检索结果-内蒙古大学图书馆

20th Annual Conference on Computer Engineering and technology, NCCET 2016

作者： Chen, Jicheng Zhao, Yaqian Shi, Hongzhi Li, Yihan State Key Laboratory of High-End Server and Storage Technology Inspur Group Company Limited Beijing China

ISBN: (纸本)9789811031588

Directory-based coherence is widely used in modern CMP systems. As the number of cores increases, it is increasingly deemed as the only candidate for on-chip cache coherence maintaining. However, limitations of traditional coherence directory pose serious challenges to deal with the ever-increasing size of the system. The hardware overhead and redundant message broadcasting problems dramatically degrade the scalability and performance of the system. In this paper, a space-efficient coherence directory BFDir is proposed. The directory dramatically reduces the directory size as the share list is shortened by Bloom filter. Also, it does not incur message broadcasting as that in limited directories. The evaluation results show, for 32-core CMP systems, compared to full-map directory, 59% overhead of share list can be avoided at the expense of 2.77% performance loss on average;compared to 16-bit coarse directory, 22% overhead of share list can be avoided at the expense of 0.16% average performance loss on average;compared to 8-bit coarse directory, 48% invalid messages are saved and the performance is improved by 2.31%. © Springer Nature Singapore Pte Ltd. 2016.

关键词： Bandpass filters

来源：评论

学校读者我要写书评

暂无评论

HPDedup: A hybrid prioritized data deduplication mechanism for primary storage in the cloud

arXiv

引用

arXiv 2017年

作者： Wu, Huijun Wang, Chen Fu, Yinjin Sakr, Sherif Zhu, Liming Lu, Kai Data61 CSIRO University of New South Wales Australia PLA University of Science and Technology China Science and Technology on Parallel and Distributed Laboratory State Key Laboratory of High Performance Computing State Key Laboratory of High-end Server & Storage Technology College of Computer National University of Defense Technology Changsha China

—Eliminating duplicate data in primary storage of clouds increases the cost-efficiency of cloud service providers as well as reduces the cost of users for using cloud services. Most existing primary deduplication techniques either use inline caching to exploit locality in primary workloads or use post-processing deduplication running in system idle time to avoid the negative impact on I/O performance. However, neither of them works well in the cloud servers running multiple services or applications for the following two reasons: Firstly, the temporal locality of duplicate data writes may not exist in some primary storage workloads thus inline caching often fails to achieve good deduplication ratio. Secondly, the post-processing deduplication allows duplicate data to be written into disks, therefore does not provide the benefit of I/O deduplication and requires high peak storage capacity. This paper presents HPDedup, a Hybrid Prioritized data Deduplication mechanism to deal with the storage system shared by applications running in co-located virtual machines or containers by fusing an inline and a post-processing process for exact deduplication. In the inline deduplication phase, HPDedup gives a fingerprint caching mechanism that estimates the temporal locality of duplicates in data streams from different VMs or applications and prioritizes the cache allocation for these streams based on the estimation. HPDedup also allows different deduplication threshold for streams based on their spatial locality to reduce the disk fragmentation. The post-processing phase removes duplicates whose fingerprints are not able to be cached due to weak temporal locality from disks. The hybrid deduplication mechanism significantly reduces the amount of redundant data written to the storage system while maintaining inline data writing performance. Our experimental results show that HPDedup clearly outperforms the state-of-the-art primary storage deduplication techniques in terms of inline

关键词： Efficiency

来源：评论

学校读者我要写书评

暂无评论

Fairness scheduling with dynamic priority for multi workflow on heterogeneous systems

Fairness scheduling with dynamic priority for multi workflow...

引用

IEEE International Conference on Cloud Computing and Big Data Analysis (ICCCBDA)

作者： Yuxin Wang Shijie Cao Guan Wang Zhen Feng Chi Zhang He Guo School of Computer Dalian University of Technology Dalian China Liaoning Police College Dalian China State Key Laboratory of High-end Server & Storage Technology Beijing China School of Software Dalian University of Technology Dalian China

ISBN: (纸本)9781509045006

In Heterogeneous Computing Systems, completion time and overall fairness are both crucial to multi workflow scheduling. Unfortunately, most dynamic workflow scheduling algorithms fail to take fairness into account or adapt unreasonable fairness policy, which will probably result in some DAGs' scheduling failure in certain scenarios when they can't complete all the tasks before deadline. Aiming at the need to solve the flaws mentioned above, we propose a novel dynamic workflow scheduling algorithm named FSDP (Fairness Scheduling with Dynamic Priority for Multi Workflow). The algorithm focuses on the deadline of each workflow to achieve a more reasonable fairness when allocating the processors so that those urgent workflows can acquire a higher priority. Also, a new metric, urgency, is proposed to update priority. Experimental results show that it is possible to meet different requirements of dynamic workflows.

关键词： Electronic mail Schedules Computers

来源：评论

学校读者我要写书评

暂无评论

Container oriented job scheduling using linear programming model

Container oriented job scheduling using linear programming m...

引用

International Conference on Information Management (ICIM)

作者： Dong Zhang Bing-Heng Yan Zhen Feng Chi Zhang Yu-Xin Wang Inspur Electronic Information Industry Co. Ltd. Jinan China State Key Laboratory of High-end Server & Storage Technology Beijing China Inspur Electronic Information Industry Co. Ltd. Jinan China School of Computer Science and Technology Dalian University of Technology Dalian China

ISBN: (纸本)9781509063079

With Docker gaining widespread popularity in the recent years, the container scheduler becomes a crucial role for the exploding containerized applications and services. In this work, the container host energy conservation, the container image pulling costs from the image registry to the container hosts and the workload network transition costs from the clients to the container hosts are evaluated in combination. By modeling the scheduling problem as an integer linear programming, an effective and adaptive scheduler is proposed. Impressive cost savings were achieved compared to Docker Swarm scheduler. Moreover, it can be easily integrated into the open-source container orchestration frameworks.

关键词： Containers Job shop scheduling Resource management servers Cloud computing Virtual machining Linear programming

来源：评论

学校读者我要写书评

暂无评论

Capture Global Kad Lookup Traffic

Capture Global Kad Lookup Traffic

引用

2016 IEEE 7th International Conference on Software Engineering and Service Science

作者： Zhibo Liu Qiang Li Dept.of Library Jinan Vocational College State Key Laboratory of High-end Server& Storage Technology

Kad is the most popular P2 P file sharing *** is very difficult to capture all the users’ lookup traffic because of its full distributed *** work could only monitors a subset of the lookup *** this paper,we propose a lightweight method to monitor the global lookup traffic by inserting two ’nearest neighbor’ peers into the peer at the last iteration of Kad lookup *** one of the inserted peer is closer than the last peer definitely,the user’s search request would be redirected to the trap *** experiment performed on eMule 0.50 a shows that this approach could capture most lookup traffic.

关键词： P2P Kad lookup capture

来源：评论

学校读者我要写书评

暂无评论

Efficient detection of dangling pointer error for C/C++ programs

引用

Journal of Physics: Conference Series 2017年第1期887卷

作者： Wenzhe Zhang Science and Technology on Parallel and Distributed Laboratory State Key Laboratory of High Performance Computing State Key Laboratory of High-end Server & Storage Technology College of Computer National University of Defense Technology Changsha PR China

Dangling pointer error is pervasive in C/C++ programs and it is very hard to detect. This paper introduces an efficient detector to detect dangling pointer error in C/C++ programs. By selectively leave some memory accesses unmonitored, our method could reduce the memory monitoring overhead and thus achieves better performance over previous methods. Experiments show that our method could achieve an average speed up of 9% over previous compiler instrumentation based method and more than 50% over previous page protection based method.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Vertical or horizontal scaling for web information systems

引用

International Journal of u- and e- Service, Science and technology 2016年第9期9卷 111-120页

作者： Liu, Wenjuan Zeng, Guosun Xiong, Huanliang Department of Computer Science and Technology Tongji University Shanghai201804 China State Key Laboratory of High-end Server & Storage Technology Jinan250101 China Jiangxi Agricultural University Nanchang330045 China

As the number of Web users and the amount of information produced keep growing rapidly, information systems need to be scaled up constantly. Usually scaling methods of information systems suffer from being chosen by experience, this paper studied on quantitative optimal selection of vertical/horizontal scaling methods. According to investigation and statistics, we assumed the cost growth formula of the key performance parameters reasonably. Then five key parameters of systems were chosen to construct Kiviat graphs and the areas of Kiviat graphs were computed to evaluate system performance. Next, the quantitative function relation between the performance of information system and the cost of system scaling was established, and then three optimal selection algorithms for system scaling were proposed. Finally, some case studies demonstrate that algorithms are feasible and have a great significance for guiding system scaling. © 2016 SERSC.

关键词： Information systems

来源：评论

学校读者我要写书评

暂无评论

The gyrokinetic particle simulation of fusion plasmas on tianhe-2 supercomputer 7

The gyrokinetic particle simulation of fusion plasmas on tia...

引用

7th Workshop on Latest Advances in Scalable Algorithms for Large-Scale Systems, ScalA 2016

作者： Wang, endong Wu, Shaohua Zhang, Qing Liu, Jun Zhang, Wenlu Lin, Zhihong Lu, Yutong Du, Yunfei Zhu, Xiaoqian State Key Laboratory of High-end Server and Storage Technology InspurJinan China Institute of Physics Chinese Academy of Science Beijing China Department of Physics and Astronomy University of California IrvineCA United States Fusion Simulation Center Peking University Beijing China School of Computer Science National University of Defense Technology Changsha China National Supercomputer Center in Guangzhou Sun Yat-sen University Guang Zhou China Academy of Ocean Science and Engineering National University of Defense Technology Changsha China

ISBN: (纸本)9781509052226

We present novel optimizations of the fusion plasmas simulation code, GTC on Tianhe-2 supercomputer. The simulation exhibits excellent weak scalability up to 3072 31S1P Xeon Phi co-processors. An unprecedented up to 5.8× performance improvement is achieved for the GTC on Tianhe-2. An efficient particle exchanging algorithm is developed that simplifies the original iterative scheme to a direct implementation, which leads to a 7.9× performance improvement in terms of MPI communications on 1024 nodes of Tianhe-2. A customized particle sorting algorithm is presented that delivers a 2.0× performance improvement on the co-processor for the kernel relating to the particle computing. A smart offload algorithm that minimizes the data exchange between host and co-processor is introduced. Other optimizations like the loop fusion and vectorization are also presented. © 2016 IEEE.

关键词： Scalability

来源：评论

学校读者我要写书评

暂无评论

19 low mass hypervelocity star candidates from the first data release of the LAMOST survey

引用

Research in Astronomy and Astrophysics 2015年第8期15卷 1364-1377页

作者： Yin-Bi Li A-Li Luo Gang Zhao You-Jun Lu Peng Wei Bing Du Xiang Li Yong-Heng Zhao Zhan-Wen Han Bo Wang Yue Wu Yong Zhang Yong-Hui Hou Yue-Fei Wang Ming Yang Key Laboratory of Optical Astronomy National Astronomical Observatories Chinese Academy of Sciences Beijing 100012 China State Key Laboratory of High-end Server & Storage Technology Jinan 250510 China Key Laboratory for the Structure and Evolution of Celestial Objects Yunnan Observatories Chinese Academy of Sciences Kunming 650216 China Nanjing Institute of Astronomical Optics & Technology National Astronomical Observatories Chinese Academy of Sciences Nanjing 210042 China

Hypervelocity stars are believed to be ejected out from the Galactic center through dynamical interactions between（binary） stars and the central supermassive black hole（s）. In this paper, we report 19 low mass F/G/K type hypervelocity star candidates from over one million stars found in the first data release of the LAMOST regular survey. We determine the unbound probability for each candidate using a MonteCarlo simulation by assuming a non-Gaussian proper-motion error distribution, and Gaussian heliocentric distance and radial velocity error distributions. The simulation results show that all the candidates have unbound possibilities over 50% as expected,and one of them may even exceed escape velocity with over 90% probability. In addition, we compare the metallicities of our candidates with the metallicity distribution functions of the Galactic bulge, disk, halo and globular clusters, and conclude that the Galactic bulge or disk is likely the birth place for our candidates.

关键词： stars： low-mass—stars： kinematics and dynamics—Galaxy： abundances—stars： fundamental parameters—stars： distances

来源：评论

学校读者我要写书评

暂无评论

Grouping-Based Elastic Striping with Hotness Awareness for Improving SSD RAID Performance

Grouping-Based Elastic Striping with Hotness Awareness for I...

引用

International Conference on Dependable Systems and Networks (DSN)

作者： Yubiao Pan Yongkun Li Yinlong Xu Zhipeng Li School of Computer Science & Technology University of Science & Technology of China The State Key Laboratory of High-end Server & Storage Technology

RAID provides a good option to provide device-level fault tolerance. Conventional RAID usually updates parities with read-modify-write or read-reconstruct-write, which may introduce a lot of extra I/Os and thus significantly degrade SSD RAID performance. The recently proposed elastic striping scheme reconstructs new stripes with updated new data chunks without updating old parity chunks. However, it necessitates RAID-level garbage collection which may incur a very high cost. In this paper, we propose a hotness-aware caching scheme to buffer incoming writes and categorize data chunks in buffers into multiple groups according to their hotness values. We then propose a grouping-based elastic striping scheme to separately write data chunks in different groups into SSDs. We deployed the proposed schemes on a RAID-5 array composed of eight commercial SSDs, and experimental results show that compared to elastic striping, our scheme reduces 26% -- 65% of chunk writes to SSDs, and also reduces the average response time by 17.2% -- 63.9%.

关键词： Arrays Time factors Performance evaluation Fault tolerance Fault tolerant systems Delays Prototypes

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

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

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

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

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：