data encryption. Our device-to-architecture co-simulation results on CNN acceleration demonstrate that MRIMA can obtain $1.7 {\times }$ better energy-efficiency and $11.2{\times }$ speed-up compared to ASICs, and $1.8 {\times }$ better energy-efficiency and $2.4 {\times }$ speed-up over the best DRAM-based PIM solutions. As an advanced encryption standard (AES) in-memory encryption engine, MRIMA shows ~77% and 21% lower energy consumption compared to CMOS-ASIC and recent domain-wall-based design, respectively.

https://par.nsf.gov/servlets/purl/10094221

MRIMA: An MRAM-Based In-Memory Accelerator

The recently emerging resistive random-access memory (RRAM) can provide nonvolatile memory storage but also intrinsic computing for matrix-vector multiplication, which is ideal for the low-power and high-throughput data analytics accelerator performed in memory. However, the existing RRAM crossbar--based computing is mainly assumed as a multilevel analog computing, whose result is sensitive to process nonuniformity as well as additional overhead from AD-conversion and I/O. In this article, we explore the matrix-vector multiplication accelerator on a binary RRAM crossbar with adaptive 1-bit-comparator--based parallel conversion. Moreover, a distributed in-memory computing architecture is also developed with the according control protocol. Both memory array and logic accelerator are implemented on the binary RRAM crossbar, where the logic-memory pair can be distributed with the control bus protocol. Experimental results have shown that compared to the analog RRAM crossbar, the proposed binary RRAM crossbar can achieve significant area savings with better calculation accuracy. Moreover, significant speedup can be achieved for matrix-vector multiplication in neural network--based machine learning such that the overall training and testing time can be both reduced. In addition, large energy savings can be also achieved when compared to the traditional CMOS-based out-of-memory computing architecture.

Distributed In-Memory Computing on Binary RRAM Crossbar

With the in-memory processing ability, ReRAM based computing gets more and more attractive for accelerating neural networks (NNs). However, most ReRAM based accelerators cannot support efficient mapping for sparse NN, and we need to map the whole dense matrix onto ReRAM crossbar array to achieve O(1) computation complexity. In this paper, we propose a sparse NN mapping scheme based on elements clustering to achieve better ReRAM crossbar utilization. Further, we propose crossbar-grained pruning algorithm to remove the crossbars with low utilization. Finally, since most current ReRAM devices cannot achieve high precision, we analyze the effect of quantization precision for sparse NN, and propose to complete high-precision composing in the analog field and design related periphery circuits. In our experiments, we discuss how the system performs with different crossbar sizes to choose the optimized design. Our results show that our mapping scheme for sparse NN with proposed pruning algorithm achieves 3 -- 5X energy efficiency and more than 2.5 -- 6X speedup, compared with those accelerators for dense NN. Also, the accuracy experiments show that our pruning method appears to have almost no accuracy loss.

Learning the sparsity for ReRAM: mapping and pruning sparse neural network for ReRAM based accelerator

The data-oriented applications have introduced increased demands on memory capacity and bandwidth, which raises the need to rethink the architecture of the current computing platforms. The logic-in-memory architecture is highly promising as future logic-memory integration paradigm for high throughput data-driven applications. From memory technology aspect, as one recently introduced nonvolatile memory device, domain-wall nanowire (or race-track) not only shows potential as future power efficient memory, but also computing capacity by its unique physics of spintronics. This paper explores a novel distributed in-memory computing architecture where most logic functions are executed within the memory, which significantly alleviates the bandwidth congestion issue and improves the energy efficiency. The proposed distributed in-memory computing architecture is purely built by domain-wall nanowire, i.e., both memory and logic are implemented by domain-wall nanowire devices. As a case study, neural network-based image resolution enhancement algorithm, called DW-NN, is examined within the proposed architecture. We show that all operations involved in machine learning on neural network can be mapped to a logic-in-memory architecture by nonvolatile domain-wall nanowire. Domain-wall nanowire-based logic is customized for in machine learning within image data storage. As such, both neural network training and processing can be performed locally within the memory. The experimental results show that the domain-wall memory can reduce 92% leakage power and 16% dynamic power compared to main memory implemented by DRAM; and domain-wall logic can reduce 31% both dynamic and 65% leakage power under the similar performance compared to CMOS transistor-based logic. And system throughput in DW-NN is improved by 11.6x and the energy efficiency is improved by 56x when compared to conventional image processing system.

An Energy-Efficient Nonvolatile In-Memory Computing Architecture for Extreme Learning Machine by Domain-Wall Nanowire Devices

Emerging resistive random-access memory (RRAM) can provide non-volatile memory storage but also intrinsic logic for matrix-vector multiplication, which is ideal for low-power and high-throughput data analytics accelerator performed in memory. However, the existing RRAM-based computing device is mainly assumed on a multi-level analog computing, whose result is sensitive to process non-uniformity as well as additional AD- conversion and I/O overhead. This paper explores the data analytics accelerator on binary RRAM-crossbar. Accordingly, one distributed in-memory computing architecture is proposed with design of according component and control protocol. Both memory array and logic accelerator can be implemented by RRAM-crossbar purely in binary, where logic-memory pairs can be distributed with protocol of control bus. Based on numerical results for fingerprint matching that is mapped on the proposed RRAM-crossbar, the proposed architecture has shown 2.86x faster speed, 154x better energy efficiency, and 100x smaller area when compared to the same design by CMOS-based ASIC.

An energy-efficient matrix multiplication accelerator by distributed in-memory computing on binary RRAM crossbar

Image processing in conventional logic-memory I/O-integrated systems will incur significant communication congestion at memory I/Os for excessive big image data at exa-scale. This paper explores an in-memory machine learning on neural network architecture by utilizing the newly introduced domain-wall nanowire, called DW-NN. We show that all operations involved in machine learning on neural network can be mapped to a logic-in-memory architecture by non-volatile domain-wall nanowire. Domain-wall nanowire based logic is customized for in machine learning within image data storage. As such, both neural network training and processing can be performed locally within the memory. The experimental results show that system throughput in DW-NN is improved by 11.6x and the energy efficiency is improved by 92x when compared to conventional image processing system.

/pdf/energy-efficient-in-memory-machine-learning-for-data-1sxzzl47cr.pdf

Energy efficient in-memory machine learning for data intensive image-processing by non-volatile domain-wall memory

The emerging resistive random-access-memory (RRAM) crossbar provides an intrinsic fabric for matrix-vector multiplication, which can be leveraged as power efficient linear embedding hardware for data analytics such as compressive sensing. As the matrix elements are represented by resistance of RRAM cells, it imposes constraints for the embedding matrix due to limited RRAM programming resolution. A random Boolean embedding can be efficiently mapped to the RRAM crossbar but suffers from poor performance. Learning-based embedding matrices can deliver optimized performance but are continuous-valued which prevents it from being mapped to RRAM crossbar structure directly. In this paper, we have proposed one algorithm that can find an optimal Boolean embedding matrix for a given learned real-valued embedding matrix, so that it can be effectively mapped to the RRAM crossbar structure while high performance is preserved. The numerical experiments demonstrate that the proposed optimized Boolean embedding can reduce the embedding distortion by 2.7x, and image recovery error by 2.5x compared to the random Boolean embedding, both mapped on RRAM crossbar. In addition, optimized Boolean embedding on RRAM crossbar exhibits 10x faster speed, 17x better energy efficiency, and three orders of magnitude smaller area with slight accuracy penalty, when compared to the optimized real-valued embedding on CMOS ASIC platform.

Optimizing Boolean embedding matrix for compressive sensing in RRAM crossbar

Data analytics such as face recognition involves large volume of image data, and hence leads to grand challenge on mobile platform design with strict power requirement. Emerging non-volatile STT-MRAM has the minimum leakage power and comparable speed to SRAM, and hence is considered as a promising candidate for data-oriented mobile computing. However, there exists significantly higher write-energy for STT-MRAM when compared to the SRAM. Based on the use of STT-MRAM, this paper introduces an energy-efficient non-volatile in-memory accelerator for a sparse-representation based face recognition algorithm. We find that by projecting high-dimension image data to much lower dimension, the current scaling for STT-MRAM write operation can be applied aggressively, which leads to significant power reduction yet maintains quality-of-service for face recognition. Specifically, compared to a baseline with SRAM, leakage power and dynamic power are reduced by 91.4% and 79% respectively with only slight compromise on recognition rate.

Chuliang Weng

Papers

An Energy-Efficient Nonvolatile In-Memory Computing Architecture for Extreme Learning Machine by Domain-Wall Nanowire Devices

An energy-efficient matrix multiplication accelerator by distributed in-memory computing on binary RRAM crossbar

Energy efficient in-memory machine learning for data intensive image-processing by non-volatile domain-wall memory

Optimizing Boolean embedding matrix for compressive sensing in RRAM crossbar

An energy-efficient non-volatile in-memory accelerator for sparse-representation based face recognition