概述 SGI的Origin2000服务器是根据客户的具体应用和问题规模而设计的。这是采用CC-NUMA(高速缓存一致性非均衡存储器访问)体系结构实现的。该体系结构把多个处理器互连在一起,连接数目随系统规模扩大而增长,从而可以避免基于总线的对称多处理(SMP)体系结构对带宽的限制,由于采用超立方体和多维互连特性,使系统的可伸缩性也达到了空前未有的水平。 基本的积木块 Overview SGI’s Origin2000 Server is designed for customer specific applications and problem size. This is achieved using the CC-NUMA (Cache Consistent Non-Balanced Memory Access) architecture. The architecture interconnects multiple processors and the number of connections grows as the system grows in size to avoid bandwidth constraints for bus-based symmetric multiprocessing (SMP) architectures. Due to the use of hypercube and multidimensional interconnect features , The scalability of the system has reached an unprecedented level. Basic building blocks