发明授权
US5280606A Fault recovery processing for supercomputer 失效
超级计算机故障恢复处理

Fault recovery processing for supercomputer
摘要:
In a high speed computer having a memory and a plurality of arithmetic processors divided into groups, the arithmetic processors of each group being connected to the memory in a hierarchical order in a master-subordinate relationship, the memory and the arithmetic processors generates an alarm signal indicating a failed part of the memory and each of the arithmetic processors. During a fault recovery process, a test program is performed on the computer to determine if it is properly functioning. If a favorable result is indicated, the computer is restarted in an original system configuration. Otherwise, part of the arithmetic processors is isolated from the computer depending on the alarm signal to degrade the computer into a first degraded system configuration. The test program is performed again on the first degraded system configuration. If the second test produces a favorable result, the computer is restarted in the first degraded system configuration. Otherwise, one or more of the arithmetic processors are isolated from the computer depending on the alarm signal so that the computer is degraded into a second degraded system configuration.
信息查询
0/0