论文部分内容阅读
问题的提出并行处理机已发展了30年,在不断追求高性能的应用需求推动下,随着众多新技术、新工艺的实现和诞生,并行处理技术取得了长足的进步。cc-NUMA(Cache Coherence-Non Uniform Memory Access,高速缓存一致性非均匀存储访问结构)解决了并行处理系统的可扩展性及可编程性问题,同时开始满足众多应用的需求。目前主要的并行处理机可以分为两类:共享存储型并行处理机及分布式存储型并行处理机。对于共享存储型并行处理机,整个系统只用一个存储空间,多个处理器共享一个存储空间,这种类型的机器
PROBLEM PROBLEM Parallel processors have been developed for 30 years. With the continuous pursuit of high-performance application requirements, parallel processing technologies have made great strides with the implementation and birth of many new technologies and new technologies. cc-NUMA (Cache Coherence-Non Uniform Memory Access) addresses the scalability and programmability of parallel processing systems and begins to address the needs of many applications. At present, the main parallel processors can be divided into two categories: shared storage parallel processors and distributed storage type parallel processors. For shared storage type parallel processors, the entire system uses only one storage space and multiple processors share one storage space. This type of machine