Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A test position for chess programs(importance of tablebases)

Author: Dieter Buerssner

Date: 07:06:18 01/19/01

Go up one level in this thread


On January 19, 2001 at 05:15:41, Uri Blass wrote:

[Please refer to the post I am answering for the board display]
>8/8/p1r5/6k1/KP6/8/8/5R2 b - - 0 1
>
>In this position Rebel century3 blundered and played 66...Rb6 when white
>probably forces mate in 48(The main line is losing a tempo by 67.Rf2 Kg4 68.Rf8
>Kg5 69.Rf1 when Rebel to mope has to play a move that desroys black's position I
>did not learn the position to understand exactly why but I believe the main line
>of chess programs)
>
>How much time does your program need to find 66...Rc3 or 66...Rc2 that are
>probably enough for a draw(I am not sure and I may be wrong about it)?
>
>How much time does it need with tablebases and how much time does it need
>without tablebases?

With TBs (KRPKR + all 4-men) Yace fails low for Rb6 after almost 5 minutes
on AMD K6-2 475.

  nodes    time   score  depth
  9364951  4:44.8  -0.89 18-- 1...Rb6 2.Rf8 Kg4 3.Rf7 Rg6
 12247832  5:53.1 -Mat48 18t  1...Rb6 2.Rf8 Kg4 3.Rf7 Kg5 4.Rf1 Kg6 5.Rf4 Rc6
                              6.Ka5 Kg7 7.Rd4 Kf7 8.Rd7+ Ke8 9.Ra7 Rc4
                              10.Rxa6 {EGTB} 10...Kd7! 11.Kb5! Rd4 12.Rb6!
                              Re4 13.Rb7+ Kc8! 14.Ka6! Re6+! 15.Ka7! Re3
                              16.b5! Ra3+! 17.Kb6 Rb3 18.Rf7 Rb4 19.Rf8+! Kd7
                              20.Rb8! Rc4 21.Ka7! Rd4 22.b6 Ra4+! 23.Kb7 Rb4!
                              24.Rf8 Rb3 25.Rf2 Rb4 26.Rd2+! Ke6! 27.Kc7
                              Rc4+! 28.Kb8! Rb4 29.b7! Rb3 30.Ra2! Kd6 31.Ra7
                              Rc3 32.Ka8! Rb3! {Not all positions found in
                              TB} {-80}
 14806072  7:18.4 -Mat49 18t+ 1...Rc3 2.Ra1 Kf6 3.Ka5 Rc6 4.Rh1 Ke7 5.Rh8 Kd7
                              6.Rf8
 17849754  9:28.6  -0.21 18t  1...Rc3 2.Ra1 Kf6 3.Ka5 Rc6 4.Re1 Kf7 5.Re5 Rh6
                              6.Re4 Rc6 7.Re1 Rh6 8.Re5 Rf6 9.Rc5 Ke7 10.Rc7+
                              Kd8 11.Ra7 Rf4 {0}

It also shows a very bad performance, due to the many TB accesses.
After 1...Rb6 it finds also a shorter mate:

  19213357  8:50.2  Mat44 19t  2.Rf3 Kg4 3.Rf2 Kg5 4.Rf1 Kg6 5.Rf4 Rc6 6.Ka5
                               Kg7 7.Rd4 Kf7 8.Rd7+ Ke8 9.Ra7 Rc4 10.Rb7 Rc1
                               11.Kxa6 {EGTB} 11...Kd8! 12.b5 Kc8! 13.Rf7 Rb1
                               14.Kb6! Rb2 15.Rf8+! Kd7 16.Rb8! Rc2 17.Ka7!
                               Rd2 18.b6 Ra2+! 19.Kb7 Rb2! 20.Rf8 Rb3 21.Rf2
                               Rb4 22.Rd2+! Ke6! 23.Kc7 Rc4+! 24.Kb8! Rb4
                               25.b7! Rb3 26.Ra2! Kd6 27.Ra7 Rc3 28.Ka8! Rb3!
                               {Not all positions found in TB} {80}

Because I do not have the promotion table, the mate cannot be shown.
The version I used, probes TBs during the whole search, but not in
quiescence search.

Without TBs, Yace cannot avoid Rb6 within 30 minutes (depth 23). The scores
after 1...Rb6 and after 1...Rc3 are identical for many depths, and for other
depths they differ by 1 centipawn. Yace eval does not have any sophisticated
knowledge about KRPKP.

-- Dieter



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.