Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: NPS

Author: Robert Hyatt

Date: 20:31:35 11/16/97

Go up one level in this thread


On November 16, 1997 at 21:49:53, David Blackman wrote:

>On November 14, 1997 at 08:16:34, Robert Hyatt wrote:
>
>
>[ much deleted ]
>>  It only means as the circumstances change, I begin to
>>be less "lazy".  This is why, on a P6/200, I search 60K in the opening,
>>80-100K in the middlegame, and 120K in the endgame.  Because I have to
>>get less lazy as the game progresses because positional scores can get
>>bigger and bigger...
>>
>
>
>This is odd. I would have thought being lazy would give you more NPS.
>That's the point isn't it? So why should the NPS increase as the
>laziness decreases?

NPS should increase dramatically as pieces come off, because the eval
should get faster and faster as should the move generator.  In the case
of Crafty, the move generator speeds up by a factor of 2-3 at least as
pieces are traded, but the eval gets slower as it has to do more work as
passed pawns are created.  So while the NPS does go up in the endgame,
it
could go *way* up.  IE Ferret speeds up by a factor of 3 or so in
endings,
while crafty is only 20% faster than in the middlegame...  there are
other
reasons ferret is faster, but this explains how it can go a little
faster
even though I said the eval is slower...  It simply ought to go a *lot*
faster than it does...



>
>My guess for the reasons:
>
>In the opening there are not many tactics that win material, so although
>the positional score range is small, the lazy eval can't kick in very
>often, since the material balance nearly always comes in equal.
>
>In the midgame a lot of lines can use lazy eval because they are either
>way ahead, or way behind materially, because of the tactics.
>
>In the endgame, i'm not sure. Endgames do have a lot of tactics,
>sometimes even more than the midgame. But maybe that wouldn't be enough
>to make up for the really huge positional factors you have to allow.
>Could it be that Crafty can
>do the endgame postitional eval really fast, compared to a midgame
>positional eval?

not as fast as I'd like, because as the endgame gets closer, there are
fewer
pieces to eval for, so that speeds up...  but the other parts of the
eval
slow down as crafty starts considering special purpose things that all
center
around passed pawns...



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.