Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Is having _all_ tablebases better than having just _most_ of them?

Author: Dieter Buerssner

Date: 09:50:26 01/23/01

Go up one level in this thread


On January 23, 2001 at 11:11:56, Brian Richardson wrote:

>This [not having a full set of EGTBs]has turned out to be a fairly significant > problem for Tinker on ICC.  I
>don't have room currently for all of the 5 piece files, but do have most of the
>pawn combinations.

And then the problem can be, that the program refuses to promote.

I have tried the following. Probe the root position of the search. If this
is a TB mate, do all legal moves, and look if probing them is successful.
If not all moves yield in a TB hit (because a promotion table is missing),
never probe TBs of this many pieces in the search. So if the root position
is KRPKR and the P can promote and KQRKR is missing, I will only probe
4-men tables in the search. Some care has to be taken with transposition
tables. If EGTB hits are stored in the hash, and the hash is not cleared between
moves, I do not allow to return the EGTB scores from the hash
with 5 pieces in the above scenario.

BTW. I think 4+1 TBs are not of much help.

Regards,
Dieter




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.