检索结果-内蒙古大学图书馆

International conference on Massively parallel processing Using Optical Interconnections

作者： R. Kannan H.F. Jordan K.Y. Lee C. Reed Department of Mathematics and Computer Science University of Denver Denver CO USA Optoelectronic Computing Systems Center University of Colorado Boulder Boulder CO USA National Science Foundation Engineering Research USA Supercomputing Research Center Institute for Defence Analysis

We present a high-bandwidth multichannel time slot permuter handling N! permutations on N=2/sup n/ optical time division multiplexed (TDM) inputs arranged as /spl radic/N parallel frames with /spl radic/N time slots each. A generalized non-blocking property of the reverse banyan network is obtained and is used to derive a self-routing space-domain network and its counterpart in the time domain. The network is fully pipelineable and self-routing with distributed control obtained dynamically at each switch. The permuter uses O(/spl radic/N(n/sup 2/)) single-bit controlled 2/spl times/2 optical switching elements with feedforward and feedback fiber delay lines, (which is a factor of O(n) reduction in hardware compared to previously proposed multichannel time slot permuters), and provides higher throughputs as compared to earlier Benes based multichannel permuters. The network preserves frame integrity and has on initial frame delay which is linear in the number of time slots.

关键词： Time division multiplexing Optical feedback Bandwidth Optical switches Delay lines Optical fiber couplers Routing distributed control Optical control Hardware

来源：评论

学校读者我要写书评

暂无评论

A study of the communication cost of the FFT on torus multicomputers

A study of the communication cost of the FFT on torus multic...

引用

IEEE International conference on Algorithms and Architectures for parallel processing (ICAP)

作者： L.D. de Cerio M. Valero-Garcia A. Gonzalez Departament d' Arqitectma de Computadors University Politiknica de Catalunya Barcelona Spain Universitat Politecnica de Catalunya Barcelona Catalunya ES Dept. d'Arquitectura de Computadors Univ. Politecnica de Catalunya Barcelona Spain

The computation of a one-dimensional FFT on a c-dimensional torus multicomputer is analyzed. Different approaches are proposed which differ in the way they use the interconnection network. The first approach is based on the multidimensional index mapping technique for the FFT computation. The second approach starts from a hypercube algorithm and then embeds the hypercube onto the torus. The third approach reduces the communication cost of the hypercube algorithm by pipelining the communication operations. A novel methodology to pipeline the communication operations on a torus is proposed. Analytical models are presented to compare the different approaches. This comparison study shows that the best approach depends on the number of dimensions of the torus and the communication start-up and transfer times. The analytical models allow us to select the most efficient approach for the available machine.< >

关键词： Costs Message passing Memory management Multidimensional systems Programming profession Hypercubes Pipeline processing distributed computing Analytical models Computer networks

来源：评论

学校读者我要写书评

暂无评论

Price and Performance of Simulating Wind Instruments

Price and Performance of Simulating Wind Instruments

引用

Supercomputing conference

作者： P.A. Skordos Massachusetts Institute of Technology Cambridge MA USA

A performance-price ratio of 2.2 Gflops per million dollars is demonstrated in simulations of wind instruments using a cluster of 20 non-dedicated Hewlett-Packard workstations and a shared-bus Ethernet network (the performance-price ratio is calculated assuming that the workstations were dedicated). The present simulations of wind instruments are based on the compressible Navier Stokes equations, and have only recently become possible through the use of parallel computing and through developments in numerical methods. The distributed simulation system has been built directly on top of UNIX and TCP/IP, and includes automatic process migration from busy workstations to free workstations.

关键词： Instruments Workstations Computational modeling Acoustic waves parallel processing TCPIP Lattice Boltzmann methods Filters Fluid dynamics Ethernet networks

来源：评论

学校读者我要写书评

暂无评论

A neurocomputer based learning controller for critical industrial applications

A neurocomputer based learning controller for critical indus...

引用

IEEE/IAS International conference on Industrial Automation and Control, IA & C

作者： K.K. Shukla Department Of Computer Engineering Institute Of Technology Banaras Hindu University Varanasi India

Because of their remarkable ability to learn nonlinear mappings from a nonexhaustive training set and generalize, neurocomputers can readily play the role of learning controllers for complex plants. Due to the parallel distributed nature of processing they introduce little time delay in the control loop. In this paper a new generalized feedforward neural network model is applied to the task of controlling an aircraft engine model. A more effective learning algorithm based on adaptive optimization is presented which compares favorably with the classical backpropagation method. Similar controllers can be designed for critical industrial applications such as nuclear power plants etc.< >

关键词： Industrial control Aircraft propulsion Control systems Artificial neural networks Automatic control Neural networks Electrical equipment industry Digital control Application software Industrial training

来源：评论

学校读者我要写书评

暂无评论

Genetic algorithm based pattern allocation schemes for training set parallelism in backpropagation neural networks

Genetic algorithm based pattern allocation schemes for train...

引用

IEEE International conference on Evolutionary Computation

作者： Shou King Foo P. Saratchandran N. Sundararajan School of Electrical & Electronic Engineering Nanyang Technological University Singapore

Training set parallelization is an efficient method to optimize the training procedure performance of the backpropagation neural network algorithm. In training set parallelism, the training patterns are distributed 'optimally' among a heterogeneous array of processors, optimality criterion obtain the minimum training time per epoch. Earlier studies on heterogeneous transputers connected in a pipeline-ring topology have indicated that the above optimization problem results in a mixed integer programming problem and results in large computation time to find the optimal pattern allocations. In this paper, a genetic algorithm is used as an optimization tool to find the optimal allocation of patterns. The approach is illustrated using two benchmark problems, the 256-8-256 Encoder and NETTALK problems. Results indicate that when 'a priori' information is not used, the computation time needed by the genetic algorithm is comparable to that obtained by mixed integer programming. However, when 'a priori' information is used, the genetic algorithm results in significant reduction in computation time for finding the optimal solution.

关键词： Genetic algorithms parallel processing Backpropagation algorithms Linear programming Neural networks Intelligent networks Optimization methods network topology Multi-layer neural network Analytical models

来源：评论

学校读者我要写书评

暂无评论

PVM-based parallel COMPUTING - A CASE-STUDY ON POWER-PLANT SIMULATION

MICROPROCESSING AND MICROPROGRAMMING

引用

MICROprocessing AND MICROPROGRAMMING 1994年第10-12期40卷 875-878页

作者： FEMMINELLA, A OMODEO, A Centro Ricerca di Automatica ENEL Via Volta 1 I-20093 Cologno Monzese Milano ITALY

parallel computing architectures are a promising answer to the growing needs of computing power in many application fields. This model can be easily fitted to applications that are structured as a collection of weakly coupled processes. Every process evolves independently and synchronises periodically with the others by exchanging relatively small messages. The power plant simulators developed at our Research Center are an example of this kind of application. Due to the absence of standards in this field, our activity was focused upon the implementation of a distributed software platform, open to heterogeneous systems, and to be used as a reference for our future developments. In this paper, after a brief description of the simulation environment, we describe a first parallel version based on a Transputer network and then the current evolution based on a network of heterogeneous workstations.

关键词： parallel processing systems

来源：评论

学校读者我要写书评

暂无评论

Modelling logic programming using a network of communicating machines 2

Modelling logic programming using a network of communicating...

引用

2nd euromicro Workshop on parallel and distributed processing, PDP 1994

作者： Bird, P.R. Cowling, A.J. Department of Computer Science University of Sheffield SheffieldS10 2TN United Kingdom

ISBN: (纸本)0818653701

As part of a larger project concerned with the unification of programming paradigms, this paper describes the development of a model of computation for the relational paradigm, as represented by Prolog. The model is based on extended finite-state machines which are formed into a parallel network of communicating machines. Three main machine types have been identified, to correspond to different control aspects within Prolog, and the different versions of these are described. © 1994 IEEE.

关键词： Logic programming

来源：评论

学校读者我要写书评

暂无评论

parallel simulation of GaAs MESFET semiconductor devices with adaptive meshing on a transputer network 2

Parallel simulation of GaAs MESFET semiconductor devices wit...

引用

2nd euromicro Workshop on parallel and distributed processing, PDP 1994

作者： Tsang-Ping, C.S. Barry, D.M. Snowden, C.M. Department of Electronic and Electrical Engineering The University of Leeds LeedsLS2 9JT United Kingdom

ISBN: (纸本)0818653701

A parallel implementation of the numerical simulation of a GaAs MESFET device based on the energy transport model using the finite difference discretization scheme and solved by a point iterative method is presented. parallel techniques targeted at message-passing MIMD distributed memory architectures and in particular Transputers, are described. Efficient parallelism is achieved by the geometric decomposition of the problem domain. Issues on the convergence and efficiency of the solution with novel strategies such as iteration ordering techniques and communication protocols in a parallel environment are discussed. The parallel implementation of the adaptive grid refinement also requires load balancing techniques for optimum efficiency. A speedup of up to 11 was obtained on a network of 16 transputers using these techniques. © 1994 IEEE.

关键词： Transputers

来源：评论

学校读者我要写书评

暂无评论

A distributed system architecture for embedded control systems 2

A distributed system architecture for embedded control syste...

引用

2nd euromicro Workshop on parallel and distributed processing, PDP 1994

作者： Ciccarella, Gianfranco Patricelli, Frédéric Scuola Superiore G. Reiss Romoli S.p.A. Str. Prov. per Coppito L'Aquila67100 Italy

ISBN: (纸本)0818653701

The paper describes a distributed system architecture based on nodes (N-cube modules) connected by a Local Area network. The system has been developed to support the implementation of new services in a B-ISDN testbed and, in general, to provide a suitable solution to many design problems of computing systems for control applications. A distributed operating system, designed by an object-model approach, supports the application processes. Particular attention was dedicated to the design of the InterProcess Communication (IPC) module and of its self-routing algorithm. The IPC module protocol architecture is based on modifications of the ISOIOSI reference model and provides unbuffered and reliable, with an exactly-once semantic, message passing and Remote Procedure Call primitives. The service provided by these primitives is an acknowledged connectionless one (datagram), and the messages are of variable length. Finally, some considerations on the implementation of distributed applications, on testing experiments, on performances and on future developments are reported. © 1994 IEEE.

关键词： network architecture

来源：评论

学校读者我要写书评

暂无评论

A parallel distributed CONTROL ARCHITECTURE FOR MULTIPLE ROBOT SYSTEMS USING A network OF MICROCOMPUTERS

引用

COMPUTERS & INDUSTRIAL ENGINEERING 1994年第1-4期27卷 63-66页

作者： YASUDA, G TACHIBANA, K SETSUNAN UNIV NEYAGAWAOSAKA 572JAPAN

This paper presents the design and implementation of a parallel distributed control architecture for industrial multiple robot systems. The design methodology is based on a concept of discrete states and actions, and a robotic task is represented as a sequence of primitive actions. For cooperative or exclusive tasks at the synchronous level of multiple robot systems, Petri net representation is applied, and discrete event-driven control is implemented as a data flow network of concurrent processes communicating with each other. Implementation of multiprocessing control on a microcomputer and a network of microcomputers is discussed.

关键词： MULTIPLE ROBOTS PETRI NET COMMUNICATION parallel processing IMPLEMENTATION

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：