Author: Robert Hyatt
Date: 14:53:40 10/31/98
Go up one level in this thread
On October 31, 1998 at 16:52:12, Inmann Werner wrote: >On October 31, 1998 at 10:31:35, Robert Hyatt wrote: > > >>Robert Hyatt wrote: >> >>easy... create a "pawn structure hash 'structure'"... in it you put the >>normal hash signature (only using the locations of pawns, of course), plus >>the output of your pawn evaluation (you have to be very careful and *not* >>include *anything* that has to do with pieces at this point). Plus you >>add any things that your pawn eval sets and passes to successive evaluation >>routines... ie a map of weak pawns, number of isolated pawns on open files, >>a map of which pawns are passed, etc. Then the hash "hit" can produce >>*everything* a call to your pawn structure evaluator normally would... only >>far faster... >> >>Just don't forget that the hash signature depends only on pawns, so the stuff >>you compute there should depend only on pawns... you can add a second pawn >>evaluator (I did in crafty) if you have some analysis that depends on where >>pieces are or how many there are... > > >I hash the pawn structure including rooks-position and the king-positions. >I do not know, if this is good, but it works fine (20% faster). I do this, >because the position of the rooks and the king is often used in my pawn >evaluation. Does anybody something like me, or do you think, it is not good? > >Werner what percentage hits do you get? with just pawns I get over 99% all the time. I'd bet that with kings and rooks factored in you drop way down... which is costly...
This page took 0.01 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.