Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: late move reductions

Author: Uri Blass

Date: 16:52:13 03/01/06

Go up one level in this thread


On March 01, 2006 at 15:26:46, Alessandro Scotti wrote:

>Hi Bob,
>well this is almost exactly what I'm doing in Kiwi, which is actually a
>straightforward implementation of an idea I got from Uri here on the CCC. The
>only difference I see is I don't have the "another move failed high" bit, I just
>count fail highs.
>
>        int hindex = hpiece * 64 + curr.getTo();
>
>        if( haveHistoryPruning &&
>            depthExtension == 0 &&
>            validMoves >= 4 &&
>            ! curr.isCaptureOrPromotion() &&
>            ! givesCheck &&
>            depth >= (4*FullPlyDepth) &&
>            true )
>        {
>            int n = histTable[ hindex ].count;
>            int f = histTable[ hindex ].fail_high;
>
>            if( f < (n / 8) ) {
>                depthExtension = -FullPlyDepth;
>            }
>        }

Thanks for sharing the way that you do it.

I do something different and more complicated and even I do not remember exactly
what I am doing and I also do not like to share everything that I am doing.

I think to try something more simple because I do not believe that I earn much
today from history based pruning and I believe it is possible to earn more from
it(I did not test it recently in games but it seems clear that I earn in average
less 0.5-1 ply from late move reduction based on test positions and it seems to
be constant because I almost do not prune when the remaining depth is high)

I will say only that I learned from this discussion because I never tried not to
reduce captures and promotions and because I had no rules not to do it I had
other rules to prevent missing tactics that limited my reductions significantly.

I tried some idea based on fruit code and use today combination of an idea from
fruit as I understood it and the idea that Alessandro got from me but I think to
try to see if I can earn more from late move reduction by something simple.

I believe that today I earn from late move reductions because in most cases I
find tactics at the same depth when I get the same depth 1.5-2 times faster but
of course it is no proof.

I know that in the past I earned from it but it does not prove that I earn from
it today because a lot of other things were changed in movei since that time.

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.