Fast Memory Research Articles

Enhancing Efficiency of the Fast Quantum Memory on Single-Atom in Cavity

Enhancing Efficiency of the Fast Quantum Memory on Single-Atom in Cavity

Ultra-durable cell-free bioactive hydrogel with fast shape memory and on-demand drug release for cartilage regeneration

Osteoarthritis is a worldwide prevalent disease that imposes a significant socioeconomic burden on individuals and healthcare systems. Achieving cartilage regeneration in patients with osteoarthritis remains challenging clinically. In this work, we construct a multiple hydrogen-bond crosslinked hydrogel loaded with tannic acid and Kartogenin by polyaddition reaction as a cell-free scaffold for in vivo cartilage regeneration, which features ultra-durable mechanical properties and stage-dependent drug release behavior. We demonstrate that the hydrogel can withstand 28000 loading-unloading mechanical cycles and exhibits fast shape memory at body temperature (30 s) with the potential for minimally invasive surgery. We find that the hydrogel can also alleviate the inflammatory reaction and regulate oxidative stress in situ to establish a microenvironment conducive to healing. We show that the sequential release of tannic acid and Kartogenin can promote the migration of bone marrow mesenchymal stem cells into the hydrogel scaffold, followed by the induction of chondrocyte differentiation, thus leading to full-thickness cartilage regeneration in vivo. This work may provide a promising solution to address the problem of cartilage regeneration.

Pengaruh Kapasitas Memori RAM (Random Access Memory) Terhadap Kecepatan Memori Pada Laptop

The memory on a laptop is very important as it is the main component of the laptop, memory which is usually used to store information for direct use on the laptop. Memory also consists of RAM (Random Access Memory), where if the RAM has a higher capacity or storage, the device operates quickly and easily. Therefore, this research will discuss memory, which aims to determine the effect of memory capacity on memory speed, which in this research uses three (3) laptops, laptop X, laptop Y and Laptop Z. Laptop X has AMD Ryzen 5 5600H processor specifications. with Radeon Graphics and 8.00 GB RAM (8.41 GB unused) and laptop Y has Intel® Core™ i3-1005G1 processor specifications and 4.00 GB RAM (3.75 GB unused), while Laptop Z has AMD A6-5200 APU processor specifications with Radeon™ HD Graphics and 2.00 GB RAM(1.47 GB unused). Testing the validity and reliability of the instrument but the description of the symbols on the model is written in sentences. Therefore, memory that has a higher RAM capacity or larger RAM has fast memory performance speed and is easy to operate, while laptops that have small RAM have low performance speed which results in low performance speed and results in slow performance.

Educational Service Model for The Special Intelligence and Special Talents of Students: Literature Review

Children with special intelligence and special talents are those who have above average intelligence, have faster memory and comprehension abilities compared to their friends. Thus, children in this category have many achievements; apart from that they also have the characteristics of self-confidence and perfectionism. Children with special intelligence and special talents certainly have intelligence or ability in one area that stands out the most. This research uses a literature review or literature review design, by examining 18 articles consisting of national and international articles related to service models for students with special intelligence and special talents. From the articles, it was found that the education service model for special intelligent children and special talents in several countries is the inclusive and accelerated model.

Strong atom capture ability and pinning effect by doping an element with large electronegativity difference in Sb2Te3 phase change material

The ultimate goal of memory researchers and developers is to devise a universal memory to replace all memories in complex hierarchies of memory systems. Phase change memory is considered by the International Semiconductor Industry Association as the most likely emerging storage technology to achieve this ambitious goal due to its excellent performance. However, the relatively slow operation speed and poor cycle endurance are stumbling blocks for this purpose. Here, we propose a new doping strategy to construct stable bonding by introducing atoms with large electronegativity differences from the matrix Te atoms enhancing the capture ability and the pinning effect to Te atoms, which facilitates fast nucleation and suppresses migration of Te atoms to achieve significant performance improvement. Using ab initio molecular dynamics simulations, we explore and compare the atomic structures, chemical bonding, and dynamics properties of Hf doped Sb2Te3 (HfST), La doped Sb2Te3 (LaST), Rb doped Sb2Te3 (RbST) and Sr doped Sb2Te3 (SrST). In the XST (X refers to Hf, Rb, La and Sr atoms) systems, the increase of the more stable ring structure fluctuation and the stronger atom capture capability of rings enable more and faster formation of nuclei during crystallization. Higher Peierls-like distortions suggest better thermal stability. The motility of Te atoms is suppressed due to the pinning effect. High coordination of atoms and the concentration reduction of the lone-pair electron and void enable small density difference. Our results show that fast universal memory with good endurance can be achieved by the construction of strong bonding by doping atoms with large electronegativity differences, providing an important guidance for experiments and industries.

Spin-based magnetic random-access memory for high-performance computing.

Spin-based memory technology is now available as embedded magnetic random access memory (eMRAM) for fast, high-density and non-volatile memory products, which can significantly boost computing performance and ignite the development of new computing architectures.

NIR light-induced functionalized MXene as a dynamic-crosslinker for reinforced polyurethane composites with shape memory and self-healing

Near infrared (NIR) light-driven shape memory can expand the potential of polyurethane as advanced materials. Herein, polyurethane composites (PU/B-THDI/FMXene) with Diels-Alder (D-A) reversible crosslinking feature were fabricated in combination with internal crosslinker bis(2-hydroxyethyl) terephthalate (BHET) modified hexamethylene-diisocyanate trimer (B-THDI), and functionalized MXene (FMXene). FMXene was prepared via 3-isocyanatopropyltriethoxysilane (IPTS) as a coupling agent between MXene and furfuryl alcohol (FA) through covalent bonds, thereby enabling dynamic crosslinking. FMXene can not only form D-A dynamic-crosslinked network with PU chain, but also retain excellent thermal conductivity. Only containing 2 wt% FMXene endows PU/B-THDI/FMXene2 composites with fast shape memory (<15 s, shape-fixity and recovery rates are 88.1 % and 97.8 %, respectively), self-healing (<100 s, the healing efficiency for tensile strength and elongation at break are 88.0 % and 91.0 %, respectively), good self-healing repeatability (79.1 %) and thermal conductivity (0.661 W m−1 K−1). Moreover, PU/B-THDI/FMXene2 exhibits remarkable mechanical performances (tensile strength of 16.9 MPa and elongation at break of 233.9 %) and thermal stability. In addition, B-THDI with crosslinking and toughening effects provides certain strength and flexibility for PU. Thus, this work may open new approaches for developing photothermal-induced shape memory polyurethane.

A novel and computationally efficient subgrid technique for modeling scattering and Doppler effects of rough dynamic sea surfaces using the finite-difference time-domain method

Over the last decade, the improvements in computational performance that multiprocessing, solid-state hard drives, and fast memories have brought to desktop workstations have made modeling and simulation methods that were previously inaccessible into the office workspace. These methods were previously restricted to large servers or computer clusters because of their demanding computation and large memory requirements when applied to underwater acoustic simulation scenarios. One such method is the Finite-difference Time-domain (FDTD) technique. The FDTD method is a discrete numeric model that provides the full pressure wave solution solved directly in the time domain. This makes it an ideal model for investigating scattering and Doppler effects of boundaries with complex geometries and motion. A fully developed sea surface generated using the Pierson–Moskowitz wavenumber spectra meets these criteria. A novel subgrid technique has been developed to improve the accuracy of signals scattered from fully developed sea surfaces. This method remains computationally efficient because the subgrid is applied only along the rough sea surface as it varies over time. The FDTD subgrid method has been applied to simulations that investigate the “frozen” sea surface assumption that is utilized by other traditional models that incorporate boundary motion.

Exploring the time-resolving capabilities of XIDer assemblies

The XIDer R&D project, between the European Synchrotron Radiation Facility (ESRF) and Heidelberg University, is developing state-of-the-art detectors for the high-energy, high-flux beamlines of the ESRF Extremely Brilliant Source (EBS). In particular, the XIDer detector aims to tackle the significant challenges posed by time-resolved experiments analysing the pattern of diffracted or scattered X-rays.To fulfil the needs of these experiments, a novel on-chip incremental digital integration readout scheme with per-pixel digital data output has been implemented. The EBS beam operates with different time structures: from a near-100% duty cycle X-ray beam to an up to 5.7MHz pulsed X-ray beam. For the latter EBS mode, the XIDer ASIC has been designed to have a fast front-end and memory storage cells, such that the pulsed EBS beam can be exploited by time-resolved experiments. Recent measurements of the time-resolving capabilities of XIDer ASICs hybridised cadmium telluride and cadmium zinc telluride sensors are presented.

Large-scale network dynamics underlying the first few hundred milliseconds after stimulus presentation: An investigation of visual recognition memory using iEEG.

Recognition memory is the ability to recognize previously encountered objects. Even this relatively simple, yet extremely fast, ability requires the coordinated activity of large-scale brain networks. However, little is known about the sub-second dynamics of these networks. The majority of current studies into large-scale network dynamics is primarily based on imaging techniques suffering from either poor temporal or spatial resolution. We investigated the dynamics of large-scale functional brain networks underlying recognition memory at the millisecond scale. Specifically, we analyzed dynamic effective connectivity from intracranial electroencephalography while epileptic subjects (n = 18) performed a fast visual recognition memory task. Our data-driven investigation using Granger causality and the analysis of communities with the Louvain algorithm spotlighted a dynamic interplay of two large-scale networks associated with successful recognition. The first network involved the right visual ventral stream and bilateral frontal regions. It was characterized by early, predominantly bottom-up information flow peaking at 115 ms. It was followed by the involvement of another network with predominantly top-down connectivity peaking at 220 ms, mainly in the left anterior hemisphere. The transition between these two networks was associated with changes in network topology, evolving from a more segregated to a more integrated state. These results highlight that distinct large-scale brain networks involved in visual recognition memory unfold early and quickly, within the first 300 ms after stimulus onset. Our study extends the current understanding of the rapid network changes during rapid cognitive processes.

Prefrontal stimulation as a tool to disrupt hippocampal and striatal reactivations underlying fast motor memory consolidation

BackgroundRecent evidence suggests that hippocampal replay in humans support rapid motor memory consolidation during epochs of wakefulness interleaved with task practice. Objectives/HypothesesThe goal of this study was to test whether such reactivation patterns can be modulated with experimental interventions and in turn influence fast consolidation. We hypothesized that non-invasive brain stimulation targeting hippocampal and striatal networks via the prefrontal cortex would influence brain reactivation and the rapid form of motor memory consolidation. MethodsTheta-burst stimulation was applied to a prefrontal cluster functionally connected to both the hippocampus and striatum of young healthy participants before they learned a motor sequence task in a functional magnetic resonance imaging (fMRI) scanner. Neuroimaging data acquired during task practice and the interleaved rest epochs were analyzed to comprehensively characterize the effect of stimulation on the neural processes supporting fast motor memory consolidation. ResultsOur results collectively show that active, as compared to control, theta-burst stimulation of the prefrontal cortex hindered fast motor memory consolidation. Converging evidence from both univariate and multivariate analyses of fMRI data indicate that active stimulation disrupted hippocampal and caudate responses during inter-practice rest, presumably altering the reactivation of learning-related patterns during the micro-offline consolidation episodes. Last, stimulation altered the link between the brain and the behavioral markers of the fast consolidation process. ConclusionThese results suggest that stimulation targeting deep brain regions via the prefrontal cortex can be used to modulate hippocampal and striatal reactivations in the human brain and influence motor memory consolidation.

FIMAR: Fast incremental memory acquisition and restoration system for temporal-dimension forensic analysis

Recent advances in anti-forensic techniques, including encryption, obfuscation, and kernel-structure manipulations, make conventional storage, network, and memory forensics difficult. For example, the code of packed malware is encrypted on storage, network packets, and system RAM to hide its existence; it appears briefly in plain text on the system RAM during its execution phase. Memory forensics, therefore, has been a vital component of today's digital forensics. Conventional memory acquisition techniques, however, capture only the current state of a system RAM; if the timing of the acquisition is late, the evidence could be lost forever. Since the state-of-the-art anti-forensic attacks leave only the slightest evidence for the shortest time, finer temporal acquisition and analysis are needed. This paper presented a Fast Incremental Memory Acquisition and Restoration (FIMAR) system that enables us to acquire multiple memory snapshots (i.e., memory dumps) over time and to construct a timeline of past activities on a computer. A thin hypervisor and its Second Level Address Translation (SLAT) were used to track memory page changes and to obtain atomic memory snapshots. The hash calculation of 4 KiB chunks was used to detect updated parts of the system RAM to reduce the transfer size. We implemented the incremental memory acquisition algorithms using a thin hypervisor called BitVisor. Finally, we tested the developed FIMAR system to analyze BlueSky ransomware with anti-forensic logic. The developed FIMAR system improved the atomicity of memory snapshots using TLB shootdown and considering per-core Extended Page Tables (EPTs) compared to previous hypervisor-based memory acquisition systems.

RHPM: Using Relative Hotness to Guide Page Migration for Hybrid Memory Systems

Modern computing systems and data-intensive applications are eager for larger and faster memory. Building hybrid memory systems with different memory technologies has become a dominant trend to satisfy these demands. For hybrid memory systems, page migration schemes that dynamically migrate frequently accessed hot pages into faster memory are crucial for improving performance. However, existing migration schemes are either too aggressive, resulting in unnecessary extra traffic, or too conservative to quickly adapt to changes in access patterns. Besides, the extra latency introduced by querying metadata is often ignored or handled in an unscalable manner. In this paper, we propose a Relative Hotness Page Migration (RHPM) strategy, which discovers hot pages in a set of pages by competing with each other rather than comparing with a threshold. The migration is performed only when a new page wins the competition. To overlap latency due to access metadata, RHPM fetches metadata and data in parallel. In addition, it enables a small metadata buffer to speed up metadata access. Compared to the state-of-the-art scheme, RHPM requires only 1/512 of the on-chip capacity, significantly reducing on-chip hardware overhead. Evaluation of RHPM with simulations of 25 workloads shows that RHPM outperforms the state-of-the-art scheme by an average of 13.34% in performance and saves 44.19% on energy, demonstrating better resilience to changes in access patterns.

Research on harmonic resonant fault mechanism and suppression measures of oilfield power grid based on fast mode analysis method

In recent years, with the use of electric drilling rig, electric fracturing pump, and other equipment, the harmonic resonance problem of oilfield power grid is becoming more and more serious. In view of the high time and space complexity and long calculation time of the traditional harmonic resonance analysis method, the improved power iteration method is combined with the traditional modal analysis method, and the harmonic resonance problem of the oilfield power grid is analyzed by the fast modal analysis method. At the same time, a new set of harmonic resonance amplification index is derived and verified, and the harmonic content amplification times in the resonant influence region and different busbars are revealed. And then a passive filter is used to suppress harmonic resonance, and the effectiveness of the scheme is verified by field tests. The results show that, in the frequency range of 0–1000 Hz, the resonant frequency points in the oil field are mainly related to the rectifier pulse number of the drilling rig and the frequency conversion device for fracturing. The proposed fast modal analysis method has the advantages of fast convergence speed and memory saving, which is helpful to accurately detect the harmonic resonance region caused by the multi-pulse frequency conversion device, and the harmonic resonance can be effectively suppressed by installing a filter at the main excitation bus of the resonant frequency.

AC2AS: Activation Consistency Coupled ANN-SNN framework for fast and memory-efficient SNN training

Spiking neural networks are efficient computation models for low-power environments. Spike-based BP algorithms and ANN-to-SNN (ANN2SNN) conversions are successful techniques for SNN training. Nevertheless, the spike-base BP training is slow and requires large memory costs, while ANN2SNN needs many inference steps to obtain good performance. In this paper, we propose an Activation Consistency Coupled ANN-SNN (AC2AS) framework to train the SNN in a fast and memory-efficient way. The AC2AS consists of two components: (a) a weight-shared architecture between ANN and SNN and (b) spiking mapping units. Firstly, the architecture trains the weight-shared parameters on the ANN branch, resulting in fast training and low memory costs for SNN. Secondly, the spiking mapping units are designed to ensure that the activation values of the ANN are the spiking features. As a result, the activation consistency is guaranteed, and the classification error of the SNN can be optimized by training the ANN branch. Besides, we design an adaptive threshold adjustment (ATA) algorithm to decrease the firing of noisy spikes. Experiment results show that our AC2AS-based models perform well on the benchmark datasets (CIFAR10, CIFAR100, and Tiny-ImageNet). Moreover, the AC2AS achieves comparable accuracy under 0.625× time steps, 0.377× training time, 0.27× GPU memory costs, and 0.33× spike activities of the Spike-based BP model. The code is available at https://github.com/TJXTT/AC2ASNN.

Deterministic Magnetization Reversal in Synthetic Antiferromagnets using Natural Light.

Traditional current-driven spintronics is limited by localized heating issues and large energy consumption, restricting their data storage density and operation speed. Meanwhile, voltage-driven spintronics with much lower energy dissipation also suffers from charge-induced interfacial corrosion. Thereby finding a novel way of tuning ferromagnetism is crucial for spintronics with energy-saving and good reliability. Here, a visible light tuning of interfacial exchange interaction via photoelectron doping into synthetic antiferromagnetic heterostructure of CoFeB/Cu/CoFeB/PN Si substrate is demonstrated. Then, a complete, reversible magnetism switching between antiferromagnetic (AFM) and ferromagnetic (FM) states with visible light on and off is realized. Moreover, a visible light control of 180° deterministic magnetization switching with a tiny magnetic bias field is achieved. The magnetic optical Kerr effect results further reveal the magnetic domain switching pathway between AFM and FM domains. The first-principle calculations conclude that the photoelectrons fill in the unoccupied band and raise the Fermi energy, which increases the exchange interaction. Lastly, a prototype device with visible light control of two states switching with a 0.35% giant magnetoresistance ratio change (maximal 0.4%), paving the way toward fast, compact, and energy-efficient solar-driven memories is fabricated.

Efficient Edge-Preserving Multi-View Stereo Network for Depth Estimation

Over the years, learning-based multi-view stereo methods have achieved great success based on their coarse-to-fine depth estimation frameworks. However, 3D CNN-based cost volume regularization inevitably leads to over-smoothing problems at object boundaries due to its smooth properties. Moreover, discrete and sparse depth hypothesis sampling exacerbates the difficulty in recovering the depth of thin structures and object boundaries. To this end, we present an Efficient edge-Preserving multi-view stereo Network (EPNet) for practical depth estimation. To keep delicate estimation at details, a Hierarchical Edge-Preserving Residual learning (HEPR) module is proposed to progressively rectify the upsampling errors and help refine multi-scale depth estimation. After that, a Cross-view Photometric Consistency (CPC) is proposed to enhance the gradient flow for detailed structures, which further boosts the estimation accuracy. Last, we design a lightweight cascade framework and inject the above two strategies into it to achieve better efficiency and performance trade-offs. Extensive experiments show that our method achieves state-of-the-art performance with fast inference speed and low memory usage. Notably, our method tops the first place on challenging Tanks and Temples advanced dataset and ETH3D high-res benchmark among all published learning-based methods. Code will be available at https://github.com/susuwj/EPNet.

Fast, noise-free atomic optical memory with 35-percent end-to-end efficiency

Coherent optical memories will likely play an important role in future quantum communication networks. Among the different platforms, memories based on ladder-type orbital transitions in atomic gasses offer high bandwidth (>100 MHz), continuous (on-demand) readout, and low-noise operation. Here we report on an upgraded setup of our previously-reported fast ladder memory, with improved efficiency and lifetime, and reduced noise. The upgrade employs a stronger control field, wider signal beam, reduced atomic density, higher optical depth, annular optical-pumping beam, and weak dressing of an auxiliary orbital to counteract residual Doppler-broadening. For a 2 ns-long pulse, we demonstrate 53% internal efficiency, 35% end-to-end efficiency, 3 × 10−5 noise photons per pulse, and a 1/e lifetime of 108 ns. This combination of performances is a record for continuous-readout memories.

High-Density Nanopatterning of SiGeAsTe Chalcogenide as Ovonic Threshold Switch Selectors for Memory Applications

Cross-point memory architecture involves high-density, fast yet cost-effective storage class memories (SCM). As a result, ovonic threshold switches (OTS) are extensively researched as they are needed in series with the SCM-based memory element for its precise functioning. In addition, these OTS-based devices exhibit the potential to be self-selecting memories too. Hence, its integration-friendly patterning becomes pivotal for large-scale manufacturing. Chalcogenide-based alloys are often the desired candidate for OTS. Here, we study the patterning of one of the widely investigated chalcogenide alloys─SiGeAsTe. While the elements of the alloy generate highly volatile halogenated etch by-products, this material is also prone to severe etch-induced lateral undercuts and material damage. Thus, its high-density patterning is challenging. A conventional method to minimize this undercut damage is to increase the hydrocarbon passivation during the etch process to prevent the exposure of the chalcogenide’s sidewalls (SWs) to any type of etchant. Inherently, this leads to the tapering of the OTS-based pillars, consequently becoming an impediment to their scaling at tighter pitches. Here, we clearly establish that OTS-based devices with well-preserved chemical and morphological properties can be fabricated using lower hydrocarbon passivation by tuning the ion energies during etching. Higher ion energies ensure better etch directionality. Concurrently, a well-controlled high energy ion bombardment is likely to enhance deposition of the passivating hydrocarbon from the etch front on to the chalcogenide’s SWs. This is achieved without any unwanted redeposition from underlying layers in the OTS-based stack. We successfully demonstrate high-density nanopatterning of chalcogenide-based OTS devices of diameter ∼65 nm at 200 nm wide pitch. This is also effectively reproduced on 300 mm semiconductor wafers. The electrically measured devices exhibit low leakage and high endurance. This distinctive yet uncomplicated etch tailoring is also discussed to be scalable for pitches <200 nm and for other chalcogenide compositions.

Tracking online low-rank approximations of higher-order incomplete streaming tensors

In this paper, we propose two new provable algorithms for tracking online low-rank approximations of high-order streaming tensors with missing data. The first algorithm, dubbed adaptive Tucker decomposition (ATD), minimizes a weighted recursive least-squares cost function to obtain the tensor factors and the core tensor in an efficient way, thanks to an alternating minimization framework and a randomized sketching technique. Under the canonical polyadic (CP) model, the second algorithm, called ACP, is developed as a variant of ATD when the core tensor is imposed to be identity. Both algorithms are low-complexity tensor trackers that have fast convergence and low memory storage requirements. A unified convergence analysis is presented for ATD and ACP to justify their performance. Experiments indicate that the two proposed algorithms are capable of streaming tensor decomposition with competitive performance with respect to estimation accuracy and runtime on both synthetic and real data.