Author: Robert Hyatt
Date: 11:47:44 02/22/99
Go up one level in this thread
On February 22, 1999 at 13:44:15, Peter McKenzie wrote: >On February 22, 1999 at 06:36:42, Peter McKenzie wrote: > >>On February 22, 1999 at 04:09:49, Will Singleton wrote: >> >><snip> > >>>7k/2pn1Q2/2r4p/p1qpp3/Pp1b3N/1P1P2PP/2P1R2B/7K w - - >> >>LambChop likes Qxd7 after completing depth 1, 2, and 3. >>Then it switches to Nf5, which it sticks with for depths 4-8. >>Then back to Qxd7 for depths 9-10. >> >>Here we have the search output for depths 8-10 (output is depth, score >>[centipawns], time in 100ths of seconds, and node count to complete depth): > >Actually, the score is in 10ths of pawns, not centipawns... > >> Running on my PII/300 notebook, crafty fails low (from +4 to +1) at depth=8, 3.6 seconds. It switches to Qd7 +5 and stays there. depth8 takes 10 secs, 9 takes 19.94, 10 takes 38.63 11 takes 1:26 and 12 takes 4:00. all with a score of > 5... The SE version still takes 8 plies on this position, but it isn't nearly so fast for plies > 8. >> 8 532 2700 295950 h4f5 c5f8 f7d7 c6g6 f5d4 f8f1 h2g1 g6g3 d7e8 h8h7 e8e7 g3g >>7 >> 9 395 28200 3776867 f7d7 c5f8 e2g2 c6d6 d7c7 d4b6 c7b7 f8f1 g2g1 f1f8 >>10 417 57000 7680080 f7d7 c5f8 h2g1 f8f1 e2g2 c6f6 h1h2 d4g1 g2g1 f1f2 g1g2 f6f >>7 d7d5 >> >>Note the huge increase in node counts between depth 8 and 9! > >I've had a better look at the position now. >The problem with Nf5 is that after Qf8 Qxd7 Rd6! black is back in the game (Nxd6 >Qf1+, or Qxc7 Qxf5). >So let me see: >Nf5 Qf8 Qxd7 Rd6 Qxc7 Qxf5 Qb8+ Kh7 Qxd6, and still a rook up. > >this line is 9 ply, minus 1 for a check extension, so 8 ply. >In this line, another ply of fullwidth is needed to see Qf1+ mating (unless you >look at checks in the q-search) which explains why our programs need 9 ply >to see that Nf5 isn't so hot. > >Anyone using singular extensions able to try this one? >For example, Qf8 looks singular. > >Peter > >> >>cheers, >er
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.