> Another question ... does more playouts really provide a *consistent* > improvement on the ELO score, especially for those strongest > programs?
It does, on the levels we've reached so far. Anyone telling you what the chart (of improvement to playouts) looks like beyond the current level of the strongest programs is guessing ;-) > I remember that some programs running on laptop rank very > high in the Olympaids, that seems imply that speed simply doesn't > matter here ... I think speed matters, but it is still clearly the case that good algorithms matter much more (*): http://www.grappa.univ-lille3.fr/icga/tournament.php?id=216 (19x19 winner has fewer cores than places 2 to 6) http://www.grappa.univ-lille3.fr/icga/tournament.php?id=215 (9x9 winner on a notebook; you can make a good argument that opening book now matters as much as algorithms on 9x9) (http://www.grappa.univ-lille3.fr/icga/tournament.php?id=226 does not show hardware for 13x13, but from the other charts: 12 cores > 56 cores > 150 cores). http://www.grappa.univ-lille3.fr/icga/tournament.php?id=193 (19x19: 4 cores > 64 cores > 16 cores > 24 cores > 8 cores) http://www.grappa.univ-lille3.fr/icga/tournament.php?id=194 (9x9: 64 cores > 16 cores > 8 cores - yay! found an "expected" result!) (Of course, statistics are easy lie with; e.g. the Mogo numbers in the above 2010 tables are unreliable as they had communication problems, but I think my conclusion, that there is only small correlation with CPU power, is valid.) Darren *: If it didn't, most of us wouldn't be here :-) -- Darren Cook, Software Researcher/Developer http://dcook.org/work/ (About me and my work) http://dcook.org/blogs.html (My blogs and articles) _______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
