FPGA Optimized Accelerator of DCNN with Fast Data Readout and Multiplier Sharing Strategy

With the continuous development of deep learning,Deep Convolutional Neural Network(DCNN)has attracted wide attention in the industry due to its high accuracy in image classification.Compared with other DCNN hard-ware deployment platforms,Field Programmable Gate Array(FPGA)has the advantages of being programmable,low power consumption,parallelism,and low cost.However,the enormous amount of calculation of DCNN and the limited logic capacity of FPGA restrict the energy efficiency of the DCNN accelerator.The traditional sequential sliding window method can improve the throughput of the DCNN accelerator by data multiplexing,but this method’s data multiplexing rate is low because it repeatedly reads the data between rows.This paper proposes a fast data readout strategy via the circular sliding window data reading method,it can improve the multiplexing rate of data between rows by optimizing the memory access order of input data.In addition,the multiplication bit width of the DCNN accelerator is much smaller than that of the Digital Signal Processing(DSP)on the FPGA,which means that there will be a waste of resources if a multiplication uses a single DSP.A multiplier sharing strategy is proposed,the multiplier of the accelerator is customized so that a single DSP block can complete multiple groups of 4,6,and 8-bit signed multiplication in parallel.Finally,based on two strategies of appeal,an FPGA optimized accelerator is proposed.The accelerator is customized by Verilog language and deployed on Xilinx VCU118.When the accelerator recognizes the CIRFAR-10 dataset,its energy efficiency is 39.98 GOPS/W,which provides 1.73×speedup energy efficiency over previous DCNN FPGA accelerators.When the accelerator recognizes the IMAGENET dataset,its energy efficiency is 41.12 GOPS/W,which shows 1.28×−3.14×energy efficiency compared with others.

A Review of Computing with Spiking Neural Networks

Artificial neural networks(ANNs)have led to landmark changes in many fields,but they still differ significantly fromthemechanisms of real biological neural networks and face problems such as high computing costs,excessive computing power,and so on.Spiking neural networks(SNNs)provide a new approach combined with brain-like science to improve the computational energy efficiency,computational architecture,and biological credibility of current deep learning applications.In the early stage of development,its poor performance hindered the application of SNNs in real-world scenarios.In recent years,SNNs have made great progress in computational performance and practicability compared with the earlier research results,and are continuously producing significant results.Although there are already many pieces of literature on SNNs,there is still a lack of comprehensive review on SNNs from the perspective of improving performance and practicality as well as incorporating the latest research results.Starting from this issue,this paper elaborates on SNNs along the complete usage process of SNNs including network construction,data processing,model training,development,and deployment,aiming to provide more comprehensive and practical guidance to promote the development of SNNs.Therefore,the connotation and development status of SNNcomputing is reviewed systematically and comprehensively from four aspects:composition structure,data set,learning algorithm,software/hardware development platform.Then the development characteristics of SNNs in intelligent computing are summarized,the current challenges of SNNs are discussed and the future development directions are also prospected.Our research shows that in the fields of machine learning and intelligent computing,SNNs have comparable network scale and performance to ANNs and the ability to challenge large datasets and a variety of tasks.The advantages of SNNs over ANNs in terms of energy efficiency and spatial-temporal data processing have been more fully exploited.And the development of programming and deployment tools has lowered the threshold for the use of SNNs.SNNs show a broad development prospect for brain-like computing.

Multilayer doped-GeSe OTS selector for improved endurance and threshold voltage stability

Selector devices are indispensable components of large-scale memristor array systems.The thereinto,ovonic threshold switching(OTS)selector is one of the most suitable candidates for selector devices,owing to its high selectivity and scalability.However,OTS selectors suffer from poor endurance and stability which are persistent tricky problems for applica-tion.Here,we report on a multilayer OTS selector based on simple GeSe and doped-GeSe.The experimental results show im-proving selector performed extraordinary endurance up to 1010 and the fluctuation of threshold voltage is 2.5%.The reason for the improvement may lie in more interface states which strengthen the interaction among individual layers.These develop-ments pave the way towards tuning a new class of OTS materials engineering,ensuring improvement of electrical perform-ance.

Sigmatropic rearrangements of B(MIDA)-propargylic alcohols towards the diverse synthesis of α-functionalized organoborons

α-Functionalized organoborons are useful building blocks and key structural elements in functional molecules.Their previous synthesis relied on the famous Matteson reaction or the late-stage borylative modification of alkynes or alkenes.Recently,the synthetic transformation of borylated building blocks offers another useful strategy and is currently actively explored.We report herein that B(MIDA)-propargylic alcohols(BPAs) are a useful type of borylated building blocks.Bearing two complementary functional group handles(alkyne and hydroxyl) in close proximity,the redox-neutral [3,3] and [2,3] sigmatropic rearrangements of BPAs allow the efficient synthesis of several types of α-functionalized boronates,including α,β-unsaturated acylborons,α-S/P-substituted allenylborons,boryl-substituted thiazoles and a borylated α,β-unsaturated hydrazine,some of which are otherwise challenging targets using other synthetic methods.

Synthesis of allenyl-B(MIDA)via hydrazination/fragmentation reaction of B(MIDA)-propargylic alcohol

Allenylboronates represent a very intriguing class of organoborons but are challenging to synthesis.In addition,these compounds are typically unstable,rendering the separation difficult.We report herein a practical and concise route to a new class of stable,easy-separable allenyl B(MIDA)via a hydrazination/fragmentation of B(MIDA)-propargylic alcohols.The synthesis of optically active allenyl B(MIDA)was also achieved.Interesting reactivity of the resulting product was observed.

Feasibility and physics potential of detecting ^(8)B solar neutrinos at JUNO

The Jiangmen Underground Neutrino Observatory(JUNO)features a 20 kt multi-purpose underground liquid scintillator sphere as its main detector.Some of JUNO's features make it an excellent location for^8B solar neutrino measurements,such as its low-energy threshold,high energy resolution compared with water Cherenkov detectors,and much larger target mass compared with previous liquid scintillator detectors.In this paper,we present a comprehensive assessment of JUNO's potential for detecting^8B solar neutrinos via the neutrino-electron elastic scattering process.A reduced 2 MeV threshold for the recoil electron energy is found to be achievable,assuming that the intrinsic radioactive background^(238)U and^(232)Th in the liquid scintillator can be controlled to 10^(-17)g/g.With ten years of data acquisition,approximately 60,000 signal and 30,000 background events are expected.This large sample will enable an examination of the distortion of the recoil electron spectrum that is dominated by the neutrino flavor transformation in the dense solar matter,which will shed new light on the inconsistency between the measured electron spectra and the predictions of the standard three-flavor neutrino oscillation framework.IfDelta m^(2)_(21)=4.8times10^(-5);(7.5times10^(-5))eV^(2),JUNO can provide evidence of neutrino oscillation in the Earth at approximately the 3sigma(2sigma)level by measuring the non-zero signal rate variation with respect to the solar zenith angle.Moreover,JUNO can simultaneously measureDelta m^2_(21)using^8B solar neutrinos to a precision of 20% or better,depending on the central value,and to sub-percent precision using reactor antineutrinos.A comparison of these two measurements from the same detector will help understand the current mild inconsistency between the value of Delta m^2_(21)reported by solar neutrino experiments and the KamLAND experiment.

JUNO sensitivity on proton decay p→νK^(+)searches

The Jiangmen Underground Neutrino Observatory(JUNO)is a large liquid scintillator detector designed to explore many topics in fundamental physics.In this study,the potential of searching for proton decay in the p→νK^(+)mode with JUNO is investigated.The kaon and its decay particles feature a clear three-fold coincidence signature that results in a high efficiency for identification.Moreover,the excellent energy resolution of JUNO permits suppression of the sizable background caused by other delayed signals.Based on these advantages,the detection efficiency for the proton decay via p→νK^(+)is 36.9%±4.9%with a background level of 0.2±0.05(syst)±0.2(stat)events after 10 years of data collection.The estimated sensitivity based on 200 kton-years of exposure is 9.6×1033 years,which is competitive with the current best limits on the proton lifetime in this channel and complements the use of different detection technologies.

Highly improved resistive switching performances of the self-doped Pt/HfO_2:Cu/Cu devices by atomic layer deposition

Metal-oxide electrochemical metallization (ECM) memory is a promising candidate for the next generation nonvolatile memory.But this memory suffers from large dispersion of resistive switching parameters due to the intrinsic randomness of the conductive filament. In this work, we have proposed a self-doping approach to improve the resistive switching characteristics. The fabricated Pt/HfO_2:Cu/Cu device shows outstanding nonvolatile memory properties, including high uniformity, good endurance, long retention and fast switching speed. The results demonstrate that the self-doping approach is an effective method to improve the metal-oxide ECM memory performances and the self-doped Pt/HfO_2:Cu/Cu device has high potentiality for the nonvolatile memory applications in the future.

Recent advances of allenes in the first-row transition metals catalyzed C-H activation reactions

Transition-metal-catalyzed C–H activation reaction has proven to be a powerful and efficient tool for the formation of diverse C-C and C–X bond and construction of functional complex molecules.From the viewpoint of sustainable chemistry,the first-row transition metals,such as Mn,Fe,Co,Ni and Cu,have been recognized as cheap,environmentally friendly and reactively effective catalysts for a number of C-H functionalization reactions.However,compared with the commonly used alkenes and alkynes in the first-row transition-metal-catalyzed C–H activations,considerable achievements have just been made by the use of structurally unique and reactively rich allenes as coupling partners in recent years.This review summarizes the recent progress of the first-row transition-metal-catalyzed C–H activations with allenes.

Synthesis of difluoromethylated benzylborons via rhodium(Ⅰ)-catalyzed fluorine-retainable hydroboration of gem-difluoroalkenes

The synthesis of bo rylated orga nofluorines is of great interest due to their potential values as synthons in modular construction of fluorine-containing molecules.Reported herein is a rhodium-catalyzed hydrobo ration of arylgem-difluoroalkenes leading to a series of α-difluoromethylated benzylborons.The use of cationic rhodium catalyst and a biphosphine ligand with large bite angle was crucial for reactivity by offering good regioselectivity and diminishing the undesired β-F elimination.Preliminary derivatizations of the products were conducted to showcase the utility of this protocol.

A boryl-migratory semipinacol rearrangement

The semipinacol rearrangement is one of the classic yet useful synthetic tools in organic synthesis.However,semipinacol rearrangements involving heteroatom-migration are rare.Reported herein is a boryl-migratory semipinacol rearrangement of α-hydroxyallylboronates and α-hydroxypropargylboronates triggered by diverse halogen-,oxygen-,sulfur-and seleniumcontaining electrophiles.The protocol leads to a mild and facile access to organoborons bearing valuable functionalities.The σ(C-B)hyperconjugation is believed to be the key factor that leads to the observed exclusive chemoselectivity and enhanced reactivity.Synthetic utilities of the formed products were demonstrated.

Sub-percent precision measurement of neutrino oscillation parameters with JUNO

JUNO is a multi-purpose neutrino observatory under construction in the south of China.This publication presents new sensitivity estimates for the measurement of the △m_(31)^(2),△m_(21)^(2),sin^(2)θ_(12),and sin^(2)θ_(13) oscillation parameters using reactor antineutrinos,which is one of the primary physics goals of the experiment.The sensitivities are obtained using the best knowledge available to date on the location and overburden of the experimental site,the nuclear reactors in the surrounding area and beyond,the detector response uncertainties,and the reactor antineutrino spectral shape constraints expected from the TAO satellite detector.It is found that the △m_(21)^(2) and sin^(2)θ_(12) oscillation parameters will be determined to 0.5%precision or better in six years of data collection.In the same period,the △m_(31)^(2) parameter will be determined to about 0.2%precision for each mass ordering hypothesis.The new precision represents approximately an order of magnitude improvement over existing constraints for these three parameters.

Voltage-control oscillator based on Pt/C/NbO_x/TiN device with highly improved threshold switching performances

Oscillator is a common key component of electronic systems.The periodic signal produced by the oscillator is generally required in various applications,such as the electronic system clock,electronic neurons,and the true random number generator system[1-6].Capacitors and inductors are usually utilized to generate periodic waveforms in the traditional oscillator,which greatly reduces integration and cannot be packaged into chips[2].Oscillators based on a memristor have been proposed as a solution to these issues[7-14].The memristor has attracted great attention and has been widely applied in many fields,such as memory,com?puting,security,etc.