检索结果-内蒙古大学图书馆

15th international conference on algorithms and architectures for parallel processing, ICA3PP 2015

作者： Fu, Hao Yu, Ce Sun, Jizhou Wang, Mengmeng Du, Jun School of Computer Science and Technology Tianjin University Tianjin300072 China

ISBN: (纸本)9783319271217

Task scheduling on multiprocessor system is a well-known problem in area of parallel computing. For this problem, many static scheduling algorithms have been reported. But in most static algorithms, only one attribute of tasks is considered when constructing a ready list, which consists of all ready tasks, and there is no evaluation for different task attributes. In this paper, a list scheduling algorithm for DAG-based parallel computing models is proposed. It is mainly designed for reducing the scheduling length of applications with regular DAG models. Eight task attributes in the DAG model are evaluated, and corresponding rules are presented, which will be used in constructing the ready list. And when scheduling tasks, its start time and communication cost on idle processors are taken into consideration. Experimental results show that the proposed algorithm can achieve a significant performance improvement, which is up to 142 %. © Springer international Publishing Switzerland 2015.

关键词： Scheduling algorithms

来源：评论

学校读者我要写书评

暂无评论

Task parallel implementation of matrix multiplication on multi-socket multi-core architectures 15th

Task parallel implementation of matrix multiplication on mul...

引用

15th international conference on algorithms and architectures for parallel processing, ICA3PP 2015

作者： Wang, Yizhuo Ji, Weixing Chen, Xu Hu, Sensen School of Computer Science and Technology Beijing Institute of Technology Beijing100081 China

ISBN: (纸本)9783319271361

Matrix multiplication is a very important computation kernel in many science and engineering applications. this paper presents a parallel implementation framework for dense matrix multiplication on multi-socket multi-core architectures. Our framework first partitions the computation between the multi-core processors. then a hybrid matrix multiplication algorithm is used on each processor, which combines the Winograd algorithm and the classical algorithm. In addition, a hierarchical work-stealing scheme is applied to achieve dynamic load balancing and enforce data locality in our framework. Performance experiments on two platforms show that our implementation gets significant performance gains compared with the state-of-the-art implementations. © Springer international Publishing Switzerland 2015.

关键词： Matrix algebra

来源：评论

学校读者我要写书评

暂无评论

Accelerated steiner tree problem solving on GPU with CUDA 15th

Accelerated steiner tree problem solving on GPU with CUDA

引用

15th international conference on algorithms and architectures for parallel processing, ICA3PP 2015

作者： Mathieu, Christian Klusch, Matthias Computer Science Department Saarland University Saarbruecken66123 Germany German Research Center for Artificial Intelligence Saarbruecken66123 Germany

ISBN: (纸本)9783319271217

the Steiner Tree Problem in Graphs (STPG) is an important NP-hard combinatorial optimization problem arising naturally in many applications including network routing, social media analysis, power grid routing and in the context of the semantic web. We present the first parallel heuristics for the solution of the STPG for GPU with CUDA, and show that the achieved speedups for different kinds of graphs are significant compared to one of the fastest serial heuristics STAR. © Springer international Publishing Switzerland 2015.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

Performance characterization and optimization for intel xeon phi coprocessor 1

引用

15th international conference on algorithms and architectures for parallel processing, ICA3PP 2015

作者： Zhang, Cheng Liu, Li Li, Ruizhe Yang, Guangwen Department of Computer Science and Technology Tsinghua University Beijing100084 China Center for Earth System Science Tsinghua University Beijing100084 China

ISBN: (数字)9783319271194

ISBN: (纸本)9783319271187

the Intel Xeon Phi is a many-core accelerator which focuses on the high performance applications. To characterize the performance of the Intel Xeon Phi, a system of dual 8-core Intel Xeon E5-2670 processors is employed as a control platform, and a subset of the PARSEC benchmark suite is selected as the benchmark applications. the first evaluation in this paper shows that the applications on the Intel Xeon Phi is averagely 2. 06x slower than on the dual Intel Xeon E5-2670. the further detailed performance characterization quantifies the performance impact of various architecture parameters on the Intel Xeon Phi. To set an example for how to improve the architecture of the Intel Xeon Phi for better performance, the hardware optimization with an additional set of vector processing units is discussed and a simple emulator is developed accordingly. the evaluation results show that this optimization can provide an average speedup of 1. 10. © Springer international Publishing Switzerland 2015.

关键词： parallel architectures

来源：评论

学校读者我要写书评

暂无评论

Can portability improve performance? An empirical study of parallel graph analytics 15

Can portability improve performance? An empirical study of p...

引用

6th ACM/SPEC international conference on Performance Engineering, ICPE 2015

作者： Varbanescu, Ana Lucia Verstraaten, Merijn De Laat, Cees Penders, Ate Iosup, Alexandru Sips, Henk University of Amsterdam Netherlands Delft University of Technology Netherlands

ISBN: (纸本)9781450332484

Due to increasingly large datasets, graph analytics-traversals, allpairs shortest path computations, centrality measures, etc.-are becoming the focus of high-performance computing (HPC). Because HPC is currently dominated by many-core architectures (both CPUs and GPUs), new graph processing solutions have to be defined to efficiently use such computing resources. Prior work focuses on platform-specific performance studies and on platform-specific algorithm development, successfully proving that algorithms highly tuned to GPUs or multi-core CPUs can provide high performance graph analytics. However, the portability of such algorithms remains an important concern for many users, especially the many companies without the resources to invest in HPC or concerned about lock-in in single-use parallel techniques. In this work, we investigate the functional portability and performance of graph analytics algorithms. We conduct an empirical study measuring the performance of 3 graph analytics algorithms (a single code implemented in OpenCL and targeted at many-core CPUs and GPUs), on 3 different platforms, using 11 real-world and synthetic datasets. Our results show that the code is functionally portable, that is, the applications can run unchanged on both CPUs and GPUs. the large variation in their observed performance indicates that portability is necessary not only for productivity, but, surprisingly, also for performance. We conjecture that the impact of datasets on performance is too high to allow platform-specific algorithms to outperform the portable algorithms by large margins, in all cases. Our conclusion is that portable parallel graph analytics is feasible without significant performance loss, and provides a productive alternative to the expensive trial-and-error selection of one algorithm for each (graph,platform) pair. Copyright © 2015 ACM.

关键词： Program processors

来源：评论

学校读者我要写书评

暂无评论

parallel pipeline implementation for moving objects detection in traffic video analysis on a heterogeneous platform 4th

Parallel pipeline implementation for moving objects detectio...

引用

4th international conference on Computer Engineering and Networks, CENet2014

作者： Li, Teng Dou, Yong Jiang, Jingfei Qiao, Peng Department of Computer Science National Laboratory for Parallel and Distributed Processing National University of Defense Technology Changsha410073 China

ISBN: (纸本)9783319111032

Moving objects detection is important in traffic video analysis, and many algorithms are being increasingly applied to moving objects detection. Most of these algorithms are time-consuming and cannot satisfy real-time demand in traffic video analysis by using a conventional central processing unit (CPU) sequential method. the emergence of the graphics processing unit (GPU) and multi-core CPU provides a method to accelerate the aforementioned algorithms to meet demand. In this study, we provide a GPU-accelerated implementation of the background subtraction algorithm and the morphological operation algorithm. then, the connected component labeling algorithm is parallelized on a multi-core CPU with Open Multi-processing (OpenMP). Furthermore, parallel pipeline implementation on a heterogeneous platform is proposed by integrating the aforementioned algorithms. Experimental results show that the proposed implementation achieves a significant speedup of up to 5×, compared with sequential implementation on a CPU. © Springer international Publishing Switzerland 2015.

关键词： Program processors

来源：评论

学校读者我要写书评

暂无评论

From Sequential Algorithm Selection to parallel Portfolio Selection 1

引用

9th international conference on Learning and Intelligent OptimizatioN (LION)

作者： Lindauer, M. Hoos, Holger H. Hutter, F. Univ Freiburg D-79106 Freiburg Germany Univ British Columbia Vancouver BC V5Z 1M9 Canada

ISBN: (数字)9783319190846

ISBN: (纸本)9783319190846;9783319190839

In view of the increasing importance of hardware parallelism, a natural extension of per-instance algorithm selection is to select a set of algorithms to be run in parallel on a given problem instance, based on features of that instance. Here, we explore how existing algorithm selection techniques can be effectively parallelized. To this end, we leverage the machine learning models used by existing sequential algorithm selectors, such as 3S, ISAC, SATzilla and ME-ASP, and modify their selection procedures to produce a ranking of the given candidate algorithms;we then select the top n algorithms under this ranking to be run in parallel on n processing units. Furthermore, we adapt the pre-solving schedules obtained by aspeed to be effective in a parallel setting with different time budgets for each processing unit. Our empirical results demonstrate that, using 4 processing units, the best of our methods achieves a 12-fold average speedup over the best single solver on a broad set of challenging scenarios from the algorithm selection library.

关键词： Algorithm selection parallel portfolios Constraint solving Answer Set Programming

来源：评论

学校读者我要写书评

暂无评论

Efficient parallel Recursive Gaussian SIFT Algorithm Based on Multi-core DSP

Efficient Parallel Recursive Gaussian SIFT Algorithm Based o...

引用

2015 5th international conference on Electronics Information and Emergency Communication (ICEIEC)

作者： Luo Yong Ye Zhengyuan Chen Yuanzhi Commun Univ China Informat Engn Sch Beijing Peoples R China Commun Univ China Engn Ctr Digital Audio & Video Beijing Peoples R China

ISBN: (纸本)9781479972845

To solve the problem of high computational complexity and real-time poor of the SIFT(Scale Invariant Feature Transform) algorithm, a parallel data streams RGF-SIFT(Recursive Gaussian Filter -SIFT) algorithm based on DSP multi-core processor is proposed. the proposed algorithm uses the forth-order recursive Gaussian filter to replace the linear Gaussian filtering of the SIFT algorithm. then the four modules of RGF-SIFT computing tasks are assigned to multiple DSP core for parallel processing, and implemented synchronization for multicore processor through inter-processor communication (IPC) and other technologies. Experimental results show that the parallel RGF-SIFT algorithm detects feature point more than the algorithm of SIFT, and the repetition rate of the correct feature point is very high. At execution time, the parallel RGF-SIFT algorithm has higher speedup ratio.

关键词： recursive Gaussian filter SIFT multi-core processor parallel

来源：评论

学校读者我要写书评

暂无评论

Scalable multi-dimensional RNN query processing

引用

CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE 2015年第16期27卷 4156-4171页

作者： Ji, Changqing Qu, Wenyu Li, Zhiyang Xu, Yujie Li, Yuanyuan Wu, Junfeng Dalian Maritime Univ Sch Informat Sci & Technol Dalian Peoples R China Dalian Univ Sch Phys Sci & Technol Dalian 116012 Peoples R China Dalian Jiaotong Univ Sch Software Dalian Peoples R China Dalian Ocean Univ Sch Educ Technol Dalian Peoples R China Dalian Ocean Univ Ctr Comp Dalian Peoples R China

Reverse nearest neighbor (RNN) queries are the complimentary problem and particular interest in the past few years, such as location-based services, profile-based marketing, resource allocation, and traffic monitoring system. the one major drawback for the existing RNN is that it has inherent sequential nature and uses in-memory algorithm, which limits its applicability to large-scale spatial data queries. this paper proposes scalable algorithms for RNN queries in a distributed environment. Firstly, we investigate the Basic-scalable reverse nearest neighbor (SRNN) initialization query method based on the inverted grid index. Secondly, two optimization methods Lazy-SRNN and Eager-SRNN are proposed to effectively process scalable multi-dimensional RNN queries. Among them, Lazy-SRNN prunes the search space when all RNN objects are discovered in one pass;Eager-SRNN attempts to prune spatial objects incrementally as soon as they are visited. In addition, the SRNN algorithm is proved to be the first attempt for the exact scalable RNN algorithms in a distributed environment on multi-dimensional data sets. We show in an extensive experimental evaluation on real-world and synthetic data the scalability and the performance of our novel approach. Copyright (c) 2015 John Wiley & Sons, Ltd.

关键词： reverse nearest neighbor big data MapReduce cloud computing

来源：评论

学校读者我要写书评

暂无评论

international Workshops and Symposiums on algorithms and architectures for parallel processing, ICA3PP 2015

International Workshops and Symposiums on Algorithms and Arc...

引用

15th international conference on algorithms and architectures for parallel processing, ICA3PP 2015

ISBN: (纸本)9783319271606

the proceedings contain 83 papers. the special focus in this conference is on Trust, Security and Privacy for Big Data. the topics include: Nth-Order Multifunction Filter Employing Current Differencing Transconductance Amplifiers;An Efficient Spatial Query processing Algorithm in Multi-sink Directional Sensor Network;An Improved Method for Reversible Data Hiding in Encrypted Image;Study on Personalized Location Privacy Preservation algorithms Based on Road Networks;A Hierarchical Identity-Based Signature from Composite Order Bilinear Groups;Towards Returning Data Control to Cloud Users;A Distributed Algorithm for Graph Edge Partitioning;Scheduling Stochastic Tasks with Precedence Constrain on Cluster Systems with Heterogenous Communication Architecture;An Output-Oriented Approach of Test Data Generation Based on Genetic Algorithm;An Efficient Pre-filter to Accelerate Regular Expression Matching;A Hybrid Optimization Approach for Anonymizing Transactional Data;Program Obfuscator for Privacy-Carrying Unidirectional One-hop Re-encryption;Predicting Severity of Software Vulnerability Based on Grey System theory;Characterization of Android Applications with Root Exploit by Using Static Feature Analysis;A Lightweighted Fine-Grained Privacy-Preserving Profile Matching Mechanism for Mobile Social Networks in Proximity;Context-Aware QoS Assurance for Smart Grid Big Data processing with Elastic Cloud Resource Reconfiguration;Continuous User Identity Verification for Trusted Operators in Control Rooms;Leakage-Resilient Anonymous Identity-Based Broadcast Encryption in the Standard Model;Scheduling Resource of IaaS Clouds for Energy Saving Based on Predicting the Overloading Status of Physical Machines and Towards Mechanised Semantics of HPC.

关键词：

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：