检索结果-内蒙古大学图书馆

2011 International Conference for High Performance computing, Networking, Storage and Analysis, SC11

作者： Solomonik, Edgar Bhatele, Abhinav Demmel, James Dept. of Electric Engineering and Computer Science University of California at Berkeley Berkeley CA 94720 United States Center for Applied Scientific Computing Lawrence Livermore National Laboratory Livermore CA 94551 United States

ISBN: (纸本)9781450307710

Recent results have shown that topology aware mapping reduces network contention in communication-intensive kernels on massively parallel machines. We demonstrate that on mesh interconnects, topology aware mapping also allows for the utilization of highly-efficient topology aware collectives. We map novel 2.5D dense linear algebra algorithms to exploit rectangular collectives on cuboid partitions allocated by a Blue Gene/P supercomputer. Our mappings allow the algorithms to exploit optimized line multicasts and reductions. Commonly used 2D algorithms cannot be mapped in this fashion. On 16,384 nodes (65,536 cores) of Blue Gene/P, 2.5D algorithms that exploit rectangular collectives are sig- nificantly faster than 2D matrix multiplication (MM) and LU factorization, up to 8.7x and 2.1x, respectively. These speed-ups are due to communication reduction (up to 95.6% for 2.5D MM with respect to 2D MM). We also derive LogP- based novel performance models for rectangular broadcasts and reductions. Using those, we model the performance of matrix multiplication and LU factorization on a hypothetical exascale architecture. Copyright 2011 ACM.

关键词： Mapping

来源：评论

学校读者我要写书评

暂无评论

Energy-efficient specialization of functional units in a coarse-grained reconfigurable array 11

Energy-efficient specialization of functional units in a coa...

引用

Proceedings of the 19th ACM/SIGDA international symposium on Field programmable gate arrays

作者： Van Essen, Brian C. Panda, Robin Wood, Aaron Ebeling, Carl Hauck, Scott Dept. of Computer Science and Engineering University of Washington Seattle WA 98195 United States Dept. of Electrical Engineering University of Washington Seattle WA 98195 United States Center for Applied Scientific Computing Lawrence Livermore National Laboratory Livermore CA 94550 United States

ISBN: (纸本)9781450305549

Functional units provide the backbone of any spatial accelerator by providing the computing resources. The desire for having rich and expensive functional units is in tension with producing a regular and energy-efficient computing fabric. This paper explores the design trade-off between complex, universal functional units and simpler, limited functional units. We show that a modest amount of specialization reduces the area-delay-energy product of an optimized architecture to 0.86× a baseline architecture. Furthermore, we provide a design guideline that allows an architect to customize the contents of the computing fabric just by examining the profile of benchmarks within the application domains. Copyright 2011 ACM.

关键词： Energy efficiency

来源：评论

学校读者我要写书评

暂无评论

Max-Min Fairness Linear Transceiver Design for a Multi-User MIMO Interference Channel

Max-Min Fairness Linear Transceiver Design for a Multi-User ...

引用

IEEE International Conference on Communications Workshops

作者： Ya-Feng Liu Yu-Hong Dai Zhi-Quan Luo State Key Lab. of Scientific and Engineering Computing Chinese Academy of Sciences Dept. of Electronic and Computer Engineering University of Minnesota

ISBN: (纸本)9781612842325

Consider the max-min fairness linear transceiver design for a multi-user MIMO interference channel. Assuming perfect channel knowledge, this problem can be formulated as the maximization of minimum SINR utility, subject to individual power constraints at each transmitter. In this paper, it is shown that when the number of antennas at each transmitter (receiver) is at least two and at each receiver (transmitter) is at least three, the problem of checking whether the given target SINR is feasible is strongly NP-hard. A cyclic coordinate ascent algorithm is also proposed for this design problem. Monotonicity and global convergence to KKT solution of the proposed algorithm are proved.

关键词： Interference channels transmitter receivers Transceivers multiple input multiple output systems Individual Power SINR global convergence Transmitters ANTENNA

来源：评论

学校读者我要写书评

暂无评论

Performance test of network simulation based on split-object model

Performance test of network simulation based on split-object...

引用

International Forum on Information Technology and applications

作者： Wu, Ping Wu, Bin Dept. Network Management Center Yunnan Academy of Scientific and Technical Information Kunming 650051 China Computer Technology Application Key Laboratory Kunming University of Science and Technology Kunming 650051 China

ISBN: (纸本)9780769541150

The network simulation (NS) is an important issue in the study of network technology. However, The error between the network performance simulation on split-object model and the true network environment is considerable apparent. In order to obtain the NS result approaching the reality, in this paper, we analyze the structure of split-object mode and find reasons to cause the error. According to the definition of NS performance's index, we remodel the split-object model by choosing the proper function and equation in the control theory and deducing the logical coefficent in them. We further evaluate the efficiency and the availability of the new model under NS-2 platform. The results show that the new mode is more suitable for network simulation with higher accuracy. © 2010 IEEE.

关键词： mathematical analysis network simulation performance test split-object model network analog Performance Tests mathematical analysis Control Theory object information model Networking Techniques

来源：评论

学校读者我要写书评

暂无评论

Detection of salient image points using principal subspace manifold structure

Detection of salient image points using principal subspace m...

引用

International Conference on Pattern Recognition

作者： Paiva, António R. C. Tasdizen, Tolga Scientific Computing and Imaging Institute University of Utah Salt Lake City UT 84112 United States Dept. of Electrical and Computer Eng. University of Utah Salt Lake City UT 84112 United States

ISBN: (纸本)9780769541099

This paper presents a method to find salient image points in images with regular patterns based on deviations from the overall manifold structure. The two main contributions are that: (i) the features to extract salient point are derived directly and in an unsupervised manner from image neighborhoods, and (ii) the manifold structure is utilized, thus avoiding the assumption that data lies in clusters and the need to do density estimation. We illustrate the concept for the detection of fingerprint minutiae, fabric defects, and interesting regions of seismic data. © 2010 IEEE.

关键词： Seismology

来源：评论

学校读者我要写书评

暂无评论

Image parsing with a three-state series neural network classifier

Image parsing with a three-state series neural network class...

引用

2010 20th International Conference on Pattern Recognition, ICPR 2010

作者： Seyedhosseini, Mojtaba Paiva, António R. C. Tasdizen, Tolga Scientific Computing and Imaging Institute University of Utah Salt Lake City UT 84112 United States Dept. of Electrical and Computer Eng. University of Utah Salt Lake City UT 84112 United States

ISBN: (纸本)9780769541099

We propose a three-state series neural network for effective propagation of context and uncertainty information for image parsing. The activation functions used in the proposed model have three states instead of the normal two states. This makes the neural network more flexible than the two-state neural network, and allows for uncertainty to be propagated through the stages. In other words, decisions about difficult pixels can be left for later stages which have access to more contextual information than earlier stages. We applied the proposed method to three different datasets and experimental results demonstrate higher performance of the three-state series neural network. © 2010 IEEE.

关键词： Neural networks

来源：评论

学校读者我要写书评

暂无评论

EMD Strategy for activation detection in functional MRI of the human brain

EMD Strategy for activation detection in functional MRI of t...

引用

作者： Zheng, Tianxiang Yangt, Lihua Jiang, Tianzi Department of Tourism Management Shenzhen Tourism College Jinan University Shenzhen China Dept. of Sci. Computing and Computer Application School of Mathematics and Computational Science Sun Yat-sen University Guangzhou China National Laboratory of Pattern Recognition Institute of Automation Chinese Academy of Sciences Beijing China

ISBN: (纸本)9781424472369

This article investigates how the Empirical Mode Decomposition (EMD) algorithm, a newly-developed and effective technique in the fields of signal processing and time-frequency analysis, is introduced and applied to activation detection in task-related functional Magnetic Resonance Imaging (fMRI) of the human brain. The method is illustrated using fMRI data simulated under one paradigm as well as one real dataset, and its performance is compared to those of a leading model based (General Linear Model - GLM) and a leading data-driven (Region Growing Method - RGM) approach. It is concluded that the proposed method achieves much better performance than the computationally inefficient RGM and is slightly more sensitive than GLM. Experimental results support its efficiency and sensitivity, which indicates that it is to become a viable alternative to the fMRI analysis. © 2010 IEEE.

关键词： Empirical mode decomposition

来源：评论

学校读者我要写书评

暂无评论

The semantic space-time models of the streamonas data stream management system

The semantic space-time models of the streamonas data stream...

引用

2009 WRI World Congress on computer Science and Information Engineering, CSIE 2009

作者： Michael, Panayiotis Adamos Parker Jr., D. Stott Computer Science Dept. - Scientific Computing Lab. UCLA-Anderson School of Management University of California Los Angeles Los Angeles CA United States

ISBN: (纸本)9780769535074

Four novel models define the fundamental temporal semantics upon which the Streamonas DSMS has been architected, as also the fundamental temporal semantics for application development on the system. Extensive experimental results demonstrate the power of the theoretical models as also the stability and scalability of the system when this is tested with a load from 2, 4, 6, 8 and 10 expressways when running the Linear Road Benchmark. The extensive experimental results demonstrate the effectiveness of the theoretical Semantic-Space Time models as the system has reached the maximum level of difficulty of the benchmark (10 expressways) with an average query latency of 0.000026 seconds, 192,307 times faster than the 5 seconds hard real-time constraint the benchmark sets. © 2008 IEEE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Machine learning based online performance prediction for runtime parallelization and task scheduling

Machine learning based online performance prediction for run...

引用

International Symposium on Performance Analysis of Systems and Software, ISPASS 2009

作者： Li, Jiangtian Ma, Xiaosong Singh, Karan Schulz, Martin De Supinski, Bronis McKee, Sally A. Dept. of Computer Science North Carolina State University Raleigh NC 27606 United States Computer Science and Mathematics Division Oak Ridge National Laboratory Oak Ridge TN 37831 United States Computer Systems Laboratory Cornell University Ithaca NY 14850 United States Center for Applied Scientific Computing Lawrence Livermore National Laboratory Livermore CA 94550 United States Department of Computer Science and Engineering Chalmers University of Technology Göteborg Sweden

ISBN: (纸本)9781424441846

With the emerging many-core paradigm, parallel programming must extend beyond its traditional realm of scientific applications. Converting existing sequential applications as well as developing next-generation software requires assistance from hardware, compilers and runtime systems to exploit parallelism transparently within applications. These systems must decompose applications into tasks that can be executed in parallel and then schedule those tasks to minimize load imbalance. However, many systems lack a priori knowledge about the execution time of all tasks to perform effective load balancing with low scheduling overhead. In this paper, we approach this fundamental problem using machine learning techniques first to generate performance models for all tasks and then applying those models to perform automatic performance prediction across program executions. We also extend an existing scheduling algorithm to use generated task cost estimates for online task partitioning and scheduling. We implement the above techniques in the pR framework, which transparently parallelizes scripts in the popular R language, and evaluate their performance and overhead with both a realworld application and a large number of synthetic representative test scripts. Our experimental results show that our proposed approach significantly improves task partitioning and scheduling, with maximum improvements of 21.8%, 40.3% and 22.1% and average improvements of 15.9%, 16.9% and 4.2% for LMM (a real R application) and synthetic test cases with independent and dependent tasks, respectively. © 2009 IEEE.

关键词： Scheduling algorithms

来源：评论

学校读者我要写书评

暂无评论

Haralick's texture features computed by GPUs for biological applications

IAENG International Journal of Computer Science

引用

IAENG International Journal of computer Science 2009年第1期36卷

作者： Gipp, Markus Marcus, Guillermo Harder, Nathalie Suratanee, Apichat Rohr, Karl König, Rainer Männer, Reinhard Institute of Computer Science V Scientific Computing Group University of Heidelberg 68161 Mannheim Germany IPMB BIOQUANT and DKFZ Heidelberg Dept. Bioinformatics and Functional Genomics University of Heidelberg Im Neuenheimer Feld 267 69120 Heidelberg Germany

This paper presents an approach to speedup the computation of co-occurrence matrices and Haralick texture features, as used for analyzing microscopy images of cells, by general-purpose graphics processing units (GPUs). The sequence of computation steps for the features is analyzed based on a graph and an optimized version of the software is derived. Afterwards, a massive parallel software version for GPUs is designed and implemented. On a single node of a cluster, a speedup of a factor of 360 was obtained compared to the original software version, and a speedup of a factor of 32 was achieved compared to the optimized software version.

关键词： Graphics processing unit

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：