Author: Andrew Williams
Date: 01:46:06 03/17/99
Go up one level in this thread
On March 16, 1999 at 17:51:47, Will Singleton wrote: > >On March 16, 1999 at 17:09:43, Andrew Williams wrote: > >>Hi Will, >> >>This is PostModernist's log file output at the time - as you can see, it >>saw the move after 9 seconds, but this is with the benefit of the TT being >>already stuffed. >> >>I've changed the eval somewhat since, but it still likes Bf5 at about depth >>8 (11 secs). >> >>I also searched to d9 after the line you mention, and I get -2.9 for Black. >> >> >> >>Andrew Williams >> >> >>--- Pondering (GWT=0), assuming move: f6 --- >> 2= 128 0 47 Na4 Ng3 Bxg3 Bxg3 Nc5 Nd8 Rac1 >> 3= 128 0 137 Na4 Ng3 Bxg3 Bxg3 Nc5 Nd8 Rac1 >> 4= 128 0 227 Na4 Ng3 Bxg3 Bxg3 Nc5 Nd8 Rac1 >> 5= 128 0 317 Na4 Ng3 Bxg3 Bxg3 Nc5 Nd8 Rac1 >> 6= 128 0 407 Na4 Ng3 Bxg3 Bxg3 Nc5 Nd8 Rac1 >> 7= 137 3 125628 Na4 Bg3 Nb6 Rb8 Bc3 Qd6 >> 8b 158 9 431077 Bf5 exf5 Nxd5 Qb8 Nb6 Rd8 Qf7 >> 8= 177 11 527141 Bf5 exf5 Nxd5 Qd6 Qxb7 Rb8 >>--- Pondering finished --- >> >>time 68000 >>otim 64400 >> >>f7f6 >> >>PONDERING MOVE RIGHT >> 2= 150 15 719124 Bf5 exf5 Nxd5 Qd6 Qxb7 Rb8 >> 3= 151 15 719227 Bf5 exf5 Nxd5 Qd6 Qxb7 Rb8 >> 4= 152 15 720429 Bf5 exf5 Nxd5 Qd6 Qxb7 Rb8 >> 5= 152 15 722614 Bf5 exf5 Nxd5 Qd6 Qxb7 Rb8 >> 6= 155 16 733703 Bf5 exf5 Nxd5 Qd6 Qxb7 Rb8 >> 7= 150 16 766337 Bf5 exf5 Nxd5 Qd6 Qxb7 Rb8 >> 8= 156 18 864018 Bf5 exf5 Nxd5 Qd6 Qxb7 Rb8 >>NODES PER SECOND 45500 > > >Andrew, > >What eval terms do you think are most responsible for choosing Bf5? > At the time this version was playing I was worried that it was over-valuing Pawns compared to minor pieces. So I think the fact that the Queen can hoover up some Pawns is playing a part. But it's really very difficult to tell. When I get home tonight, I'll set up the position at the end of the PV and see what it thinks of the various pieces. >I notice you handle pondering differently than I. If my guess is correct, I >just continue the search at that point. You restart. Any reason to restart the >search? > No. It's a dumb idea and I can't really remember why I don't continue the search from (completeddepth+1). I'll have to have a look at that at some point. Andrew
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.