A big multicore program cant repeat the move. Timing differences between nodes and communication delays can make it nondeterministic. For any program, keeping data from prior searches makes it hard to do a new search in isolation and get the same result. If random seeds are not kept for each move, the random search will be different. It's not unusual for some of the top few moves to be much better than others.
David > -----Original Message----- > From: computer-go-boun...@computer-go.org [mailto:computer-go- > boun...@computer-go.org] On Behalf Of Jacques Basaldúa > Sent: Monday, February 02, 2009 1:20 PM > To: computer-go@computer-go.org > Subject: Re: [computer-go] Rules for remote play at the Computer Olympiad > > > About the "thinking process" log. > > Enabling debugging options can result in serious performance loss. In my > system > only the "admin thread" can do such things as tree dumps and that makes > all other > "pawn threads" idle. I don't think such preventive measures are > justified. In case > of doubt, it should be enough if the author can show that the program > can repeat > any suspectful move (even if it does not always play the same move, the > played > move should at least be among the best). If the program is local that > should be > enough. Remote programs cannot be controlled anyway. I think adding > constraints to local programs makes the unfairness vs remote programs even > worse. In case something has to be implemented it must be announced in > advance. > > Questions: > > 1. What are the time settings for 19x19? > > 2. What are the days for 19x19? > > 3. Is hardware available from the organizers? At least, monitors and > keyboards to > avoid flying with non-critical and voluminous equipment. > > > Jacques. > > _______________________________________________ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ _______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
