首页> 外国专利> PRIVATE CACHE MISS AND ACCESS MANAGEMENT IN A MULTIPROCESSOR COMPUTER SYSTEM EMPLOYING PRIVATE CACHES FOR INDIVIDUAL CENTRAL PROCESSOR UNITS AND A SHARED CACHE

PRIVATE CACHE MISS AND ACCESS MANAGEMENT IN A MULTIPROCESSOR COMPUTER SYSTEM EMPLOYING PRIVATE CACHES FOR INDIVIDUAL CENTRAL PROCESSOR UNITS AND A SHARED CACHE

机译:多处理器计算机系统中的私人缓存错误和访问管理,该系统采用单个中央处理器单元和共享缓存的私人缓存

摘要

Computer system including group of CPUs (22), and CPU bus (10) coupled to a private caches in the CPUs (22) and to shared cache (11). Each private cache includes a cache controller having a processor directory (31) for identifying information blocks resident in cache memory (40), a cache miss output buffer (32) for storing the identifications of blocks to be swapped out of cache memory (40), a cache miss input buffer stack (35) for storing the identifications of all blocks to be swapped out from all the CPUs (22), a comparator (34) for comparing the identifications in the cache miss output buffer stack (32) with the identifications in the cache miss input buffer stack (35) and control logic, that responsive to the comparator (34) sensing a compare inhibits the broadcast of a swap operation onto the CPU bus (10) and converts the swap operation into a 'siphon' operation to the requesting CPU.
机译:计算机系统包括一组CPU(22),以及CPU总线(10),CPU总线(10)耦合到CPU(22)中的专用缓存和共享缓存(11)。每个专用高速缓存包括:高速缓存控制器,其具有用于识别驻留在高速缓存存储器(40)中的信息块的处理器目录(31);用于存储要从高速缓存存储器(40)中交换出的块的标识的高速缓存未命中输出缓冲器(32)。高速缓存未命中输入缓冲器栈(35),用于存储将要从所有CPU(22)换出的所有块的标识;比较器(34),用于将高速缓存未命中输出缓冲器栈(32)中的标识与处理器进行比较。高速缓存未命中输入缓冲区堆栈(35)和控制逻辑中的标识,响应于比较器(34)感测到比较,禁止在CPU总线(10)上广播交换操作,并将交换操作转换为“虹吸”对请求CPU的操作。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号