检索结果-内蒙古大学图书馆

Reinventing Memory system Design for Many-Accelerator architecture

Journal of computer Science & Technology 2014年第2期29卷 273-280页

作者：王颖张磊韩银和李华伟 State Key Laboratory of Computer Architecture Institute of Computing TechnologyChinese Academy of Sciences University of Chinese Academy of Sciences

The many-accelerator architecture, mostly composed of general-purpose cores and accelerator-like function units （FUs）, becomes a great alternative to homogeneous chip multiprocessors （CMPs） for its superior power-efficiency. However, the emerging many-accelerator processor shows a much more complicated memory accessing pattern than general purpose processors （GPPs） because the abundant on-chip FUs tend to generate highly-concurrent memory streams with distinct locality and bandwidth demand. The disordered memory streams issued by diverse accelerators exhibit a mutual- interference behavior and cannot be efficiently handled by the orthodox main memory interface that provides an inflexible data fetching mode. Unlike the traditional DRAM memory, our proposed Aggregation Memory system （AMS） can function adaptively to the characterized memory streams from different FUs, because it provides the FUs with different data fetching sizes and protects their locality in memory access by intelligently interleaving their data to memory devices through sub-rank binding. Moreover, AMS can batch the requests without sub-rank conflict into a read burst with our optimized memory scheduling policy. Experimental results from trace-based simulation show both conspicuous performance boost and energy saving brought by AMS.

关键词： many-accelerator chip multiprocessor memory general purpose processor

来源：评论

学校读者我要写书评

暂无评论

A High-Performance and Cost-Efficient Interconnection Network for High-Density Servers

引用

Journal of computer Science & Technology 2014年第2期29卷 281-292页

作者：包雯韬付斌章陈明宇张立新 State Key Laboratory of Computer Architecture Institute of Computing TechnologyChinese Academy of Sciences University of Chinese Academy of Sciences

The high-density server is featured as low power, low volume, and high computational density. With the rising use of high-density servers in data-intensive and large-scale web applications, it requires a high-performance and cost-efficient intra-server interconnection network. Most of state-of-the-art high-density servers adopt the fully-connected intra-server network to attain high network performance. Unfortunately, this solution costs too much due to the high degree of nodes. In this paper, we exploit the theoretically optimized Moore graph to interconnect the chips within a server. Accounting for the suitable size of applications, a 50-size Moore graph, called Hoffman-Singleton graph, is adopted. In practice, multiple chips should be integrated onto one processor board, which means that the original graph should be partitioned into homogeneous connected subgraphs. However, the existing partition scheme does not consider above problem and thus generates heterogeneous subgraphs. To address this problem, we propose two equivalent-partition schemes for the Hoffman-Singleton graph. In addition, a logic-based and minimal routing mechanism, which is both time and area efficient, is proposed. Finally, we compare the proposed network architecture with its counterparts, namely the fully-connected, Kautz and Torus networks. The results show that our proposed network can achieve competitive performance as fully-connected network and cost close to Torus.

关键词： high-density server interconnection network Moore graph Hoffman-Singleton graph equivalent partition

来源：评论

学校读者我要写书评

暂无评论

Optimizing Parallel S_n Sweeps on Unstructured Grids for Multi-Core Clusters

引用

Journal of computer Science & Technology 2013年第4期28卷 657-670页

作者：闫洁谭光明孙凝晖 State Key Laboratory of Computer Architecture Institute of Computing Technology Chinese Academy of Sciences University of Chinese Academy of Sciences

In particle transport simulations, radiation effects are olden described by the discrete ordinates （Sn） form of Boltzmann equation. In each ordinate direction, the solution is computed by sweeping the radiation flux across the grid. Parallel Sn sweep on an unstructured grid can be explicitly modeled as topological traversal through an equivalent directed acyclic graph （DAG）, which is a data-driven algorithm. Its traditional design using MPI model results in irregular communication of massive short messages which cannot be efficiently handled by MPI runtime. Meanwhile, in high-end HPC cluster systems, multicore has become the standard processor configuration of a single node. The traditional data-driven algorithm of Sn sweeps has not exploited potential advantages of multi-threading of multicore on shared memory. These advantages, however, as we shall demonstrate, could provide an elegant solution resolving problems in the previous MPI-only design. In this paper, we give a new design of data-driven parallel Sn sweeps using hybrid MPI and Pthread programming, namely Sweep-H, to exploit hierarchical parallelism of processes and threads. With special multi-threading techniques and vertex schedule policy, Sweep-H gets more efficient communication and better load balance. We further present an analytical performance model for Sweep-H to reveal why and when it is advantageous over former MPI counterpart. On a 64-node multicore cluster system with 12 cores per node, 768 cores in total, Sweep-H achieves nearly linear scalability for moderate problem sizes, and better absolute performance than the previous times speedup on 64 nodes）. MPI algorithm on more than 16 nodes （by up to two

关键词： parallel Sn sweep unstructured grid data-driven algorithm

来源：评论

学校读者我要写书评

暂无评论

Accelerating DNN-based 3D point cloud processing for mobile computing

引用

Science China(Information Sciences) 2019年第11期62卷 40-50页

作者： Bosheng LIU Xiaoming CHEN Yinhe HAN Jiajun LI Haobo XU Xiaowei LI State Key Laboratory of Computer Architecture Institute of Computing TechnologyChinese Academy of Sciences University of Chinese Academy of Sciences

3D point cloud data, which are produced by various 3 D sensors such as LIDAR and stereo cameras, have been widely deployed by industry leaders such as Google, Uber, Tesla, and Mobileye, for mobile robotic applications such as autonomous driving and humanoid robots. Point cloud data, which are composed of reliable depth information, can provide accurate location and shape characteristics for scene understanding, such as object recognition and semantic segmentation. However, deep neural networks(DNNs), which directly consume point cloud data, are particularly computation-intensive because they have to not only perform multiplication-and-accumulation(MAC) operations but also search neighbors from the irregular 3 D point cloud data. Such a task goes beyond the capabilities of general-purpose processors in realtime to figure out the solution as the scales of both point cloud data and DNNs increase from application to application. We present the first accelerator architecture that dynamically configures the hardware onthe-fly to match the computation of both neighbor point search and MAC computation for point-based DNNs. To facilitate the process of neighbor point search and reduce the computation costs, a grid-based algorithm is introduced to search neighbor points from a local region of grids. Evaluation results based on the scene recognition and segmentation tasks show that the proposed design harvests 16.4× higher performance and saves 99.95% of energy than an NVIDIA Tesla K40 GPU baseline in point cloud scene understanding applications.

关键词： deep neural network acceleration point cloud data neighbor point search mobile robotics hardware architecture

来源：评论

学校读者我要写书评

暂无评论

Prescribed Performance Path-Following Control for Rotor-Assisted Vehicles via an Improved Reinforcement Learning Mechanism

引用

IEEE Transactions on Neural Networks and Learning systems 2025年 PP卷 PP页

作者： Zhang, Guoqing Li, Zhihao Li, Jiqiang Zhang, Weidong Qiu, Bin Navigation College Liaoning Dalian116026 China State Key Laboratory of Maritime Technology and Safety Liaoning Dalian116026 China Hainan University School of Information and Communication Engineering Hainan Haikou570228 China Shanghai Jiao Tong University Department of Automation Shanghai200240 China Guilin University of Technology School of Computer Science and Engineering Guangxi Key Laboratory of Embedded Technology and Intelligent System Guilin541004 China

This article investigates an adaptive prescribed performance path-following control algorithm for rotor-assisted vehicles, incorporating reinforcement learning (RL) to execute energy-saving cruising missions. For obtaining a high-performance path-following controller, a concise prescribed performance control (PPC) algorithm is designed to tightly constrain the output errors within the defined boundaries, while a shifting function is introduced to solve the problem of initial condition restrictions. Furthermore, through integrating the Backstepping method and the optimal control technique, an improved RL with the form of actor–critic neural networks (AC-NNs) is proposed to offer an innovative approach to the challenges of the model uncertainties and external disturbances. In this approach, the actor NN is employed to create an appropriate control policy, while the critic NN is aimed at evaluating the cost-to-go function to modify the system action. Semi-global uniform ultimate bounded (SGUUB) stable properties of the proposed algorithm are guaranteed via the Lyapunov theory. Finally, the superiority and feasibility of the proposed algorithm are verified by two numerical experiments. © 2012 IEEE.

关键词： Cost functions

来源：评论

学校读者我要写书评

暂无评论

Diffusion-Based Model for Parametric Ki Generation From Total-Body Dynamic PET of Short-Duration Scan

引用

IEEE Transactions on Radiation and Plasma Medical Sciences 2025年

作者： Wen, Meiyuan Wu, Yaping Huang, Zhenxing He, Xiangjian Tesema, Fiseha B. Chen, Zixiang Gao, Yunlong Li, Wenbo Yang, Xinlan Yang, Yongfeng Zheng, Hairong Liang, Dong Wang, Meiyun Hu, Zhanli Chinese Academy of Sciences Research Center for Medical AI Shenzhen Institute of Advanced Technology China State Key Laboratory of Biomedical Imaging Science and System Key Laboratory of Biomedical Imaging Science and System Chinese Academy of Sciences China University of Nottingham Ningbo China School of Computer Science Ningbo315104 China Henan Provincial People's Hospital People's Hospital of Zhengzhou University Department of Medical Imaging Zhengzhou450003 China United Imaging Healthcare Group Central Research Institute Shanghai201807 China

Dynamic positron emission tomography (PET) parametric imaging typically requires a 60-minute acquisition period, causing patient discomfort and reducing clinical efficiency. This study explores the feasibility of generating parametric Ki images from 10-minute dynamic PET images acquired in the early or late scanning phases employing a multi-channel feature fusion cold sampling (MCFFCoS) framework. PET data from 103 patients are acquired using the uEXPLORER total-body PET/CT scanner during 60-minute scans. This study conducts deep learning experiments, taking early-phase or late-phase PET images as input, respectively. The generated Ki images are evaluated by visual quality and quantitative metrics, including root mean squared error (RMSE), structural similarity index (SSIM), and peak signal-to-noise ratio (PSNR). Volumes of interest (VOIs) analysis is performed using linear regression and Bland-Altman plots. In the quantitative evaluation of total-body data, the parametric Ki images generated from late-phase PET data generally outperform those derived from early-phase data. The analysis of VOIs indicates that the appropriate scanning protocol for PET parametric imaging may vary for different body regions. The deep learning approach is able to generate high-quality parametric Ki images from 10-minute dynamic PET scans, bypassing the requirements of long acquisition time for the estimation of blood input function in kinetic modeling. © 2017 IEEE.

关键词： Positron emission tomography

来源：评论

学校读者我要写书评

暂无评论

YOLO-LLTS: Real-Time Low-Light Traffic Sign Detection via Prior-Guided Enhancement and Multi-Branch Feature Interaction

arXiv

引用

arXiv 2025年

作者： Lin, Ziyu Wu, Yunfan Ma, Yuhang Chen, Junzhou Zhang, Ronghui Wu, Jiaming Yin, Guodong Lin, Liang Guangdong Key Laboratory of Intelligent Transportation System School of intelligent systems engineering Sun Yat-sen University Guangzhou510275 China Department of Architecture and Civil Engineering Chalmers University of Technology Sven Hultins gata 6 GothenburgSE-412 96 Sweden School of Mechanical Engineering Southeast University Nanjing211189 China School of Computer Science and Engineering Sun Yat-sen University Guangzhou510275 China

Detecting traffic signs effectively under low-light conditions remains a significant challenge. To address this issue, we propose YOLO-LLTS, an end-to-end real-time traffic sign detection algorithm specifically designed for low-light environments. Firstly, we introduce the High-Resolution Feature Map for Small Object Detection (HRFM-TOD) module to address indistinct small-object features in low-light scenarios. By leveraging high-resolution feature maps, HRFM-TOD effectively mitigates the feature dilution problem encountered in conventional PANet frameworks, thereby enhancing both detection accuracy and inference speed. Secondly, we develop the Multi-branch Feature Interaction Attention (MFIA) module, which facilitates deep feature interaction across multiple receptive fields in both channel and spatial dimensions, significantly improving the model’s information extraction capabilities. Finally, we propose the Prior-Guided Enhancement Module (PGFE) to tackle common image quality challenges in low-light environments, such as noise, low contrast, and blurriness. This module employs prior knowledge to enrich image details and enhance visibility, substantially boosting detection performance. To support this research, we construct a novel dataset, the Chinese Nighttime Traffic Sign Sample Set (CNTSSS), covering diverse nighttime scenarios, including urban, highway, and rural environments under varying weather conditions. Experimental evaluations demonstrate that YOLO-LLTS achieves state-of-the-art performance, outperforming the previous best methods by 2.7% mAP50 and 1.6% mAP50:95 on TT100K-night, 1.3% mAP50 and 1.9% mAP50:95 on CNTSSS, and achieving superior results on the CCTSDB2021 dataset. Moreover, deployment experiments on edge devices confirm the real-time applicability and effectiveness of our proposed approach. Copyright © 2025, The Authors. All rights reserved.

关键词： Traffic signs

来源：评论

学校读者我要写书评

暂无评论

Resource scheduling virtualization in service-oriented future Internet architecture

引用

The Journal of China Universities of Posts and Telecommunications 2015年第4期22卷 92-100页

作者： Li Lan Chen Jianya Cui Hongyan Huang Tao Liu Yunjie State Key Laboratory of Networking and Switching Technology Beijing University of Posts and Telecommunications Beijing Key Laboratory of Network System Architecture and Convergence Beijing University of Posts and Telecommunications

Service-oriented future internet architecture（SOFIA） is a clean-slate network architecture. In SOFIA, a service request is mainly processed through service resolution and network resource allocation. To realize the network resource allocation, we reference the idea of network virtualization and propose resource scheduling virtualization. In resource scheduling virtualization, a service request is abstracted as a virtual network（VN） and the network resources are allocated by mapping the VN onto the physical network. Resource scheduling virtualization provides centralized resource scheduling control within an autonomous system（AS） and achieves better controllability compared with the distributed schemes. Besides, resource scheduling virtualization supports multi-site selection as well. Meanwhile, we propose a collection of resource scheduling algorithms based on maximum resource tree（MRT） adapting to different scenarios. According to the simulation results, the proposed algorithms show good performance on the key metrics, such as acceptance ratio, revenue, cost and utilization. Moreover, the simulation results reveal that our algorithm is more efficient than the traditional ones.

关键词： SOFIA resource scheduling virtualization virtual network embedding maximum resource tree（MRT）

来源：评论

学校读者我要写书评

暂无评论

Interference of human's ankle to the channel characteristics of near surface WUSN

引用

The Journal of China Universities of Posts and Telecommunications 2014年第1期21卷 36-42页

作者： LI Li WU Wei CHEN Jian-ya LIU Yun-jie State Key Laboratory of Networking and Switching Technology Beijing University of Posts and Telecommunications Beijing Key Laboratory of Network System Architecture and Convergence Beijing University of Posts and Telecommunications

Because of its potential applications in agriculture, environment monitoring and so on, wireless underground sensor network（WUSN） has been researched more and more extensively in recent years. The main and most important difference of WUSN to terrestrial wireless sensor network（WSN） is the channel characteristics, which determines the design methodology of it. In this paper, the propagation character of electromagnetic（EM） wave in the near surface WUSN is analyzed, as well as the path loss model of it is given. In addition, the influence of human＇s ankle to the channel characteristics of near surface WUSN is investigated by electromagnetic theory analysis, simulation and experiment. A novel path loss model of near surface WUSN which takes the interference of human＇s ankle into consideration is proposed. It is verified that the existing of human above the WUSN system may cause additional attenuation to the signal of near surface WUSN which propagates as lateral wave along the ground. Moreover, the relation of the attenuation and operating frequency is deduced, which gives a reference to extend the frequency band applied in WUSN.

关键词： WUSN lateral wave human tissue path loss model EM wave attenuation

来源：评论

学校读者我要写书评

暂无评论

Temperature behaviors of transparent solar PV panels under various operating modes: an experimental and numerical study

引用

Renewable Energy 2025年 250卷

作者： Dong, Qichang Wu, Xuejiao Song, Ye Du, Yaohan Qi, Jiacheng Huang, Lin Li, Wengui Huang, Youbo Shi, Long State Key Laboratory of Fire Science University of Science and Technology of China Anhui Hefei230026 China GCL System Integration Technology Co Ltd. Jiangsu Suzhou China School of Architecture and Civil Engineering Chengdu University Sichuan Chengdu610106 China Centre for Infrastructure Engineering and Safety School of Civil and Environmental Engineering The University of New South Wales NSW2052 Australia College of Safety Engineering Chongqing University of Science and Technology Chongqing401331 China

Temperature is the key factor in determining the practical performance and energy production prediction of solar photovoltaic (PV) panels. Transparent solar PV panels, in contrast to traditional opaque solar PV panels, have grown rapidly. However, the temperature behaviors of transparent PV panels and the effects of operating modes (i.e., operating and non-operating) on their temperature behaviors remain unidentified. Therefore, this study investigated the temperature behaviors of transparent PV panels (i.e., hollow and non-hollow) and the effect of different operating modes on their thermal performance. The hollow layer is the primary component influencing the temperature behaviors of hollow and non-hollow PV panels. At a solar irradiance of 1000 W/m2, the hollow layer resulted in the steady-state temperature and the mean temperature rise rate on the exposed surface of hollow PV panels being 12.0 % and 5.6 % lower, respectively, compared to non-hollow PV panels. Hollow PV panels have better heat resistance under high solar irradiance conditions. Experimental results showed that the effects of the operating mode on the rate of change of the PV panel surface temperature can be ignored. A theoretical model was also developed for the first time, and its predictions obeyed well with the experimental data. These findings and the developed theoretical model offer a foundation for forecasting the temperature and energy generation rate of transparent solar PV panels, helping with future policymaking. © 2025 Elsevier Ltd

关键词： Energy generation Online or offline Solar irradiance Temperature characteristics Transparent solar PV panels

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：