检索结果-内蒙古大学图书馆

12th GAMM-IMACS International Symposium on scientific computing, computer Arithmetic and Validated Numerics, SCAN 2006

作者： Frommer, Andreas Hüsken, Matthias Bergische Universität Wuppertal Applied Computer Science and Scientific Computing D-42097 Wuppertal Germany

ISBN: (纸本)076952821X

Certain numerically intensive applications executed within a grid computing environment crucially depend on the properties of floating-point arithmetic implemented on the respective platform. Differences in these properties may have drastic effects. This paper identifies the central problems related to this situation. We propose an approach which gives the user valuable information on the various platforms available in a grid computing environment in order to assess the numerical quality of an algorithm run on each of these platforms. In this manner, the user will at least have very strong hints whether a program will perform reliably in a grid before actually executing it. Our approach extends the existing IeeeCC754 test suite by two "grid-enabled" modes: The first mode calculates a "numerical checksum" on a specific grid host and executes the job only if the checksum is identical to a locally generated one. The second mode provides the user with information on the reliability and IEEE 754-conformity of the underlying floating-point implementation of various platforms. Furthermore, it can help to find a set of compiler options to optimize the application's performance while retaining numerical stability.

关键词： Digital arithmetic

来源：评论

学校读者我要写书评

暂无评论

Analysis of Cache-Coherence Bottlenecks with Hybrid Hardware/Software Techniques

引用

ACM Transactions on Architecture and Code Optimization 2006年第4期3卷 390-423页

作者： Marathe, Jaydeep Mueller, Frank De Supinski, Bronis R. North Carolina State University Department of Computer Science Raleigh NC 27695-7534 United States Lawrence Livermore National Laboratory Center for Applied Scientific Computing Livermore L-557 CA 94551 United States

Application performance on high-performance shared-memory systems is often limited by sharing patterns resulting in cache-coherence bottlenecks. Current approaches to identify coherence bottlenecks incur considerable run-time overhead and do not scale. We present two novel hardwareassisted coherence-analysis techniques that reduce trace sizes by two orders of magnitude over full traces. First, hardware performance monitoring is combined with capturing stores in software to provide a lossy-trace mechanism, which is an order of magnitude faster than softwareinstrumentation- based full-tracing and retains accuracy. Second, selected long-latency loads are instrumented via binary rewriting, which provides even higher accuracy and control over tracing, but requires additional overhead. © 2006, ACM. All rights reserved.

关键词： Cache analysis Coherence protocols Dynamic binary rewriting Hardware performance monitoring Measurement Performance Program instrumentation Smps

来源：评论

学校读者我要写书评

暂无评论

Ensuring Numerical Quality in Grid computing

Ensuring Numerical Quality in Grid Computing

引用

International Symposium on scientific computing, computer Arithmetic and Validated Numerics (SCAN)

作者： Andreas Frommer Matthias Husken Bergische Universität Wuppertal Applied Computer Science and Scientific Computing Wuppertal Germany

ISBN: (纸本)9781424431410;9780769528212;076952821X

关键词： Grid computing Floating-point arithmetic Application software Mesh generation Software libraries Testing Optimizing compilers Hardware Standards development computer science

来源：评论

学校读者我要写书评

暂无评论

Ensuring numerical quality in grid computing

Ensuring numerical quality in grid computing

引用

International Symposium on High Performance Distributed computing

作者： A. Frommer M. Husken Applied Computer Science and Scientific Computing Bergische Universität Gesamthochschule Wuppertal Wuppertal Germany

We propose an approach which gives the user valuable information on the various platforms avail able in a grid in order to assess the numerical quality of an algorithm run on each of these platforms. In this manner, the user is provided with at least very strong hints whether a program performs reliably in a grid before actually executing it. Our approach extends IeeeCC754 by two "grid-enabled" modes: The first mode calculates a "numerical checksum" on a specific grid host and executes the job only if the check sum is identical to a locally generated one. The second mode provides the user with information on the reliability and IEEE 754-conformity of the underlying floating-point implementation of various platforms. In addition, it can help to find a set of compiler options to optimize the application's performance while retaining numerical stability

关键词： Grid computing Testing Application software Floating-point arithmetic Hardware Software libraries computer science scientific computing Programming Software standards

来源：评论

学校读者我要写书评

暂无评论

On the optimal working set size in serial and parallel support vector machine learning with the decomposition algorithm

Conferences in Research and Practice in Information Technolo...

引用

Conferences in Research and Practice in Information Technology Series 2006年 61卷 121-128页

作者： Eitrich, Tatjana Lang, Bruno Central Institute for Applied Mathematics Research Centre Juelich Germany Applied Computer Science and Scientific Computing Group Department of Mathematics University of Wuppertal Germany

The support vector machine (SVM) is a wellestablished and accurate supervised learning method for the classification of data in various application fields. The statistical learning task - the so-called training - can be formulated as a quadratic optimization problem. During the last years the decomposition algorithm for solving this optimization problem became the most frequently used method for support vector machine learning and is the basis of many SVM implementations today. It is characterized by an internal parameter called working set size. Usually small working sets have been assigned. The increasing amount of data used for classification led to new parallel implementations of the decomposition method with efficient inner solvers. With these solvers larger working sets can be assigned. It was shown, that for parallel training with the decomposition algorithm large working sets achieve good speedup values. However, the choice of the optimal working set size for parallel training is not clear. In this paper, we show how the working set size influences the number of decomposition steps, the number of kernel function evaluations and the overall training time in serial and parallel computation. © 2006, Australian computer Society, Inc.

关键词： Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Large-scale electronic structure calculations of high-Z metals on the BlueGene/L platform

Large-scale electronic structure calculations of high-Z meta...

引用

2006 ACM/IEEE Conference on Supercomputing, SC'06

作者： Gygi, Francois Draeger, Erik W. Schulz, Martin De Supinski, Bronis R. Gunnels, John A. Austel, Vernon Sexton, James C. Franchetti, Franz Kral, Stefan Ueberhuber, Christoph W. Lorenz, Juergen Department of Applied Science University of California Davis Davis CA 95616 Center for Applied Scientific Computing Lawrence Livermore National Laboratory Livermore CA 94551 IBM Thomas J. Watson Research Center Yorktown Heights NY 10598 Department of Electrical and Computer Engineering Carnegie Mellon University Pittsburgh PA 15213 Institute of Analysis and Scientific Computing Vienna University of Technology Vienna Austria

ISBN: (纸本)0769527000

First-principles simulations of high-Z metallic systems using the Qbox code on the BlueGene/L supercomputer demonstrate unprecedented performance and scaling for a quantum simulation code. Specifically designed to take advantage of massively-parallel systems like BlueGene/L, Qbox demonstrates excellent parallel efficiency and peak performance. A sustained peak performance of 207.3 TFlop/s was measured on 65,536 nodes, corresponding to 56.5% of the theoretical full machine peak using all 128k CPUs. © 2006 IEEE.

关键词： Electronic structure

来源：评论

学校读者我要写书评

暂无评论

Annotating user-defined abstractions for optimization 20

Annotating user-defined abstractions for optimization

引用

20th IEEE International Parallel and Distributed Processing Symposium, IPDPS 2006

作者： Quinlan, Dan Schordan, Markus Vuduc, Richard Yi, Qing Center for Applied Scientific Computing Lawrence Livermore National Laboratory Livermore CA 94550 United States Institute of Computer Languages Vienna University of Technology Vienna Austria Dept. of Computer Science University of Texas at San Antonio San Antonio TX 78249 United States

ISBN: (纸本)1424400546

Although conventional compilers implement a wide range of optimization techniques, they frequently miss opportunities to optimize the use of abstractions, largely because they are not designed to recognize and use the relevant semantic information about such abstractions. In this position paper, we propose a set of annotations to help communicate high-level semantic information about abstractions to the compiler, thereby enabling the large body of traditional compiler optimizations to be applied to the use of those abstractions. Our annotations explicitly describe properties of abstractions that are needed to guarantee the applicability and profitability of a broad variety of such optimizations, including memoization, reordering, data layout transformations, and Mining and specialization. © 2006 IEEE.

关键词： Program compilers

来源：评论

学校读者我要写书评

暂无评论

Enabling scientific Workflow Reuse through Structured Composition of Dataflow and Control-Flow 22

Enabling Scientific Workflow Reuse through Structured Compos...

引用

22nd International Conference on Data Engineering Workshops, ICDEW 2006

作者： Bowers, Shawn Ludäscher, Bertram Ngu, Anne H.H. Critchlow, Terence UC Davis Genome Center University of California Davis United States Deptartment of Computer Science UC Davis Genome Center University of California Davis United States Department of Computer Science Texas State University San Marcos United States Center for Applied Scientific Computing Lawrence Livermore National Laboratory United States

ISBN: (纸本)0769525717

Data-centric scientific workflows are often modeled as dataflow process networks. The simplicity of the dataflow framework facilitates workflow design, analysis, and optimization. However, modeling "control-flow intensive" tasks using dataflow constructs often leads to overly complicated workflows that are hard to comprehend, reuse, and maintain. We describe a generic framework, based on scientific workflow templates and frames, for embedding control-flow intensive subtasks within dataflow process networks. This approach can seamlessly handle complex control-flow without sacrificing the benefits of dataflow. We illustrate our approach with a real-world scientific workflow from the astrophysics domain, requiring remote execution and file transfer in a semi-reliable environment. For such workflows, we also describe a 3-layered architecture based on frames and templates where the top-layer consists of an overall dataflow process network, the second layer consists of a tranducer template for modeling the desired control-flow behavior, and the bottom layer consists of frames inside the template that are specialized by embedding the desired component implementation. Our approach can enable scientific workflows that are more robust (faulttolerance strategies can be defined by control-flow driven transducer templates) and at the same time more reusable, since the embedding of frames and templates yields more structured and modular workflow designs. © 2006 IEEE.

关键词： Astrophysics

来源：评论

学校读者我要写书评

暂无评论

On the Advantages of Weighted L1-Norm Support Vector Learning for Unbalanced Binary Classification Problems

On the Advantages of Weighted L1-Norm Support Vector Learnin...

引用

International IEEE Conference on Intelligent Systems, IS

作者： Tatjana Eitrich Bruno Lang Central Institute for Applied Mathematics Research Center Juelich GmbH Germany Applied Computer Science and Scientific Computing Group Department of Mathematics University of Wuppertal Germany

In this paper we analyze support vector machine classification using the soft margin approach that allows for errors and margin violations during the training stage. Two models for learning the separating hyperplane do exist. We study the behavior of the optimization algorithms in terms of training characteristics and test accuracy for unbalanced data sets. The main goal of our work is to compare the features of the resulting classification functions, which are mainly defined by the support vectors arising during the support vector machine training

关键词： Support vector machine classification Support vector machines Learning systems Kernel Testing Supervised learning Mathematics Intelligent systems Machine learning Machine learning algorithms

来源：评论

学校读者我要写书评

暂无评论

Annotating user-defined abstractions for optimization 06

Annotating user-defined abstractions for optimization

引用

International Symposium on Parallel and Distributed Processing (IPDPS)

作者： D. Quinlan M. Schordan R. Vuduc Qing Yi Lawrence Livermore National Laboratory Center for Applied Scientific Computing Livermore CA USA Institute of Computer Languages University of Technology Vienna Vienna Austria Department of Computer Science University of Texas San Antonio San Antonio TX USA

ISBN: (纸本)9781424400546

关键词： Optimizing compilers Design optimization scientific computing Laboratories Profitability computer languages computer science Costs US Department of Energy Containers

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：