Author: Martin Giepmans
Date: 17:10:29 11/22/02
Go up one level in this thread
On November 22, 2002 at 19:55:23, Omid David Tabibi wrote: >On November 22, 2002 at 19:43:10, Martin Giepmans wrote: > >>On November 22, 2002 at 18:09:28, Uri Blass wrote: >> >>>previous version found Rxb2 only at depth 14 >>>With verified null move pruning movei finds wac 2 at depth 12. >>> >>>[D]8/7p/5k2/5p2/p1p2P2/Pr1pPK2/1P1R3P/8 b - - 0 1 >>> >>>depth=12 +0.64 b3b2 >>>Nodes: 6559172 NPS: 215975 >>>Time: 00:00:30.37 >>> >>>depth=12 +2.67 b3b2 d2b2 c4c3 b2b6 f6g7 b6b7 g7h6 b7b6 h6h5 b6b5 c3c2 b5f5 h5g6 >>>f5g5 g6h6 g5c5 d3d2 c5c6 h6h5 c6c5 h5h4 c5c2 d2d1q c2e2 >>>Nodes: 17674291 NPS: 212303 >>>Time: 00:01:23.25 >>> >>>depth=13 +3.54 b3b2 d2b2 c4c3 b2b6 f6g7 b6b7 g7h6 b7b6 h6h5 b6b5 c3c2 b5f5 h5g6 >>>f5g5 g6f6 g5c5 d3d2 c5c6 f6e7 c6c7 e7d6 c7c2 d2d1q c2e2 h7h5 >>>Nodes: 30679118 NPS: 220032 >>>Time: 00:02:19.43 >>> >>>Uri >> >>I don't think that pure nullmove (without any verification) is safe >>in an endgame like this. So I'm not amazed that verified nullmove does better >>here. >> > >I don't think the improvement comes from zugzwang detection here, since I can >see no zugzwang after Rxb2. > >This can be checked by turning off zugzwang detection (the re-search part). > > >>Martin It's not necessarily zugzwang. Nullmove easily overlooks nasty passed pawn things in a position like this. I don't use nullmove at all in such endgames but switch to other pruning methods. On an AMD 2000 XP my program needs 3 seconds to find Rxb2 with a winning score. Martin
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.