论文部分内容阅读
曙光2000-Ⅰ是国家智能计算机研究开发中心研制的可扩展超级服务器系统,它以曙光天演工作站为计算节点机,节点 CPU 采用 Motorola 公司的300MHz PowerPC 604e,节点之间通过10/100Mbps 高速以太网和100MBps 的蛀洞路由芯片组成二维网络互连。系统可配置4到128个计算节点,一个控制台节点。在32个计算节点省缺配置下,整机峰值速度达到200GFLOPS(即每秒200亿次综合运算)。1月12日,在“曙光2000-1超级服务器暨’98新成果汇报会”上,曙光信息产业有限公司总工程师徐志伟介绍了该机的设计思想和体系结构。徐总介绍说,曙光采用了可扩展机群体系结构,实现了单一系统映象。简单地讲,可扩展机群体系结构是指将一组节点计算机通过高性能互
Shuguang 2000-Ⅰ is a scalable super server system developed by National Intelligent Computer Research and Development Center, which uses Shuguang TianYing workstation as a computing node machine and a node CPU adopts Motorola’s 300MHz PowerPC 604e. The nodes communicate with each other through 10 / 100Mbps high speed Ethernet And 100MBps wormhole routing chip composed of two-dimensional network interconnection. The system can be configured from 4 to 128 computing nodes, a console node. In the absence of 32 compute nodes configuration, the total peak speed of 200GFLOPS (that is, 20 billion times per second integrated computing). On January 12, at “Dawning 2000-1 super server and ’98 new results briefing session,” Xu Zhiwei, chief engineer of Shuguang Information Industry Co., Ltd., introduced the design idea and architecture of the machine. Xu said the introduction of the dawn of a scalable cluster architecture to achieve a single system image. Simply put, scalable cluster architecture refers to a set of node computers through high-performance mutual