Sakaji
User
Junior Member
| Posts: 28 |   |
Karma: 0
|
re:X3D Photos of the actual X3D Fritz machine - 2006/12/10 19:37
I'm not trying to shoot the messenger!
But, whoa there! An eighteen-ply search in four minutes *from the start position*. I assume this must be without the opening book turned on, right? Because if the opening book is being used, this oft-quoted figure of eighteen ply in four minutes is completely misleading.
But even without the book, it's misleading. There are only twenty poss- ible first moves for each player, and an average of about 23 for the second move[1]. In a typical middle-game position, there's an average of about forty legal moves, I think. This means that searching from the start position is about twelve times easier than searching from a general position, which equates to about an extra 3.5 ply.
I grant that it's not entirely obvious from the text that Chuck in Minot quoted whether `in the starting position' applies to just the figure of
3.4Mnodes/sec or also to the 18-ply search in four minutes. I think it's meant to apply to both, though -- the third-to-last picture on the
Chessbase webpage shows the results of a Fritz search from the starting position that ran at 3435knodes/sec and in which every line was analysed to depth 18.
Dave.
[1] To arrive at this figure, I counted the total number of possible moves for white after each of white's possible first moves, assuming that black doesn't move at all. For white's second move, black's first move has very little effect. Sure, after 1.e4 e5, white can't play e5 himself but, on the other hand, after 1.e4 d5 or 1.e4 f5, white has a capture that I'm not counting. So I think my estimate is good enough for the purposes at hand. Even if you disagree with 23, go for some- thing like 25 and you'll get the same kind of answer..
---------
Death is nothing, but to live defeated and inglorious is to die daily.
Popular posts by Sakaji
Will Chessbase address software ...
Computer defeated
X3DFritz vs Kasparov, what concl...
|
post new topic
