并行计算机

One important parallel-processing approach is the neural network, which mimics the architecture of the nervous system.

一种重要的并行处理方法是模仿神经系统结构的神经网络。另一个持续不断的趋势是计算机联网的增加。

It was composed of four parts,named control center,physical tester,virtual tester,and remote observer.

该平台由控制中心、真实试验机、虚拟试验机和远程观察器4个模块组成，各模块由计算机网络连接在一起，采用并行通讯模式实现控制中心与试验机和观察器之间的数据传输。

However, the capability of a single computer cannot fulfill the requirements of the large scale computing, and High Performance Computing becomes the only approach for it.

然而，单个计算机的能力远远不能满足大规模科学与工程计算的要求，高性能并行计算成为实现这种计算的唯一途径。

For the need of large scale parallel computing based on Data Grid that is composed of the multi-clusters, the single cluster grouping communication model for multi-level grouping communication has been researched.

软件简介：针对基于多计算机机群构成的网格的大规模并行计算的需要，对多级分组通信模型的单一机群分组通信进行了研究。

This research is supported by IBM and other outside sponsors, and requires the use of Rensselaer's Computational Center for Nanotechnology Innovations, which provides more than 100 teraflops of computing power through massively parallel Blue Gene supercomputers, POWER-based Linux clusters, and AMD Opteron processor-based clusters.

这项研究是支持由IBM和其他外界赞助，并要求使用伦斯勒的计算为中心的纳米科技创新，其中规定，超过100万亿次的运算能力，通过大规模并行的Blue Gene超级计算机，电力为基础的Linux集群，和AMD Opteron处理器为基础的集群。

The research of neural network realization can be traced back to 60's. It processes along with two avenues: fully hardware realization and virtual realization. Scientists have developed many neurochips supporting fully hardware realization. Neurocomputing simulating supporting environments, neurocomputing accelerators and parallel neurocomputing systems also have been developed for virtual realization of neural network.

神经网络实现技术的研究可以追朔到六十年代，该项研究一直是沿两个方向开展，即神经网络的全硬件实现和神经网络的虚拟实现，科学加们也研制出了许多支持神经网络全硬件实现的神经芯片，以及支持神经网络虚拟实现的各种神经计算软件模拟环境、加速器和并行神经计算机系统。

Busparallel expansion and pipeline technique are adopted to improve the average inputand output speed of recorder based on FLASH memory. 2 A concept of 'cluster' isintroduced. The using state message of each cluster is loaded in the spare area,following the target datum, to keep the pipeline sequence of multiple clusters' loading, which guarantee the average input speed of system. 3 Based on the using statemessage of every cluster, the Cluster Allocation Table established in memory enablethe system to manage the recorder flexibly in high speed, similar to a file managementsystem. 4 PowerPC 8245 based embedded system, along with Vxworks real-timeoperating system is introduced to make the system more easier to be applied to theprojects in the future.

本文所述的关键技术特点主要包括：采用了并行扩展以及流水线操作技术以提高基于FLASH芯片的存储区平均输入输出能力；引入了&簇&的概念，为了保证连续写入过程中的流水线顺序，系统向存储区加载每一簇有效数据后，紧接着还将向相应簇的空余区中加载该簇的使用状态信息；根据所有簇的使用状态信息在系统内存中建立&簇分配表&，可以支持系统快速灵活地管理存储区，即，类文件化的数据管理；采用基于PowerPC8245的嵌入式计算机硬件系统及基于Vxworks的嵌入式实时操作系统，提高了系统的整体性能及系统研发的延续性。

They could assemble a people's supercomputer by networking together a million Macintoshes, each one computing a coordinated little part of a huge, distributed decryption program.

他们可以集合一个民族的超级计算机，也就是把上百万台MAC电脑联结在一起，每台都只运行超大的分布式解密程序中并行的一小部分。

They could assemble a people's supercomputer by networking together a million Macintoshes, each one computing a coordinated little part of a huge, distributed decryption program.

他们可以集合一个民族的超级计算机，也就是把上百万台麦金托什联结在一起，每台都只运行超大的分布式解密程序中并行的一小部分。

These problems require great computer sources, so the technique of large-scale parallel process on multiprocessors has become the key to solve them.

这类问题对计算机资源的要求高，而且求解规模比较大，基于多个处理器的大规模并行处理技术已成为解决这类问题的关键技术。

This brought fixed cost, but it also is meant at the same time use a source to won't make you singlehanded assume a problem.

这带来了一定的成本，但它同时也意味着使用开源不会让你独力承担问题。

He gained a small fortune in real estate.

他在房地产上发了一笔小财。