Parallel computer, a control method of the control program and the parallel computer of parallel computer

摘要

PROBLEM TO BE SOLVED: To restart a jab having been stopped during execution of barrier synchronization.;SOLUTION: A control method is implemented by a first node as one of a plurality of nodes of a parallel computer having the plurality of nodes. The control method includes the steps of: gathering information associated with progress states of barrier synchronism at the plurality of nodes from the respective nodes when a program of a job is stopped from being executed at each of the plurality of nodes; and determining a restart position of the program of the job at the first node on the basis of the stop position of the program of the job at the first node and the information associated with the progress states of barrier synchronism at the plurality of nodes.;COPYRIGHT: (C)2014,JPO&INPIT