Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: an evaluation problem of chess programs

Author: blass uri

Date: 01:59:55 10/04/98

Go up one level in this thread



On October 04, 1998 at 04:56:09, blass uri wrote:

>
>On October 03, 1998 at 22:07:23, Robert Hyatt wrote:
>
>>On October 03, 1998 at 19:09:59, Roberto Waldteufel wrote:
>>
>>>
>>>Here's another trick. I don't know if anybody else has tried this, but it might
>>>help improve efficiency when there are large numbers of tablebase hits in the
>>>search. Normally the tablebase would store an evaluation like draw or +mate in N
>>>or -mate in N. Now the search only needs to know win/draw/loss except at the
>>>root, where the best move is also needed. We can store win/draw/loss in 2 bits
>>>per position, which is less than the space to store mate in n type scores. By
>>>making the tablebase very compact, we may be able to store it in RAM, increasing
>>>speed by avoiding many disk accesses. When we actually reach a tablebase
>>>position at the root, the 2 bits of information are insufficient, so we store a
>>>second, larger tablebase on disk with a move for each position, and just look up
>>>the move at the root (we could also store number of moves to mate, but that is
>>>not necessary). Does this make sense?
>>>
>>>Best wishes,
>>>Roberto
>>
>>
>>It really wouldn't help, because this would reduce the size of the current
>>tablebases by 1/4, because they use 8 bit values.  And at 2+ gigs just
>>for KRPKR + promotions, such a savings wouldn't help as the resulting
>>databases would still be over 512mb.
>>
>>and the values would be very difficult to use
>
>
>I think that not storing the number of moves to mate  can help
>to store in hard disk the tablebases that may be too large to store in the
>harddisk (even if today there is no need to do it tommorow there will be more
>tablebases for 5 or 6 pieces)
>
>against humans you can replace the -mate in N by -1 if N is large enough
>(if the opponent needs more than 30 moves without captures and moving a pawn to
>win)
>because there is a good chance that the opponent will not find the right move
>
>you can do it also against many computer programs that do not use tablebases
>
>you need only 2 bits to represent the score from the hash tables

I mean score from the tablebases
of course you need  bits for the right move in the position but you do not need
bits for the number of moves that you do a checkmate.

 because there
>are only 4 cases:
>1)win
>2)draw
>3)losing slow(evaluation -1)
>4)losing fast
>
>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.