公开(公告)号：US20100017735A1

公开(公告)日：2010-01-21

申请号：US12218382

申请日：2008-07-15

申请人： Philip J. Erickson ,  Terrence Vincent Powderly ,  Andrew F. Sanderson ,  Gary L. Schwartz ,  Sebastian Serrano ,  James A. Sievert

发明人： Philip J. Erickson ,  Terrence Vincent Powderly ,  Andrew F. Sanderson ,  Gary L. Schwartz ,  Sebastian Serrano ,  James A. Sievert

IPC分类号： G06F15/76 ,  G06F3/048 ,  G06F9/02

CPC分类号： G06F9/5077

摘要： In general, techniques for performing decentralized hardware partitioning within a multiprocessing computing system are described herein. More specifically, the multiprocessing computing system comprises first and second independent computing cells, where the first cell comprises a first processor that calculates a partition identifier. The partition identifier uniquely identifies a partition to which the first cell belongs. The first cell also comprises a second processor that establishes the partition within the multiprocessing computing system based on the partition identifier, and executes a single operating system across the partition. In the event the established partition successfully includes the first and second cells, the first and second cells execute the single operating system across the partition. Because the cells themselves perform the partitioning process, scalability may be achieved more easily. Moreover, the overall cost of the system may be reduced because a dedicated processor may no longer be required to perform partitioning.

摘要翻译： 通常，这里描述了用于在多处理计算系统内执行分散硬件分区的技术。 更具体地，多处理计算系统包括第一和第二独立计算单元，其中第一单元包括计算分区标识符的第一处理器。 分区标识符唯一地标识第一小区所属的分区。 第一单元还包括基于分区标识符在多处理计算系统内建立分区的第二处理器，并跨越分区执行单个操作系统。 在建立的分区成功地包括第一和第二小区的情况下，第一和第二小区跨越分区执行单个操作系统。 因为单元本身执行分区过程，所以可以更容易地实现可扩展性。 此外，由于专用处理器可能不再需要执行分区，所以系统的整体成本可能会降低。