首页> 外国专利> Apparatus and method for improving cache consistency using a single copy of cache tag memory in a multiprocessor computer system

Apparatus and method for improving cache consistency using a single copy of cache tag memory in a multiprocessor computer system

机译:用于在多处理器计算机系统中使用高速缓存标签存储器的单个副本来改善高速缓存一致性的设备和方法

摘要

In a multiprocessor computer system, an apparatus and method for a cache controller is disclosed that maintains cache coherency in the structure of a cache memory having a single copy of a cache tag while supporting multiple good operations. The CPU includes a small internal cache memory structure.;Almost large external cache arrays are coupled to the CC via the CPU and the primary integrated address and databus.;The CC is in turn coupled to a second bus that interconnects with the processor, the I / O device, and the main memory, among other devices. The external cache is subblocked. The CC's cache directory tracks the use of external caches. The input buffer of the CC is connected to a first bus that provides buffering of instructions sent by the CPUs. The output buffer of the CC is coupled to the second bus to buffer the instructions indicated by the CC with the device operating on the second bus.;The virtual bus interface (VBI) receives an entry formed in the input buffer, thus reducing the input buffer to accept other commands. The cache invalidation queue (CIQ) register stores the address of the cache subblock that goes to the import invalidation operation. The address of the destination device is also read into the output buffer. If the address of the destination device stored in the output buffer matches the address of the CIQ register, the CC issues a read-invalidate instruction and the invalid block of the cache is filled with the data corresponding to the priority-access processor. Override intervening redundancy issued by the CPU accessing. Response times for roaming requests are bound and data consistency between cache and handler is preserved.
机译:在多处理器计算机系统中,公开了一种用于高速缓存控制器的设备和方法,该设备和方法在具有多个高速缓存标签的单个副本的高速缓存存储器的结构中保持高速缓存一致性,同时支持多个良好操作。 CPU包括一个小的内部高速缓存存储器结构。几乎所有大的外部高速缓存阵列都通过CPU以及主要的集成地址和数据总线耦合到CC。CC依次耦合到与处理器互连的第二条总线。 I / O设备以及主内存等设备。外部缓存被子块化。 CC的缓存目录跟踪外部缓存的使用。 CC的输入缓冲区连接到第一总线,该总线提供CPU发送的指令的缓冲。 CC的输出缓冲区耦合到第二总线,以在设备在第二总线上运行时缓冲CC指示的指令。虚拟总线接口(VBI)接收在输入缓冲区中形成的条目,从而减少输入缓冲区以接受其他命令。高速缓存无效队列(CIQ)寄存器存储要进行导入无效操作的高速缓存子块的地址。目标设备的地址也被读入输出缓冲区。如果存储在输出缓冲区中的目标设备的地址与CIQ寄存器的地址匹配,则CC发出读取无效指令,并且高速缓存的无效块将填充与优先级访问处理器相对应的数据。覆盖由CPU访问发出的中间冗余。漫游请求的响应时间受到限制,并且缓存和处理程序之间的数据一致性得以保留。

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