发明名称 Cache affinity scheduling method for multi-processor nodes in a split transaction bus architecture
摘要 A method of selecting a process to run within a multi-processor computer system in which the processors are arranged in nodes defined by node buses. A global run queue for queuing unaffined processes of the computer system is provided. A nodal run queue for each node bus of the computer system is provided. Highest priority levels in the global run queue is determined by one of the processors. Highest priority levels in the nodal run queue of the one processor is determined by the one processor. A process having a highest priority within the highest priority levels of the global and nodal run queues is determined by the processor. Finally, the process is executed by the one processor. Only when the processor finishes searching the global run queue and its nodal run queue for a process to run and does not find one, does it search within other nodal run queues that are found to be busy.
申请公布号 US5784614(A) 申请公布日期 1998.07.21
申请号 US19970841183 申请日期 1997.04.28
申请人 NCR CORPORATION 发明人 DAVIS, TODD C.
分类号 G06F9/50;G06F12/08;(IPC1-7):G06F15/16 主分类号 G06F9/50
代理机构 代理人
主权项
地址