[CHiPPS] How To Record Information of Alps Processed nodes on A parallel Cluster?
Yan Xu
Yan.Xu at sas.com
Fri Jul 6 09:21:53 EDT 2012
In AlpsKnowledgeBroker.h, there are members like
/** The number of nodes that have been processed. */
int nodeProcessedNum_;
/** The number of nodes that have been branched. */
int nodeBranchedNum_;
/** The number of nodes that have been discarded before processing. */
int nodeDiscardedNum_;
/** The number of nodes that are pregnant. */
int nodePartialNum_;
/** To record how many nodes processed by the system
(used in parallel code). */
int systemNodeProcessed_;
Probably, you can also take a look at AlpsKnowledgeBrokerMPI.cpp and find out how the systemNodeProcessed_ gets computed.
From: chipps-bounces at list.coin-or.org [mailto:chipps-bounces at list.coin-or.org] On Behalf Of ???
Sent: Wednesday, July 04, 2012 4:44 AM
To: chipps at list.coin-or.org
Cc: geogrid at gmail.com
Subject: [CHiPPS] How To Record Information of Alps Processed nodes on A parallel Cluster?
Dear professors,
I am know working on record the information of processed nodes in parallel ALPS.
I print the information of every node in AlpsSubTree.cpp, exploreunitwork() and ramup() where there is "++numNodesProcessed" to the stdout.
But finally, the number of nodes printed(the cluster will collect the stdout of all machines and write to one file ) is much less than that in the log file where search result is summarized.
While in a serial system, the two number are the same.
I don't know where is wrong. Could you help me?
Best Wishes!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/chipps/attachments/20120706/e82fc57f/attachment.html>
More information about the CHiPPS
mailing list