Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / New bug in Rybka 2.3.1?
- - By Voodoo Master [de] Date 2007-03-06 09:42 Edited 2007-03-06 10:31
Hello Vas,

here is a position, created by W. Eigenmann (2005)

1....Kh4 (Black will win)

8/p7/8/pPp1p1k1/2PbK3/8/8/3B4 b - - 0 1


Analysis by Rybka 2.3.1 mp 64-bit on Windows XP-64 (F10-GUI):

1...Kf6 2.Bc2 Ke6 3.Bd1 Bf2 4.Bg4+ Kd6 5.Bd1 Bh4 6.Kf5 Be1 7.Ke4 Bf2 8.Bc2
  ³  (-0.67)   Tiefe: 2   00:00:00
  µ  (-0.74)   Tiefe: 37   00:00:44  14420kN, tb=3154

Further analysing had no progress after the above evaluation. I stopped evaluation after 1 hour.

The evaluation so far was 1....Kh4 (10/12)
-/+ (-0.74); Depth: 37; time: 00:00:44; 14420KN; tb=3154

It seems Rybka hangs somewhere in a loop. Please can you check this behavior?.

The correct solution (German notation) is:

1...Kh4! [1...Kf6? 2.La4 Ke6 3.Lb3 Kd6 4.La4 Kc7 5.Lb3 Kb6 6.La4 a6 7.bxa6 Kxa6 8.Lb3=] 2.Kf3 [2.Lc2 Kg3 3.Ld1 Kf2 4.Kd3 Ke1 5.La4 e4+ 6.Kxe4 Kd2-+] 2...Kh3 3.Lc2 [3.Lb3 Kh2-+; 3.Ke4 Kg3-+; 3.La4 Kh2-+] 3...Kh2 4.Le4 [4.Ld1 Kg1 5.Ke2 (5.Ke4 Kf2-+) 5...Kg2 6.Lb3 a) 6.Lc2 Kg3-+; b) 6.Kd3 Kf2 7.Lc2 (7.Ke4 Ke1 8.Lb3 Kd2-+) 7...Kf3 8.Ld1+ Kf4-+; 6...Kg3 7.Lc2 (7.La2 Kf4 8.Lb1 e4-+) 7...Kf4-+; 4.La4 Kg1 5.Ke2 Kg2-+; 4.Lb1 a4-+; 4.Ke4 Kg2-+] 4...a4 5.Ke2 [5.Lb1 a3 6.La2 Kg1-+] 5...Kg3 6.Lc2 [6.Lb1 a3 7.La2 (7.Kd3 Kf2-+) 7...e4-+] 6...a3 7.Lb1 Kf4 8.La2 e4-+ (W.Eigenmann)

Here is the result testet with Rybka 2.2n2 mp:

Neue Partie - Rybka 2.3.1 mp

Analysis by Rybka 2.2n2 mp :

1...Kf6 2.Lc2 Ke6 3.Lb3 Lb2 4.Ld1 Lc1 5.Lc2 Lf4 6.La4 Ld2 7.Lb3 Kd6 8.Ld1
  ³  (-0.52)   Tiefe: 2   00:00:00
  ³  (-0.66)   Tiefe: 23   00:00:26  7012kN
1...Kh4 2.Kf3 Kh3 3.Lc2 Kh2 4.Ld1 Kg1 5.Ke4 Kf1 6.Kd3 Kf2 7.Lc2 Kf3 8.Ld1+
  µ  (-1.03)   Tiefe: 23   00:00:39  11927kN
  µ  (-1.03)   Tiefe: 26   00:01:24  25606kN

Volker.
Parent - - By bnc (***) Date 2007-03-06 16:19
Yes this happens on my system also .. Opteron 4 core (2x275), 1gb hash, Rybka 2.3.1 mp x64, Vista Ultimate x64, 3-4-5-some 6 EGTB normal setting

After reaching depth 37 in 1 min 27 sec ...

  (-0.74)   Depth: 37   00:01:27  9910kN, tb=2950
    1...Kf6 2.Bc2 Ke6 3.Bd1 Bg1 4.Bb3 Bh2 5.Bd1 Bf4 6.Bg4+ Kd6 7.Bd1 Bg5

Rybka 2.3.1 starts to ponder on 1...Kh4 . I let it run for over 2 hours ..Still "stuck" on 1..Kh4 at depth 37

Switching back to Rybka 2.2n2 and 1..Kh4 is found in 1 min 26 sec

  (-0.69)   Depth: 21   00:01:10  761kN
    1...Kf6 2.Bc2 Ke6 3.Bb3 Bc3 4.Bd1 Be1 5.Kd3 Kd6 6.Bc2 Ke7 7.Ke2 Bc3 8.Ke3 Ke6

  (-0.86)   Depth: 21   00:01:26  3481kN
    1...Kh4 2.Kf3 Kh3 3.Bc2 Kh2 4.Bb3 Kg1 5.Ke2 e4 6.Bc2 e3 7.Bd1 Kg2 8.Bb3 Kg3
Parent - By Vasik Rajlich (Silver) [hu] Date 2007-03-08 13:33
Thanks for the position. I can confirm that the same thing happens in the single-CPU, which means that it's reproducible. Although it won't be easy to get to the bottom of it.

Vas
Up Topic Rybka Support & Discussion / Rybka Discussion / New bug in Rybka 2.3.1?

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill