检索结果-内蒙古大学图书馆

Annual international symposium on Computer Architecture, ISCA

作者： Xiaqang Qiu M. Dubois Department of Electrical Engineering-Systems University of Southern California Los Angeles CA USA

ILP processors can execute a large number of instructions at the same time. Thus it becomes more and more difficult to support traps efficiently. On the other hand a current trend in architecture is to support various memory functions in software rather than hardware, usually by trapping the execution processor on a cache miss, TLB miss or a failed access to a local or remote memory. These late memory traps block the faulty instruction at the top of the active list, backing up the pipeline. Moreover the support for late memory traps may affect the performance of non-faulting memory instructions as well. In this paper we analyze the overhead caused by late memory traps in ILP processors and define several measures for this overhead. In order to tolerate late memory traps, we propose hardware prefetching of exception conditions and a tagged store buffer to implement deferred traps on stores. We show that, with these hardware optimizations, the overhead added by the lateness of traps is significantly reduced relative to the overhead of early traps. Because of caching effects the frequency of late memory traps usually decreases as they are taken deeper in the memory hierarchy and their overall impact on the execution time becomes negligible.

关键词： Concurrent computing High performance computing Computer aided instruction distributed computing Computer architecture Computer languages parallel processing Software performance Writing

来源：评论

学校读者我要写书评

暂无评论

On the duality of data-intensive file system design: Reconciling HDFS and PVFS

On the duality of data-intensive file system design: Reconci...

引用

Supercomputing Conference

作者： Wittawat Tantisiriroj Seung Woo Son Swapnil Patil Samuel J. Lang Garth Gibson Robert B. Ross Carnegie Mellon University USA Argonne National Laboratory USA

Data-intensive applications fall into two computing styles: Internet services (cloud computing) or high-performance computing (UPC). In both categories, the underlying file system is a key component for scalable application performance. In this paper, we explore the similarities and differences between PVFS, a parallel file system used in UPC at large scale, and HDFS, the primary storage system used in cloud computing with Hadoop. We integrate PVFS into Hadoop and compare its performance to HDFS using a set of data-intensive computing benchmarks. We study how HDFS-specific optimizations can be matched using PVFS and how consistency, durability, and persistence tradeoffs made by these file systems affect application performance. We show how to embed multiple replicas into a PVFS file, including a mapping with a complete copy local to the writing client, to emulate HDFS's file layout policies. We also highlight implementation issues with HDFS's dependence on disk bandwidth and benefits from pipelined replication.

关键词： Servers Layout Semantics Web and internet services distributed databases Prefetching Cloud computing

来源：评论

学校读者我要写书评

暂无评论

Scalable matching and ranking for network search

Scalable matching and ranking for network search

引用

international Conference on Network and Service Management

作者： Misbah Uddin Rolf Stadler Alexander Clemm ACCESS Linnaeus Center KTH Royal Institute of Technology Cisco Systems San Jose California USA

Network search makes operational data available in real-time to management applications. In contrast to traditional monitoring, neither the data location nor the data format needs to be known to the invoking process, which simplifies application development, but requires an efficient search plane inside the managed system. The search plane is realized as a network of search nodes that process search queries in a distributed fashion. This paper introduces matching and ranking for network search queries. We are proposing a semantic for matching and ranking, which is configurable to support different types of management applications-from exact matching for database-style queries to loose, approximate matching, which is appropriate for exploratory purposes. We describe an echo protocol for efficient distributed query processing that supports matching and ranking. Further, we present the design of a search node, which maintains a real-time database of operational information and allows for parallel processing of search queries. A prototype implementation on a cloud testbed shows that the network search system, on a 9-node cluster with 24 core servers, executes 200 global search queries/sec with the 75th percetile latency below 100 milliseconds and with a CPU utilization below 5%. The performance measurements, together with our design, suggest that a system of 100,000 servers processing the same load would exhibit the same overhead per server and a query latency of below 1 sec.

关键词： Measurement Search problems Peer-to-peer computing databases Protocols Web search Semantics

来源：评论

学校读者我要写书评

暂无评论

Analyzing the impact of computational heterogeneity on runtime performance of parallel scientific components

Analyzing the impact of computational heterogeneity on runti...

引用

2007 Agent Directed Simulation symposium, ADS 2007

作者： Chandra, Sumir Parashar, Manish Ray, Jaideep Dept. of Electrical and Computer Engineering Rutgers University NJ United States Advanced Software R and D Sandia National Laboratories CA United States

ISBN: (纸本)1565553136

Scientific simulations modeling complex physical phenomena exhibit varying degrees of spatiotemporal and computational heterogeneity, which can pose significant challenges in algorithmic efficiency and runtime performance. Addressing these challenges often requires an understanding of application behavior and the impact of heterogeneity on the simulation, especially when analytical approaches are not feasible. Runtime calibration is used to analyze the impact of computational heterogeneity on the performance of two different load balancing strategies applied to two different problems of 2-D methane-air combustion using different chemistry models. Experimental evaluation demonstrates that such empirical approaches are inevitable in component-based scientific computations, where performance is determined by the problem characteristics and the particular connectivity of components in the simulation code, none of which are known before runtime.

关键词： Combustion

来源：评论

学校读者我要写书评

暂无评论

Immersive and interactive exploration of billion atom systems

Immersive and interactive exploration of billion atom system...

引用

IEEE Annual international symposium Virtual Reality

作者： A. Sharma Xinlian Liu P. Miller A. Nakano R.K. Kalia P. Vashishta Wei Zhao T.J. Campbell A. Haas Concurrent Computing Laboratory for Materials Simulations Biological Computation and Visualization Center Department of Computer Science Department of Physics & Astronomy Louisiana State University LA USA Naval Oceanographic Office Major Shared Resource Center Stennis Space Center Logicon Inc. MS USA

ISBN: (纸本)0769514928

Recent advances in parallel computing have made it possible for scientists to perform atomistic simulations of materials involving billions of atoms. An immersive and interactive virtual environment such as ImmersaDesk is an ideal platform for exploring complex material processes in these simulations. However rendering such large datasets at an interactive speed is a major challenge. To solve this problem we have developed a visualization system by incorporating parallel and distributed computing paradigms. The system uses a parallelized fast visibility-culling algorithm based on the octree data structure to reduce the number of atoms sent to the graphics pipeline. An adaptive multiresolution algorithm based on atomic density is employed to further reduce the load on the graphics pipeline. The resulting system renders a billion-atom system at nearly interactive frame rates on a dual processor SGI Onyx2 with an InfiniteReality2 graphics pipeline connected to a 4-node PC cluster.

关键词： Graphics Pipelines Computational modeling parallel processing Virtual environment Data visualization distributed computing Data structures Clustering algorithms Rendering (computer graphics)

来源：评论

学校读者我要写书评

暂无评论

Job scheduling that minimizes network contention due to both communication and I/O

Proceedings of the International Parallel Processing Symposi...

引用

proceedings of the international parallel Processing symposium, IPPS 2000年 457-463页

作者： Mache, Jens Lo, Virginia Garg, Sharad Lewis & Clark Coll Portland United States

As communication and I/O traffic increase on the interconnection network of high-performance systems, network contention becomes a critical problem drastically reducing performance. Whereas earlier allocation strategies were either sensitive to communication alone or sensitive to I/O alone, we present a new strategy that is sensitive to both communication and I/O. Our new strategy, MC-Elongated, strives to achieve (1) the compactness needed to minimize communication-based contention as well as (2) the balance and orientation relative to I/O nodes needed to minimize I/O-based contention. We tested our new strategy using synthetic workloads and a real workload trace of 6087 jobs captured from a 400 node Intel Paragon. Our results show that with respect to system throughput and average job turnaround time, in environments with varying degree of communication and I/O traffic, MC-Elongated outperforms previous allocation strategies that are in use today. Regarding the tension between communication and I/O, our results show that spatial layout is more critical for I/O-intensive jobs at lower utilization levels and more critical for communication-intensive jobs at higher utilization levels;and that in general, the impact of I/O traffic is dominant.

关键词： distributed computer systems

来源：评论

学校读者我要写书评

暂无评论

Adaptive admission control of multimedia traffic in high-speed networks

Adaptive admission control of multimedia traffic in high-spe...

引用

IEEE international symposium on Intelligent Control (ISIC)

作者： S. Jagannathan A. Tohmaz A. Chronopoulos H.G. Cheung Embedded Systems and Networking Laboratory Department of Electrical and Computer Engineering University of Missouri Rolla MO USA Parallel and Distributed Systems Lab Department of Computer Science University of Texas-San Antonio TX USA

This paper proposes a novel real-time adaptive admission control (AAC) scheme with a desired quality of service (QoS) guarantee and high network utilization in high speed networks. The QoS is given in terms of service delay, which is defined as the time it takes for a source to get admitted into the network after it initiates its intended request, packet/cell losses, and transmission delay (time taken to complete transmission from its initiation). AAC uses the following information - the available capacity from a novel adaptive bandwidth estimation scheme, a congestion indicator derived from a congestion controller, peak cell rate estimate from new sources, along with the desired QoS metrics, and outputs an 'admit' or 'reject' decision signal to the new sources while guaranteeing QoS and network utilization. Simulation results are presented by streaming ON/OFF and video data into the network. Results show that the proposed AAC admits significantly more traffic compared to other available admission control schemes thereby guaranteeing high network utilization while maintaining the desired QoS.

关键词： Programmable control Adaptive control Admission control Communication system traffic control High-speed networks Quality of service Delay effects Propagation losses Bandwidth Streaming media

来源：评论

学校读者我要写书评

暂无评论

A 3-phase series active power filter with compensate voltage drop and voltage unbalance

A 3-phase series active power filter with compensate voltage...

引用

IEEE international symposium on Industrial Electronics (ISIE)

作者： Seok-Woo Han Seung-Yo Lee Gyu-Ha Choe Department of Electrical Engineering Konkuk University Seoul South Korea

ISBN: (纸本)0780370902

In this paper, a three-phase series active power filter to compensate harmonic current, voltage drop and unbalanced voltage in power distribution systems is studied. The main circuit of the APF consists of a voltage source inverter with space vector modulation and a high pass filter connected in parallel to the power system. The APF is connected in series to the transmission line through a single-phase transformer with a turns ratio of 1:1. Here, the load is a six-pulse thyristor rectifier. The phase angle detected in order to generate the reference voltage at the load terminal is synchronized with the positive sequence component of the unbalanced source by using symmetrical component transformation. The proposed system has harmonic isolation between the source and load, voltage regulation and unbalance compensation. To verify the validity of the proposed compensating system, both computer simulations and experiments are carried out.

关键词： Active filters Voltage Power harmonic filters Power system harmonics Power distribution Inverters Power transmission lines distributed parameter circuits Phase transformers Thyristors

来源：评论

学校读者我要写书评

暂无评论

Design and development of high performance, scalable content based publish subscribe system: doctoral symposium 16

Design and development of high performance, scalable content...

引用

proceedings of the 10th ACM international Conference on distributed and Event-based systems

作者： M. A. Shah D. B. Kulkarni Walchand College of Engineering Vishrambag Sangli

ISBN: (纸本)9781450340212

In many application scenarios, content based pub/sub systems are required to provide stringent service guarantees such as reliable delivery, high performance in terms of throughput and low latency for event notification to interested subscribers. Matching algorithm play a critical role in content based pub/sub systems. The aim of our work is design and development of parallel, scalable and high performance content based publish subscribe system. We parallelize event processing using thread based and multi GPU approaches. We achieved low latency and high throughput when pub/sub is deployed on Apache Storm, a real time event processing system. Throughput gain and reduction in matching time is nearly 48% and 40% respectively in multi GPGPU approach of event processing compared to earlier work mentioned in [1].

关键词： throughput matching algorithm matching time multiple GPGPU real time event processing system data decomposition

来源：评论

学校读者我要写书评

暂无评论

System for database reports generating

System for database reports generating

引用

international Forum on Strategic Technology, IFOST

作者： P.F. Tarassenko M.S. Bukharova Department of Applied Mathematics and Cybernetics Tomsk State University Russia Sibgeoinf ormatics Limited Tomsk Russia

Flexible report generation is a typical problem for database applications in business, management and administration. The subject of the paper is the new system developed at Tomsk State University and Sibgeoinformatics Ltd. The system is targeted to small enterprises and offices. It can produce sophisticated ready-to-use reports, which could be created and modified easily without the necessity for the user to make routine work and learn much about SQL queries. As a result, the development of a report is focused on describing tables with columns/rows and sub-columns/sub-rows, rules of data selection and summarising for the report, particularly, table or column/row, and formatting the report in the destination document. The distinctive features of the system are: an XML-based report definition language; report interpreter, independent of data source and destination; adjustable data interface tested with popular data sources (MS Excel, Oracle, MS Access) and report destinations (MS Excel, MS Word); WYSIWYG report description editor (under construction); ability to describe multi-dimensional tables in the report; ability to create tables with nested columns and rows; a wide set of traditional data selectors, statistics, functions, and expressions; usual set of sorting and formatting features; performing optimisation of SQL code by minimisation of selection conditions and splitting or joining SQL queries.

关键词： databases Geology Data warehouses distributed information systems Biomedical engineering Data engineering Humans Aerospace engineering Laboratories Seminars

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：