Author: Uri Blass
Date: 21:51:17 02/09/01
Go up one level in this thread
On February 09, 2001 at 11:36:04, Heiner Marxen wrote: >On February 09, 2001 at 06:51:15, leonid wrote: > >>On February 09, 2001 at 06:47:28, leonid wrote: >> >>>Hi! >>> >>>If you like to solve a forced mate, you can try this position: >>> >>>[D]1bqQBnRn/3N2Qb/Q1QN2np/1Q1Q1qpk/4Qqbn/2B1Qrnn/2Q5/K1R5 w - - >>> >>>Please, indicate your result. > >Chest find a mate in 7 with 1.Qxg5+: >Qxg5+ Nxg5 Qdxf5 Nhxf5 Qh2+ Bh3 Nf6+ Kh4 Qxh6+ Nxh6 Q4xf4+ Ng4 Qfxg5# > >(3.3 hours, K7/600 350MB hash) > >>>By curiosity I looked what is the NPS (nodes/second) for this position. Since >>>the number of moves, during the solution of mate went quit often up to 130 for >>>whites, NPS was more that impressive. If it could be the same all the time! On >>>It went up in around 60% as average. It is the same for your program? NPS depend >>>greatly on number of average moves existing in each ply? >>> >>>Thanks, >>>Leonid. > >For Chest on your crowded boards NPS greatly decreases, by a factor around >2 or 3. Some of the heuristics to gain speed, do not pay off, anymore. >Hmmm, may be I should improve my heuristics ;-) >The reduced NPS does not worry me, as long as the job is done quickly. > >Heiner I believe that removing part of the white pieces may help chest to find the solution more quickly I did not investigate exactly all the pieces that you can remove but without the white queen at a6 there is also a mate in 7. Here is a position that chess programs accept(I put a rook at g5 instead of a pawn and removed the white queen at a6) [D]1bqQBnRn/3N2Qb/2QN2np/1Q1Q1qrk/4Qqbn/2B1Qrnn/2Q5/K1R5 w - - 0 1 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.