检索结果-内蒙古大学图书馆

Interference Management for distributed parallel Applications in Consolidated Clusters

ACM SIGPLAN NOTICES 2016年第4期51卷 443-456页

作者： Han, Jaeung Jeon, Seungheun Choi, Young-ri Huh, Jaehyuk Korea Adv Inst Sci & Technol Sch Comp Daejeon South Korea UNIST Sch Elect & Comp Engn Ulsan South Korea

Consolidating multiple applications on a system can improve the overall resource utilization of data center systems. However, such consolidation can adversely affect the performance of some applications due to interference caused by resource contention. Despite many prior studies on the interference effects in single-node systems, the interference behaviors of distributed parallel applications have not been investigated thoroughly. With distributed applications, a local interference in a node can affect the whole execution of an application spanning many nodes. This paper studies an interference modeling methodology for distributed applications to predict their performance under interference effects in consolidated clusters. This study first characterizes the effects of interference for various distributed applications over different interference settings, and analyzes how diverse interference intensities on multiple nodes affect the overall performance. Based on the characterization, this study proposes a static profiling-based model for interference propagation and heterogeneity behaviors. In addition, this paper presents use case studies of the modeling method, two interference-aware placement techniques for consolidated virtual clusters, which attempt to maximize the overall throughput or to guarantee the quality-of-service.

关键词： Resource contention consolidated system cloud computing distributed parallel application interference model placement algorithm

来源：评论

学校读者我要写书评

暂无评论

Automatic Refactoring of Legacy Fortran Code to the Array Slicing Notation

Automatic Refactoring of Legacy Fortran Code to the Array Sl...

引用

international conference on parallel and distributed systems (ICPADS)

作者： Chandrasehar Rajaseharan Wen Jun Tan Wai Teng Tang Stephen John Turner Shyh-Hao Kuo Rick Siow Mong Goh Weng-Fai Wong School of Computer Engineering Nanyang Technological University Singapore Agency for Science Technology and Research Institute of High Performance Computing Singapore Agency for Science Technology and Research Singapore SG Department of Computer Science National University of Singapore Singapore

There are many legacy Fortran programs still in use today, especially scientific codes which were written decades ago. Many of these codes use explicit DO-loops in programs that tend to clutter the code and make it harder to understand and maintain. Modern features of the Fortran language, such as the array slicing notation and introduction of commonly used intrinsic functions, go a long way in helping programmers write code that is easier to read and maintain. We introduce a refactoring tool that can help to transform code to make use of the array slicing notation and related intrinsic functions.

关键词： Arrays Educational institutions Jacobian matrices Optimization Java Clutter Transforms

来源：评论

学校读者我要写书评

暂无评论

Interactive Sound Generation System Based on Instruction Recognition of Human Body Motion

Interactive Sound Generation System Based on Instruction Rec...

引用

ACIS international conference on Software Engineering, Artificial Intelligence, Networking, and parallel/distributed computing (SNPD)

作者： Hotaka Kitabora Yoichiro Maeda Yasutake Takahashi Dept. of Human and Artificial Intelligent Systems Graduate School of Engineering University of Fukui Fukui Japan

Recently, the interactive sound is researched in the field of an artificial life or Kansei engineering as one of interactive arts. An interactive sound changes the aspect variously by the interaction of human and a system, and it aims at generating the sound which was full of the complexity and diversity beyond expectations of human. In our laboratory, the chaotic sound generation system which shows various aspects has been developed by using the Globally Coupled Map (GCM) which puts plural chaos elements in order and operates each chaos and the whole synchronicity. However, in the present system, what we can operate is only a chaos parameter. Therefore this is the system which it is hard to reflect the human's intention in the generated sound. So, in this research, the system which analyzes human motions and instructions with an external camera and defines the directivity of the output sound is proposed. Motion capture system detects coordinates of human's joint. We designed the sound generation system to which the element of sound pitch, length, volume, and tonality are changed by the detected motion. The intention of desired sound by human is able to be reflected to the system, and we can expect to realize the large interaction effect. Moreover, it is also reported by Kansei evaluation to confirm the efficiency of the designed system.

关键词： Software engineering Artificial intelligence distributed computing

来源：评论

学校读者我要写书评

暂无评论

Using Supercomputer to Speed up Neural Network Training

Using Supercomputer to Speed up Neural Network Training

引用

international conference on parallel and distributed systems (ICPADS)

作者： Yue Yu Jinrong Jiang Xuebin Chi Department of High Performance Computing Technology and Application Development University of Chinese Academy of Sciences Beijing China Department of High Performance Computing Technology and Application Development Center of Scientific Computing Applications & Research Beijing China Department of High Performance Computing Technology and Application Development Chinese Academy of Sciences Beijing China

ISBN: (纸本)9781509053827

Recent works in deep learning have shown that large models can dramatically improve performance. In this paper, we accelerated the deep network training using many GPUs. We have developed a framework based on Caffe called Caffe-HPC that can utilize computing clusters with multiple GPUs to train large models. Caffe[6] provides multimedia scientists and practitioners with a clean and modifiable framework for state-of-the-art deep learning algorithms and a collection of reference models. And Caffe-HPC retains all the features of the original Caffe, the model trained on original Caffe can be continue to trained on Caffe-HPC. It provides a convenient solution for people who are using Caffe and want to speed up the training. Using an Asynchronous Stochastic Gradient Descent optimizer, We made a good acceleration on training a CNN model on ILSVRC[5] 2012 dataset. And we have compared the convergence of different SGD algorithms. We believe our work will makes it possible to train larger networks on larger training sets in a reasonable amount of time.

关键词： Servers Training Graphics processing units Neural networks Machine learning Computational modeling Acceleration

来源：评论

学校读者我要写书评

暂无评论

Scalability issues in cloud computing

Scalability issues in cloud computing

引用

international conference on Advanced computing, ICAC

作者： Thamarai Selvi Somasundaram V. Prabha Mahesh Arumugam MIT Campus CDAC Alpha Cloud Labs LLC & Pvt Ltd

The following topics are dealt with: ad hoc & sensor networks; agent based systems; autonomous computing; body area network; cloud computing; cluster computing; compiler optimizations; data mining; digital image computing; distributed databases; e-learning; e-commerce; e-governance; green IT; grid computing; image processing; intelligent agents; knowledge management; theoretical computer science; virtualization; machine learning; mobile computing; multicore architecture; multimedia streaming; natural language processing; network security; network management; neural networks; parallel computing; pattern recognition; pervasive computing; power aware computing; semantic Web; service oriented architecture; soft computing; software architecture; social computing; storage management; ubiquitous computing; and Web intelligence.

关键词： Cloud computing Scalability Educational institutions Abstracts Databases Virtualization Organizations

来源：评论

学校读者我要写书评

暂无评论

Toward comprehensive and accurate simulation performance prediction of parallel file systems

Toward comprehensive and accurate simulation performance pre...

引用

international conference on Dependable systems and Networks (DSN)

作者： Miguel A. Erazo Ting Li Jason Liu Stephan Eidenbenz School of Computer and Information Sciences Florida International University USA Information Sciences (CCS-3) Los Alamos National Laboratory USA

We present the design and implementation of FileSim, a simulation framework with detailed models of parallel file systems, capable of reproducing the complex I/O behavior at scale. FileSim aims to support comprehensive and accurate end-to-end I/O performance prediction and evaluation of exascale high-end computing systems. To this end, FileSim provides several key features, including detailed, pluggable models of contemporary parallel file systems, the support of trace-driven simulation, and the capability of running large-scale I/O systems using parallel and distributed *** conducted extensive validation and performance studies, through which we show that the simulator is capable of reproducing important I/O system behaviors comparable to those measured from the real systems. We demonstrate the capabilities of FileSim as a tool for exploring the parameter space and design alternatives of large-scale parallel file systems.

关键词： Computational modeling Servers Redundancy Computer architecture Predictive models Performance evaluation Linux

来源：评论

学校读者我要写书评

暂无评论

Weight adjustment of the particle filter on distributed computing systems

Weight adjustment of the particle filter on distributed comp...

引用

international conference on Information Fusion

作者： Shin'ya Nakano Tomoyuki Higuchi The Institute of Statistical Mathematics Tachikawa Tokyo Japan.

ISBN: (纸本)9781467304177

The particle filter (PF) is a state estimation algorithm that is inherently suitable for parallel computing. When the PF is implemented on a parallel computer, it is crucial to reduce the number of data transfers in the resampling procedure. One effective way to do this is to divide the particles into multiple groups. If the resampling is then performed only within each group, data transfers are reduced effectively. However, when the resampling is limited to within a small group, the imbalance of the weights of the particles cannot be resolved sufficiently, and this can depress the estimation accuracy. To evaluate this imbalance, we introduce a metric based on the entropy and observe that the accuracy actually does become worse as the imbalance of weights becomes more evident. We then propose a recipe in which the imbalance of weights is resolved when the metric of the imbalance is less than a predetermined threshold value. Finally, we demonstrate that this recipe notably improves the estimation accuracy without requiring substantial additional computational cost.

关键词： Particle filter filtering parallel computing

来源：评论

学校读者我要写书评

暂无评论

An Execution Environment for Robust parallel computing on Volunteer PC Grids

An Execution Environment for Robust Parallel Computing on Vo...

引用

international conference on parallel Processing (ICPP)

作者： Hien Nguyen Eshwar Pedamallu Jaspal Subhlok Edgar Gabriel Qian Wang Margaret S. Cheung David Anderson Department of Computer Science University of Houston USA Department of Physics University of Houston USA Space Sciences Laboratory University of California Berkeley USA

A pool of distributed volunteer PCs presents an extremely hostile environment for execution of communicating parallel codes due to system and network heterogeneity, varying availability, and frequent failures. Well known methods for fault tolerance, specifically replication and check pointing, are challenging to deploy and not sufficient individually to provide continuous forward application progress. As the failure of a single logical process leads to application failure, the degree of redundancy needed for long running applications is too large to be practical. Check pointing and rollback does not provide protection against slow and variable speed nodes and is impractical when system wide MTBF is in minutes or less, common for a moderate size volunteer computing pool. The approach taken in this research is to exploit both, but that presents formidable challenges, efficient check pointing of distributed replicated processes, dynamic management of redundancy, quick restart in a distributed environment, and others. Proposed solution also leverages node selection based on availability prediction. The integrated runtime system is shown to effectively execute moderate size, coarse grain, communicating codes on a worldwide distributed volunteer environment, a new milestone in volunteer computing. The results provide new insight into how multiple techniques interact and contribute to robustness. The programming model is based on one-sided Put/Get calls to an abstract global shared space that works seamlessly with replicated processes. A Replica Exchange Molecular Dynamics code is employed to drive evaluation. The execution environment includes hosts on a University campus as well as hosts distributed around the world.

关键词： Checkpointing Servers Availability Redundancy Programming Fault tolerant systems

来源：评论

学校读者我要写书评

暂无评论

Managing Data-Movement for Effective Shared-Memory parallelization of Out-of-Core Sparse Solvers 12

Managing Data-Movement for Effective Shared-Memory Paralleli...

引用

ACM/IEEE international conference for High Performance computing, Networking, Storage, and Analysis

作者： Haim Avron Anshul Gupta IBM T.J. Watson Research Center

ISBN: (纸本)9781467308052

Direct methods for solving sparse linear systems are robust and typically exhibit good performance, but often require large amounts of memory due to fill-in. Many industrial applications use out-of-core techniques to mitigate this problem. However, parallelizing sparse out-of-core solvers poses some unique challenges because accessing secondary storage introduces serialization and I/O overhead. We analyze the data-movement costs and memory versus parallelism trade-offs in a shared-memory parallel out-of-core linear solver for sparse symmetric systems. We propose an algorithm that uses a novel memory management scheme and adaptive task parallelism to reduce the data-movement costs. We present experiments to show that our solver is faster than existing out-of-core sparse solvers on a single core, and is more scalable than the only other known shared-memory parallel out-of-core solver. This work is also directly applicable at the node level in a distributed-memory parallel scenario.

关键词： node level distributed-memory parallel out-of-core Solvers distributed memory parallel Lines AUXILIARY STORAGE serialize Memory Industrial applications

来源：评论

学校读者我要写书评

暂无评论

Self-Stabilizing Micro Controller for Large-Scale Sensor Networks in Spite of Program Counter Corruptions Due to Soft Errors

Self-Stabilizing Micro Controller for Large-Scale Sensor Net...

引用

international conference on parallel and distributed systems (ICPADS)

作者： Christopher Boelmann Torben Weis Michael Engel Arno Wacker Distributed Systems Group University of Duisburg-Essen Duisburg Germany Design Automation for Embedded Systems Group TU Dortmund University Dortmund Germany Software-engineering for Ubiquitous-Computing Applications University of Kassel Kassel Germany

For large installations of networked embedded systems it is important that each entity is self-stabilizing, because usually there is nobody to restart nodes that have hung up. Self-stabilization means to recover from temporary failures (soft errors) and adapt to a change of network topology caused by permanent failures. On the software side self-stabilizing algorithms must assume that the hardware is executing the software correctly. In this paper we discuss cases in which soft errors invalidate this assumption, especially in cases where CPU registers or the watchdog timer are affected by the fault. Based on the observation that a guaranteed self-stabilization is only possible as long as the watchdog-timer is working properly after temporary failures, we propose and compare three different approaches that meet the requirements of sensor networks, to solve this problem with a combination of hardware- and software-modifications: 1) A run-time verification of every watchdog access 2) A completely hardware-based approach, without any software modifications 3) A 2 X byte code alignment, to realign a corrupted program counter Furthermore we determine the average code-size increase and evaluate necessary hardware-modifications that come along with each approach.

关键词： Registers Hardware Random access memory Read only memory Microprocessors Embedded systems

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：