Author: Uri Blass
Date: 01:45:32 02/23/05
Go up one level in this thread
On February 23, 2005 at 04:18:03, Dann Corbit wrote: >On February 23, 2005 at 03:07:35, Uri Blass wrote: > >>On February 23, 2005 at 00:51:25, Matthew Hull wrote: >> >>>I've noticed that GNUChess ponders from the very first move of the game, even >>>though it is still in book. Is this a usual practice? >>> >>>Does this give some advantage to the search when the game drops out of book, >>>since the hash table may have some relevant entries? >> >>I do not do it in Movei. >> >>Movei is too weak to even consider to implement such stuff that maybe gives 1 >>rating point. >> >>I think that it is significant work to check for no bugs in such code and in the >>same time of working on the code I can get more rating points from implementing >>different stuff. > >The point of first exit from the book is very important. Many programs give a >double time increment on book exit. This is a very clever strategy by GnuChess >and I think it may be more valuable than you imagine. > >Of course, you can only do it in ponder mode. I do not think that the first move out of book is more important to use time in it than other moves(I see no logical reason to use in move 11 twice the time that you use in move 12 and not using in move 12 twice the time that you use in move 13 and the fact that the program is out of book first in move 11 and not in move 12 may be dependent in the choice of the book). In part of the cases the program does not predict correctly the opponent move when it is out of book so my estimate is that it is not going to give you more than 1 elo. Uri
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.