关键词： 并行处理系统   计算机   性能  

摘要： An important application area of modem computer systems is that of digital signal processing. This discipline is concerned with the analysis or modification of digitally represented signals, through tiie use of simple matiiematical operations. A primary need of such systems is that of high data throughput. Although optimised programmable processors are available, system designers are now looking towards parallel processing to gain further performance increases. Such parallel systems may be easily constructed using the transputer family of processors. However, although these devices are comparatively easy to program, they possess a general von Neumann core and so are relatively inefficient at implementing digital signal processing algorithms. The power of the transputer lies in its ability to communicate effectively, not in its computational capability. The converse is true of specialised digital signal processors. These devices have been designed specifically to implement the type of small data intensive operations required by digital signal processing algorithms, but have not been designed to operate efficiendy in a multiprocessor environment This thesis examines the performance of both types of processors witii reference to a common signal processing application, multichannel filtering. The transputer is examined in both uniprocessor and multiprocessor configurations, and its performance analysed. A theoretical model of program behaviour is developed, in order to assess the performance benefits of particular code structures and the effects of such parameters as data block size. The transputer implementation is contrasted witii that of the Motorola DSP56001 digital signal processor. This device is found to be much more efficient at implementing such algorithms on a single device, but provides limited multiprocessor support. Using tiie conclusions of tiiis assessment, a hybrid multiprocessor has been designed. This consists of a transputer controlling a number of si

关键词： multiprozessor  

ISBN: (数字)9781439864937 ISBN: (纸本)9780867202298

摘要： This comprehensive work merges two of the hottest topics in computer science: parallel computing and computer graphics. Selected Topics from the Table of Contents: -Overview of Accelerated Rendering Techniques -Overview of Parallel Methods for Image Generation -Issues in Parallel Algorithm Development -Overview of Base Level Implementation -Compari

关键词： 多处理器   千兆级   总线   存贮器系统   存贮系统   处理器系统   VME   接口板  

摘要： 当前的VME卡盒中的千兆级翻转(Gigaflop)性能可能归功于Mercurry Compute-System公司的板级多处理器设计。采用4个i860处理器在9UVME卡上使用模块就可以作成MC860VS单板机(见图)。每一个处理卡能够运行到320MFLops,在一个以耦合处理的环境中,VME总线上可以具有高达8个的MC860 VS板,32个i860这一最大模式提供了2.5GFLops的最大处理能力。

关键词： 多处理器   图   最优算法   互连网络  

摘要： 已知一个无向图G(V,E),|V|=n.本文在SIMD机器-Hype-rcube上提出了计算图的连通分支和最小生成树的两个最优算法.若Hypercu-be由P个处理器组成,则上述两个算法的时间复杂性都是O(n^2/p),1≤p且PlogP≤n.

关键词： 多处理器   超级计算机   欧州  

关键词： 图象处理   多处理器  

摘要： The increasing speed of the most powerful computers, especially multiprocessors, makes it difficult to provide sufficien t I/O bandwidth to keep them running at full speed for the largest problems. T rends sho w that the di?erence in the sp eed of disk hardw are and the sp eed of pro cessors is increasing? with I?O sev erely limiting the p erformance of otherwise fast mac hines? This widening access?time gap is kno wn as the ?I?O b ottlenec k crisis?? One solution to the crisis? suggested b y man y researc hers? is to use man y disks in parallel to increase the o v erall bandwidth? This dissertation studies some of the ?le system issues needed to get high p erformance from parallel disk systems? since parallel hardw are alone cannot guaran tee go o d p erformance? The target systems are large MIMD m ultipro cessors used for scien ti?c applications? with large ?les spread o v er m ultiple disks attac hed in parallel? The fo cus is on automatic cac hing and prefetc hing tec hniques? W e sho w that cac hing and prefetc hing can transparen tly pro vide the p o w er of parallel disk hardw are to b oth sequen tial and parallel applications using a con v en tional ?le system in terface? W e also prop ose a new ?le system in terface ?compatible with the con v en tional in terface? that could mak e it easier to use parallel disks e?ectiv ely ? Our metho dology is a mixture of implemen tation and sim ulation? using a soft w are testb ed that w e built to run on a BBN GP???? m ultipro cessor? The testb ed sim ulates the disks and fully implemen ts the cac hing and prefetc hing p olicies? Using a syn thetic w orkload as input? w e use the testb ed in an extensiv e set of exp erimen ts? The results sho w that prefetc hing and cac hing impro v ed the p erformance of parallel ?le systems? often dramatically ?

关键词： 黑板结构   专家系统   多处理器  

摘要： 黑板结构是专家系统的一种流行构造框架。用这种结构建造的专家系统一般由知识源、控制器及称为黑板的共享数据区三大部分组成,其中知识源由控制器调度并通过黑板相互通讯。本文描述了黑板结构的一种扩充,使得知识源可以并发执行,因而适用于多处理器环境。