A5079471306- Advanced Memory and Neural Computing
- Parallel Computing and Optimization Techniques
- Ferroelectric and Negative Capacitance Devices
- Embedded Systems Design Techniques
- Advanced Data Storage Technologies
- Advanced Vision and Imaging
- Advanced Graph Neural Networks
- Advanced Algorithms and Applications
- Advanced Neural Network Applications
- Interconnection Networks and Systems
- Perovskite Materials and Applications
- Advanced Sensor and Control Systems
- Caching and Content Delivery
- Distributed and Parallel Computing Systems
- Advanced Computational Techniques and Applications
- Image Processing Techniques and Applications
- CCD and CMOS Imaging Sensors
- Algorithms and Data Compression
- Radio Astronomy Observations and Technology
- Graph Theory and Algorithms
- Analog and Mixed-Signal Circuit Design
- Low-power high-performance VLSI design
- Ferroelectric and Piezoelectric Materials
- Galaxies: Formation, Evolution, Phenomena
- Neural Networks and Reservoir Computing
University of Science and Technology of China
2016-2025
Academy of Mathematics and Systems Science
2024
Chinese Academy of Sciences
2016-2024
Beijing Institute of Technology
2024
Sun Yat-sen University
2024
Songshan Lake Materials Laboratory
2024
South China Normal University
2024
Institute of Microelectronics
2008-2023
National University of Singapore
2023
Sungkyunkwan University
2023
Abstract Next-generation non-volatile memories with ultrafast speed, low power consumption, and high density are highly desired in the era of big data. Here, we report a performance memristor based on Ag/BaTiO 3 /Nb:SrTiO ferroelectric tunnel junction (FTJ) fastest operation speed (600 ps) highest number states (32 or 5 bits) per cell among reported FTJs. The sub-nanosecond resistive switching maintains up to 358 K, write current is as 4 × 10 A cm −2 . functionality spike-timing-dependent...
The rapid development of neuro-inspired computing demands synaptic devices with ultrafast speed, low power consumption, and multiple non-volatile states, among other features. Here, a high-performance device is designed established based on Ag/PbZr0.52Ti0.48O3 (PZT, (111)-oriented)/Nb:SrTiO3 ferroelectric tunnel junction (FTJ). advantages (111)-oriented PZT (~1.2 nm) include its switching dynamics, ultrafine domains, small coercive voltage. FTJ shows high-precision (256 8 bits), reproducible...
Abstract Perovskite (PVK) solar cells (PSCs) have garnered considerable research interest owing to their cost-effectiveness and high efficiency. A systematic annual review of the on PSCs is essential for gaining a comprehensive understanding current trends. Herein, analysis papers reporting key findings in 2023 was conducted. Based results, were categorized into six classifications, including regular n–i–p PSCs, inverted p–i–n PVK-based tandem cells, PVK modules, device stability, lead...
Flexible ferroelectric devices have been a hot-spot topic because of their potential wearable applications as nonvolatile memories and sensors. Here, high-quality (111)-oriented BiFeO3 films are grown on flexible mica substrates through an appropriate design SrRuO3/BaTiO3 double buffer layers. exhibits the largest polarization (saturated Ps ≈ 100 μC/cm2, remnant Pr 97 μC/cm2) among all reported films, is very stable in 104 bending cycles under 5 mm radius. Accordingly, memristor behaviors...
A perovskite precursor stabilization and defect passivation strategy is developed by coordination engineering with a series of structurally similar crown ethers.
The illumination side of perovskite solar cells is more vulnerable to external impact (such as hail, flying rocks, snow, hurricanes, etc.) than the rear side, leading likelihood of...
Existing Hospital Information System (HIS) is only shared within separated hospitals or districts, which resulted in high construction costs and waste of resources, especially for small township other health care systems are incapable to set up an independent information platform. We propose a solution simplify this by using network module attached existing that connected the hospitals. The proposal based on concepts cloud computing so sharing high-end processing available "cloud"....
Abstract Memristive logic‐in‐memory (LIM) is an attractive candidate for in‐memory computing and thus would overcome some of the issues concerning von Neumann bottleneck, which has been intensively investigated. Here, a multi‐bit functionally complete LIM strategy designed to efficiently perform multi logical functions in parallel. This experimentally demonstrated by ferroelectric tunnel junction (FTJ) memristor with sub‐nanosecond operation speed nearly two orders magnitude faster than...
Aiming at discriminating different gravitational potential models of the Milky Way, we perform tests based on kinematic data powered by Gaia DR2 astrometry, over a large range $(R,z)$ locations. Invoking complete form Jeans equations that admit three integrals motion, use independent $R$- and $z$-directional as two discriminators ($T_R$ $T_z$). We apply formula for spatial distributions radial vertical velocity dispersions proposed Binney et al., successfully extend it to azimuthal...
Graph attention networks (GATs) are a mainstream graph neural network (GNN) model. They have better performance in some tasks compared to other GNN models. The challenge is that data structures irregular and dependency GATs complex. General-purpose hardware cannot provide enough or energy efficiency. Therefore, specialized accelerator for needed. In this brief, we propose FTW-GAT accelerate GAT inference. key idea of our approach quantize the weights ternary values, which can greatly...
A carry-select adder(CSA) can be implemented by using single ripple carry adder and an add-one circuit instead of dual ripple-carry adders to reduce the area power but with speed penalty. This paper proposes a new circuits fast all-one finding low-delay multiplexers accelerate CSA, no restrictions are imposed on design blocks. For bit length n = 64, this requires approximate 38 percent fewer transistors 16 shorter delay than original adder.
Three-dimensional convolutional networks (3D CNNs) are used efficiently in various video recognition applications. Compared to traditional 2D CNNs, extra temporal dimension causes 3D CNNs more computationally intensive and have a larger memory footprint. Therefore, the optimization is extremely crucial this case. This paper presents design space exploration of access for FPGA-based CNN accelerator. We present non-overlapping data tiling method contiguous off-chip explore on-chip reuse...
As a modified-gravity proposal to handle the dark matter problem on galactic scales, Modified Newtonian Dynamics (MOND) has shown great success. However, the<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M2"><mml:mrow><mml:mi>N</mml:mi></mml:mrow></mml:math>-body MOND simulation is quite challenged by its computation complexity, which appeals acceleration of calculation. In this paper, we present highly integrated accelerating solution for<mml:math...
In recent years, solid-state drives (SSDs) have been widely deployed in modern storage systems. To increase the performance of SSDs, prefetchers for SSDs designed both at operating system (OS) layer and flash translation (FTL). Prefetchers FTL many advantages like OS-independence, easy-using, compatibility. However, due to limitation computing capabilities memory resources, existing merely employ simple sequential prefetching which may incur high penalty cost I/O access stream with complex...
In high-performance computing systems, each node communicates via a high-speed serial bus to ensure sufficient data transfer bandwidth. However, of different protocols is very difficult communicate directly, which not conducive the extensibility HPC (High performance computing) clusters. this paper, we propose UPI, inter-node communication interface based on FPGA, can transmit (PCIe protocol and Ethernet protocol) simultaneously. More importantly, many bus-supported nodes be connected same...
Convolutional Neural Networks (CNNs) have been widely used in various image recognition applications due to their high precision and versatility. In order improve real-time performance, many efficient CNNs with low computational complexity proposed. However, compared the reduction amount of calculation, demand for memory access has not decreased simultaneously. How balance utilization resources become a challenge. This brief deeply deconstructs CNN network analyzes relationship among...