Author: Robert Hyatt
Date: 09:52:47 04/20/04
Go up one level in this thread
On April 20, 2004 at 11:50:29, Dieter Buerssner wrote: >On April 20, 2004 at 06:10:05, Omid David Tabibi wrote: > >>In his article "PEASANT: An endgame program for kings and pawns", Newborn >>writes: "Position 70 would require a 30-ply search (25,000 hours)" > >I did the experiment. A search without transposition tables, without >pruning/extensions and with material only eval (I forgot, if I used qsearch or >not). A pawn capture was found at depth 26 (after 8 hours, IIRC). I assume you mean depth=26, not ply=26? IE white wins the pawn and I had thought that this happens on ply=27, which means the first ply of q-search. I will try to run this myself as it would be nice to know exactly how deep this is precisely, verified by multiple programs... > With hash, it >is almost guaranteed, that you find it at lower depth. Every second ply, you >will have to search all moves, and many inferior moves will be refuted by seeing >the pawn capture earlier. These refutations will be in the HT, and will be >grabbed in the other more decent lines, to find the solution at lower depth. > >For my engine, even 1000 entries in the HT is enough, to solve the problem in >practically no time. Theoretically if you search a perfectly ordered tree, the hash table should not let you solve it at a shallower than normal depth, although it should cut the time dramatically as we all see... > >Regards, >Dieter
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.