Not logged inRybka Chess Community Forum
Forum RybkaChess Help Search Register Login
Up Topic Rybka Support & Discussion / Rybka Discussion / Endgame knowledge of Rybka
- - By ChessISDraw Date 2008-03-11 11:37
White to move and white wins

1.Bxd3 Kxh5 2.Kc3 Kg5 3.b6 Bxb6 4.Kxb6 leads to draw by 5-men tablebases.

1.Bd1 white win.

8/b5p1/6Pk/1P1K3P/8/3p4/4B3/8 w - - 0 56


Analysis by Rybka 2.3.2a mp (with TBs access Never):

1.Bxd3 Kxh5
  ±  (1.40)   Depth: 5   00:00:00
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8
  +-  (1.61)   Depth: 7   00:00:00  1kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8
  +-  (1.65)   Depth: 9   00:00:00  1kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8
  +-  (1.65)   Depth: 9   00:00:00  1kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4
  +-  (2.75)   Depth: 10   00:00:00  5kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8
  +-  (2.77)   Depth: 11   00:00:00  5kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8
  +-  (2.77)   Depth: 13   00:00:00  5kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8
  +-  (2.77)   Depth: 13   00:00:00  5kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bc4
  +-  (2.85)   Depth: 14   00:00:00  8kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bc4
  +-  (2.85)   Depth: 15   00:00:00  8kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bb5 Kd4
  +-  (3.07)   Depth: 16   00:00:00  11kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bb5 Kd4
  +-  (3.06)   Depth: 17   00:00:00  15kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bb5 Kd4
  +-  (3.17)   Depth: 18   00:00:00  22kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bb5 Kd4
  +-  (3.19)   Depth: 19   00:00:00  33kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bb5 Kd4
  +-  (3.21)   Depth: 20   00:00:00  49kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bb5 Kd4
  +-  (3.27)   Depth: 21   00:00:00  71kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bb5 Kd4
  +-  (3.28)   Depth: 22   00:00:00  99kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bb5 Kd4
  +-  (3.30)   Depth: 23   00:00:00  138kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bb5 Kd4
  +-  (3.32)   Depth: 24   00:00:00  193kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bb5 Kd4
  +-  (3.63)   Depth: 25   00:00:00  310kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bb5 Kd4
  +-  (3.73)   Depth: 26   00:00:00  429kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Ke3 7.Bb5 Kd4
  +-  (4.03)   Depth: 27   00:00:01  599kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Kg5 7.b8Q Bxb8
  +-  (4.17)   Depth: 28   00:00:13  9891kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Kg5 7.b8Q Bxb8
  +-  (4.17)   Depth: 29   00:00:15  12601kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Kg5 7.b8Q Bxb8
  +-  (4.18)   Depth: 30   00:00:16  13219kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Kg5 7.b8Q Bxb8
  +-  (4.18)   Depth: 31   00:00:17  13876kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Kg5 7.b8Q Bxb8
  +-  (4.18)   Depth: 32   00:00:19  15013kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Kg5 7.b8Q Bxb8
  +-  (4.18)   Depth: 33   00:00:45  29892kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf4 5.Kd7 Be5 6.Kc8 Kg5 7.b8Q Bxb8
  +-  (4.18)   Depth: 34   00:00:59  37979kN
Parent - - By Felix Kling (Gold) [de] Date 2008-03-11 12:44
finding fortresses is not possible for Rybka and the other engines...
Parent - - By BB (****) [de] Date 2008-03-11 12:51

> finding fortresses is not possible for Rybka and the other engines...


Some engines can find some fortresses.

Patzer is the first program to solve the famous Behting study (1.Kc6 draw) because it knows that it cannot win with a lonely queen. [Quote from here].

8/8/7p/3KNN1k/2p4p/8/3P2p1/8 w - - 0 1
Parent - By Felix Kling (Gold) [de] Date 2008-03-11 13:21
the question is how much of it is specific knowledge, maybe the author tried everything to get the engine solving this puzzle. Of course you could also save the position in the code and tell the engine that it is a draw. It won't solve other puzzles with fortresses though...

Btw., the idea seems to be Kc6 g1Q Nh4 followed by Nf3; Kc6 because otherwise there's a check (and with Kc6 Qh1 Nf3 we get the fortress), e.g. Kxc4 Nxh4 Qf1 followed by Kxh4 wins.
Parent - By ChessISDraw Date 2008-03-11 15:22
Lets see what Vasik thinks about this.
Parent - - By Gaмßito (****) [cr] Date 2008-03-11 23:50
Not for all the other engines. 
 
Fritz 11 can avoid Bxd3 that leads to a draw and can see that Bf3 is much better. Once Fritz searches deeply here, it already understand that the resulting position is a draw, thanks to their endgame knowledge. I hope the same for Rybka 3.  
 
I think this endgame is interesting and relatively easy to solve for a human, even for a weak player. Note that without the h5 white pawn, it is impossible for white to win here. The only way to win here is with trying to promote some pawn. And white can do it, by means of h6 gxh6 and g7 and white wins.  
 
After 1. Bxd3? Kxh6 and the pawn disappear. Now the black king will hide in g8 and h8, not allowing white to make any progress.
 
1. Bxd3? Kxh5 2. Kc6 Bb8 3. b6 Kg5 4. b7 Kf6 5. Kd7 Be5 6. Kc8 Ke7 7. b8=Q Bxb8 8. Kxb8 Kf8 with a draw, but many programs display here a big white advantage due the lack of knowledge.   
 
1. Bf3 Kg5 2. Ke4 d2 3. Kd3 d1=Q 4. Bxd1 Kf6 5. Ke4 Ke7 6. Kd5 Kf8 7. Kc6 Bb8 8. b6 Be5 9. b7 Bf4 10. Kd7 Bb8 11. Kc8 Bf4 12. b8=Q Bxb8 13. Kxb8 Kg8 14. Kc8 Kf8 15. Kd8 Kg8 16. Ke8 Kh8 17. h6 gxh6 18. Kf7 h5 19. g7 1-0  

8/b5p1/6Pk/1P1K3P/8/3p4/4B3/8 w - - 0 1


Analysis by Fritz 11:

1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf6 5.Bc4 Ke5
  +/-  (1.38)   Depth: 9/13   00:00:00  6kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf6 5.Be2 Ke5 6.Bc4
  +/-  (1.16)   Depth: 10/14   00:00:00  8kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf6 5.Kd7 Be5 6.Be4 Bf4
  +-  (1.52)   Depth: 11/16   00:00:00  12kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf6 5.Kd7 Be5 6.Be4 Bf4 7.Kc8
  +-  (1.47)   Depth: 12/17   00:00:00  18kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf6 5.Kd7 Be5 6.Be4 Bf4 7.Bd5 Kf5
  +-  (1.49)   Depth: 13/19   00:00:00  27kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf6 5.Kd7 Be5 6.Be4 Bf4 7.Bd5 Kf5 8.Kc8
  +-  (1.44)   Depth: 14/18   00:00:00  42kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.b6 Bb8 4.b7 Kf6 5.Kd7 Be5 6.Be4 Bf4 7.Kc8 Ke7 8.Bd5 Be5
  +/-  (1.37)   Depth: 15/22   00:00:00  64kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.Kd7 Bf2 4.Kc7 Kf6 5.Kb7 Ke7 6.Kc6 Be3 7.Kd5 Kf6 8.Bc4 Kxg6 9.Bd3+ Kf6 10.Bc4 Bf2 11.Kc6 Ke5 12.Bd3 g5
  =  (0.13)   Depth: 16/25   00:00:00  365kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.Kd7 Bf2 4.Kc7 Kf6 5.Kc6 Ke7 6.Be4 Kf6 7.Bd3
  =  (0.13)   Depth: 17/22   00:00:00  589kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.Kd7 Bf2 4.Kc7 Kf6 5.Kc6 Ke7 6.Be4 Kf6 7.Bd3
  =  (0.13)   Depth: 18/24   00:00:00  795kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.Kd7 Bf2 4.Kc7 Kf6 5.Kc6 Ke7 6.Be4 Kf6 7.Kb7 Ke7 8.Bd3 Bd4 9.Kc6 Be3 10.Be4 Kf6 11.Bc2 Ke7 12.Bb3
  =  (0.13)   Depth: 19/24   00:00:01  1164kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.Kd7 Bf2 4.Kc7 Kf6 5.Kc6 Ke7 6.Be4 Kf6 7.Bd3
  =  (0.13)   Depth: 20/26   00:00:01  1592kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.Kd7 Bf2 4.Kc7 Kf6 5.Kc6 Ke7 6.Be4 Kf6 7.Kc7 Ke7 8.Kc6
  =  (0.13)   Depth: 21/26   00:00:02  2196kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.Kd7 Bf2 4.Kc7 Kf6 5.Kc6 Ke7 6.Be4 Kf6 7.Bc2 Ke7 8.Kd5 Be3 9.Be4 Kf6 10.Kc6 Kg5 11.Bc2 Kf6 12.Kd5 Ke7
  =  (0.13)   Depth: 22/26   00:00:03  2994kN
1.Bxd3 Kxh5 2.Kc6 Kg5 3.Kd7 Bf2 4.Kc7 Kf6 5.Kc6 Ke7 6.Be4 Kf6 7.Bc2 Ke7 8.Be4
  =  (0.13)   Depth: 23/28   00:00:04  4148kN
1.Bf3 d2 2.Kc6 Kg5 3.b6 Bb8 4.Kd5 Kf4 5.Be2 Kg5 6.Ke4 Bg3 7.b7 Bb8 8.Kd3 Bf4 9.Ke4 Bb8
  +/=  (0.41)   Depth: 23/41   00:00:22  22791kN
1.Bf3 d2 2.Kc6 Kg5 3.b6 Bb8 4.Kd5 Kf4 5.Be2 Kg5 6.Ke4 Bg3 7.b7 Bb8 8.Kd3 Bf4 9.Ke4 Bb8
  +/=  (0.41)   Depth: 24/39   00:00:33  32526kN
1.Bf3 d2 2.Kc6 Kg5 3.b6 Bb8 4.Kb7 Be5 5.Ka8 Kf4 6.Bd1 Kg5 7.b7 Kf6 8.b8Q Bxb8 9.Kxb8 Ke6 10.Kc7 Ke7 11.Kc6 Ke6 12.Kc5 Ke5 13.Kc4 Ke4 14.Kc3
  +-  (1.70)   Depth: 25/45   00:03:26  205mN
1.Bf3 Kg5 2.Ke4 Bf2 3.Kxd3 Kf6 4.Ke4 Ke7 5.Kd5 Be3 6.Kc6 Kd8 7.Bg4 Bf4 8.b6 Ke7 9.b7 Bg3 10.Be2 Bb8 11.Bc4 Kf6 12.Kd7 Ba7 13.Be2 Ke5 14.h6 gxh6 15.g7
  +-  (3.89)   Depth: 26/45   00:06:59  435mN
1.Bf3 Kg5 2.Ke4 Bf2 3.Kxd3 Kf6 4.Ke4 Ke7 5.Kd5 Be3 6.Kc6 Kd8 7.Bg4 Bf4 8.b6 Ke7 9.b7 Bg3 10.Be2 Bb8 11.Bc4 Kf6 12.Kd7 Ba7 13.Be2 Ke5 14.h6 gxh6 15.g7
  +-  (3.89)   Depth: 27/41   00:07:19  456mN
1.Bf3 Kg5 2.Ke4 Bf2 3.Kxd3 Kf6 4.Ke4 Ke7 5.Kd5 Be3 6.Kc6 Kd8 7.Bg4 Bf4 8.b6 Ke7 9.b7 Be5 10.Be2 Bf4 11.Bg4
  +-  (3.89)   Depth: 28/43   00:08:27  527mN
1.Bf3 Kg5 2.Ke4 Bf2 3.Kxd3 Kf6 4.Ke4 Ke7 5.Kd5 Be3 6.Kc6 Kd8 7.Bg4 Bf4 8.b6 Ke7 9.b7 Be5 10.Be2 Bf4 11.Kd5 Kf6 12.Kc6 Ke7
  +-  (3.89)   Depth: 29/47   00:10:28  657mN
1.Bf3 Kg5 2.Ke4 Bf2 3.Kxd3 Kf6 4.Ke4 Ke7 5.Kd5 Be3 6.Kc6 Kd8 7.Bg4 Bf4 8.b6 Ke8 9.b7 Ke7 10.Be2 Bg3 11.Kd5 Kf6 12.Kc6 Ke7
  +-  (4.05)   Depth: 30/49   00:25:59  1711mN
1.Bf3 Kg5 2.Ke4 Bf2 3.Kxd3 Kf6 4.Ke4 Ke7 5.Kd5 Be3 6.Kc6 Kd8 7.Bg4 Bf4 8.b6 Ke8 9.b7 Ke7 10.Be2 Be5 11.Bc4 Kf6 12.Be2 Ke7
  +-  (4.05)   Depth: 31/48   00:31:14  2070mN
1.Bf3 Kg5 2.Ke4 Bf2 3.Kxd3 Kf6 4.Ke4 Ke7 5.Kd5 Be3 6.Kc6 Kd8 7.Bg4 Bf4 8.b6 Ke8 9.b7 Ke7 10.Be2 Bg3 11.Kd5 Kf6 12.Kc4 Ke7 13.Kd5
  +-  (4.05)   Depth: 32/51   00:39:34  2631mN
1.Bf3 Kg5 2.Ke4 Bf2 3.Kxd3 Kf6 4.Ke4 Ke7 5.Kd5 Be3 6.Kc6 Kd8 7.Bg4 Bf4 8.b6 Ke8 9.b7 Ke7 10.Be2 Bg3 11.Bf3 Bf4 12.Bg4 Be5 13.Kd5 Bf4 14.Kc6
  +-  (4.05)   Depth: 33/53   01:02:30  4196mN

(, AMD 11.03.2008)

Regards,
Gaмßito.
Parent - - By Felix Kling (Gold) [de] Date 2008-03-12 12:07
That's just the build in knowledge i talked about, somehow like built in tablebases. I think it's more cleaner to seperate those things, so downloading tablebases is the way to go at the moment. I guess Rybka 3 will include all 3-4-5 men tablebases on DVD.
Parent - - By InspectorGadget (****) [za] Date 2008-03-12 14:23
"I guess Rybka 3 will include all 3-4-5 men tablebases on DVD"

All? Is that possible?
Parent - - By Uly (Gold) [mx] Date 2008-03-12 14:36
It would be 8 GB, so maybe a two sides DVD. But I wonder if the 5 ELO points are worth the hassle.
Parent - - By h1a8 (***) [us] Date 2008-03-12 21:48 Edited 2008-03-12 22:09
They have dual layer discs out. They hold more than 9GB per side. So it is possible to have more than 18GB of tablebases on 1 DVD.
Also it is false to say tablebases only provide rybka with 5elo. The true estimate is 30elo.
Just think about it. There are many 5 piece and some 4 piece endgames Rybka do not know how to win.
So even if Rybka only uses the tablebases when 5 pieces or less are on the board then it would provide significant strength.
That is to say Rybka doesn't hit for the tablebases before 5 men are on board so it doesn't waste resources hitting for them.
But even if it did hit for them the we easily get 30elo improvement.
Parent - By Permanent Brain (*****) Date 2008-03-13 07:56
The problem with +5 versus +30 may be, that it depends on how this is tested. Or in other words, which 'environment' the improvement refers to, because: In computer chess, the opponent will often resign a lost late endgame, not forcing Rybka to actually demonstrate the win.

While I cannot acknowledge nor disprove the exact figures, I think it is logical that the gain an engine with some endgame weaknesses gets from tablebases, is bigger against opponents which do never resign, and smaller against opponents which resign 'anyway', obscuring parts of the problem.

(This could of course be the same opponents each, and resign or not is a configuration option.)
Parent - - By Felix Kling (Gold) [de] Date 2008-03-12 17:21
aha, you may be right that it could become critical, so 3-4 men only maybe, in the end it will be a question if enough space is left on the DVD. I don't know much about DVD etc. , sorry :) I thought they would have a bigger capcity
Parent - By Permanent Brain (*****) Date 2008-03-12 19:03 Edited 2008-03-12 19:11
It should be possible to find a reasonable subset of all 5-piece tablebases which fits on a DVD. For example, all 4 vs 1 tables are not necessary. I only ask for the precision to include all tables which can arise from all underpromotions, from endings with pawns which are included...

Typical useful 5-piece endings would be RP-R, RB-R, RN-R, QP-Q, NN-P.
Parent - - By Dadi Jonsson (Silver) [is] Date 2008-03-13 12:47
Convekta has been selling a complete collection of 3-4-5 tablebases on a single DVD for many years.

You can see it at the bottom of this page:

http://store.convekta.com/shop_model.asp?gid=121&sView=Catalog
Parent - By Felix Kling (Gold) [de] Date 2008-03-13 13:42
yes, the only question is if there is enough space left if the GUI, Rybka and some databases/opening books are already on the DVD.
Parent - - By George Tsavdaris (****) Date 2008-03-11 15:40
Why you didn't use tablebases? When a tool is available you should use it. :-)
It's like having an umbrella and when it rains you don't want to use it because you want to test your own strength at rain conditions. :-)

Analysis by Rybka 1.2i 32-bit:

1.Be2xd3 Kh6xh5 2.Kd5-c6 Kh5-g5 3.Bd3-b1 Ba7-g1 4.Bb1-c2 Bg1-d4
  +/=  (0.48)   Depth: 14   00:00:01  162kN, tb=354
1.Be2-d1 d3-d2 2.Kd5-c6 Ba7-b8 3.b5-b6 Kh6-g5 4.b6-b7 Bb8-a7 5.Kc6-d5 Kg5-f5 6.Kd5-c4 Ba7-b8 7.Kc4-d4 Bb8-e5+
  +/=  (0.66)   Depth: 14   00:00:01  207kN, tb=373
1.Be2-d1 d3-d2 2.Kd5-c6 Ba7-b8 3.b5-b6 Kh6-g5 4.b6-b7 Kg5-f6 5.Kc6-d5 Kf6-f5 6.Kd5-d4 Bb8-e5+ 7.Kd4-d3 Be5-f4
  +/=  (0.66)   Depth: 15   00:00:01  237kN, tb=381
1.Be2-d1 d3-d2 2.Kd5-c6 Ba7-b8 3.b5-b6 Kh6-g5 4.b6-b7 Kg5-f6 5.Kc6-d5 Kf6-f5 6.Kd5-d4 Bb8-f4 7.Kd4-d3 Kf5-g5
  +/=  (0.66)   Depth: 16   00:00:01  272kN, tb=419
1.Be2-d1 d3-d2 2.Kd5-c6 Ba7-b8 3.b5-b6 Kh6-g5 4.b6-b7 Kg5-f6 5.Kc6-d5 Kf6-f5 6.Kd5-d4 Bb8-f4 7.Kd4-d3 Kf5-g5
  +/=  (0.68)   Depth: 17   00:00:01  335kN, tb=507
1.Be2-d1 d3-d2 2.Kd5-c6 Ba7-b8 3.b5-b6 Kh6-g5 4.b6-b7 Kg5-f6 5.Kc6-d5 Kf6-f5 6.Kd5-d4 Bb8-f4 7.Kd4-d3 Kf5-g5
  +/=  (0.67)   Depth: 18   00:00:02  430kN, tb=563
1.Be2-d1 d3-d2 2.Kd5-c6 Ba7-b8 3.b5-b6 Kh6-g5 4.b6-b7 Kg5-f6 5.Kc6-d5 Kf6-f5 6.Kd5-d4 Bb8-f4 7.Kd4-d3 Kf5-g5
  +/=  (0.67)   Depth: 19   00:00:02  536kN, tb=688
1.Be2-d1 Ba7-f2 2.Kd5-c4 d3-d2 3.Kc4-d3 Bf2-a7 4.Kd3xd2 Kh6-g5 5.Kd2-d3 Kg5-f5 6.Bd1-g4+ Kf5-g5 7.Kd3-e4 Ba7-f2
  +/-  (1.35)   Depth: 20   00:00:10  2146kN, tb=2116
1.Be2-d1 Ba7-f2 2.Kd5-c6 Kh6-g5 3.b5-b6 Bf2-g3 4.b6-b7 Kg5-f5 5.Kc6-d7 Kf5-g5 6.Kd7-c8 Kg5-f6 7.b7-b8Q Bg3xb8
  +-  (3.58)   Depth: 21   00:00:16  3155kN, tb=3240
1.Be2-d1 Ba7-f2 2.Kd5-c6 Kh6-g5 3.b5-b6 Bf2-g3 4.b6-b7 Kg5-f5 5.Kc6-d7 Kf5-g5 6.Kd7-c8 Kg5-f6 7.b7-b8Q Bg3xb8
  +-  (3.58)   Depth: 22   00:00:17  3206kN, tb=3486
1.Be2-d1 Ba7-f2 2.Kd5-c6 Kh6-g5 3.b5-b6 Bf2-g3 4.b6-b7 Kg5-f5 5.Kc6-d7 Kf5-g5 6.Kd7-c8 Kg5-f6 7.b7-b8Q Bg3xb8
  +-  (3.58)   Depth: 23   00:00:18  3308kN, tb=3910
1.Be2-d1 Ba7-f2 2.Kd5-c6 Kh6-g5 3.b5-b6 Bf2-g3 4.b6-b7 Kg5-f5 5.Kc6-d7 Kf5-g5 6.Kd7-c8 Kg5-f6 7.b7-b8Q Bg3xb8
  +-  (3.58)   Depth: 24   00:00:19  3454kN, tb=4479
1.Be2-d1 Ba7-f2 2.Kd5-c6 Kh6-g5 3.b5-b6 Bf2-g3 4.b6-b7 Kg5-f5 5.Kc6-d7 Kf5-g5 6.Kd7-c8 Kg5-f6 7.b7-b8Q Bg3xb8
  +-  (3.58)   Depth: 25   00:00:22  3673kN, tb=5431
1.Be2-d1 Ba7-f2 2.Kd5-c6 Kh6-g5 3.b5-b6 Bf2-g3 4.b6-b7 Kg5-f5 5.Kc6-d7 Kf5-g5 6.Kd7-c8 Kg5-f6 7.b7-b8Q Bg3xb8
  +-  (3.58)   Depth: 26   00:00:24  3928kN, tb=6464
1.Be2-d1 Ba7-f2 2.Kd5-c6 Kh6-g5 3.b5-b6 Bf2-g3 4.b6-b7 Kg5-f5 5.Kc6-d7 Kf5-g5 6.Kd7-c8 Kg5-f6 7.b7-b8Q Bg3xb8
  +-  (3.77)   Depth: 27   00:00:35  4566kN, tb=9695
1.Be2-d1 Ba7-f2 2.Kd5-c6 Kh6-g5 3.b5-b6 Bf2-g3 4.b6-b7 Kg5-f5 5.Kc6-d7 Kf5-g5 6.Kd7-c8 Kg5-f6 7.b7-b8Q Bg3xb8
  +-  (3.77)   Depth: 28   00:00:42  5003kN, tb=12469
1.Be2-d1 Ba7-f2 2.Kd5-c6 Kh6-g5 3.b5-b6 Bf2-g3 4.b6-b7 Kg5-f5 5.Kc6-d7 Kf5-g5 6.Kd7-c8 Kg5-f6 7.b7-b8Q Bg3xb8
  +-  (11.80)   Depth: 29   00:02:28  14920kN, tb=56072
1.Be2-d1 Ba7-f2 2.Kd5-c6 Kh6-g5 3.b5-b6 Bf2-g3 4.b6-b7 Kg5-f5 5.Kc6-d7 Kf5-g5 6.Kd7-c8 Kg5-f6 7.b7-b8Q Bg3xb8
  +-  (11.80)   Depth: 30   00:02:45  15951kN, tb=63081
Parent - - By ChessISDraw Date 2008-03-12 11:23 Edited 2008-03-12 11:26
Yes sure but I cannot use a 1.2i version with a Q6600 computer and
here is how Rybka 2.3.2a mp analysis using tablebases accessing Normally with full 5-men and some of 6-men.

Analysis by Rybka 2.3.2a mp :

56.Bxd3 Kxh5
  ±  (1.40)   Depth: 5   00:00:00
56.Bxd3 Kxh5 57.Kc6
  ±  (1.22)   Depth: 6   00:00:00  1kN
56.Bxd3 Kxh5 57.Kc6 Bb8
  ±  (1.23)   Depth: 7   00:00:00  1kN
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6
  +-  (1.61)   Depth: 8   00:00:00  2kN
56.Bxd3 Kxh5 57.Kc6 Kg5 58.Kc7 Bg1
  ±  (1.23)   Depth: 9   00:00:00  5kN, tb=1
56.Bxd3 Kxh5 57.Kc6 Kg5 58.Kc7 Bd4 59.b6
  +-  (2.75)   Depth: 10   00:00:00  15kN, tb=3
56.Bxd3 Kxh5 57.Kc6 Kg5 58.Kc7 Bd4 59.b6 Kf4
  +-  (2.75)   Depth: 11   00:00:00  17kN, tb=4
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3
  +-  (2.75)   Depth: 12   00:00:00  19kN, tb=8
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2
  ±  (1.17)   Depth: 13   00:00:00  38kN, tb=14
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (2.86)   Depth: 14   00:00:00  47kN, tb=16
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (2.86)   Depth: 16   00:00:00  47kN, tb=16
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (2.86)   Depth: 16   00:00:00  47kN, tb=16
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (3.00)   Depth: 17   00:00:00  51kN, tb=19
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (3.00)   Depth: 18   00:00:00  52kN, tb=21
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (2.90)   Depth: 19   00:00:01  116kN, tb=89
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (3.62)   Depth: 21   00:00:01  131kN, tb=103
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (3.62)   Depth: 21   00:00:01  131kN, tb=103
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (3.62)   Depth: 22   00:00:02  131kN, tb=103
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (3.62)   Depth: 23   00:00:02  131kN, tb=103
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (3.62)   Depth: 24   00:00:02  131kN, tb=103
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (3.62)   Depth: 25   00:00:02  131kN, tb=103
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ke3 61.Kc8 Bd6 62.Bc2 Kd2
  +-  (3.62)   Depth: 26   00:00:02  131kN, tb=103
56.Bxd3 Kxh5 57.Kc6 Bb8 58.b6 Kg4 59.b7 Kf4 60.Kd7 Ba7 61.Kc8 Kg5 62.b8Q Bxb8
  +-  (2.99)   Depth: 27   00:00:44  2272kN, tb=3391
56.Bd1 d2 57.Kc6 Bb8 58.b6 Bg3 59.Kd7 Kg5 60.b7 Kh6 61.Kc8 Kg5 62.b8Q Bxb8
  +-  (4.79)   Depth: 27   00:00:48  2541kN, tb=3817

(, Home 12.03.2008)

I cannot get 48 seconds to solve this in a 3min blitz.
Parent - By Eduard (***) [de] Date 2008-03-12 11:49 Edited 2008-03-12 11:51
That is much better:

[Event "Wertungspartie, 5m + 0s"]
[Site "Maschinenraum"]
[Date "2008.03.11"]
[Round "?"]
[White "Nvv777Fritz, Rybka 2.3.2a mp"]
[Black "Wolkenlos, Toga II 3.1.3SE MP"]
[Result "0-1"]
[ECO "B22"]
[WhiteElo "2564"]
[BlackElo "2551"]
[Annotator "-0.07;0.11"]
[PlyCount "96"]
[EventDate "2008.03.11"]
[TimeControl "300"]

{Rybka 2.3.2a mp: 17.3 ply; 211kN/s Intel(R) Core(TM)2 CPU          6320  @ 1.
86GHz 1867MHz, (2 threads), MyBig.ctg, à_[} 1. e4 {B/0 0} c5 {B/0 0} 2. c3 {
B/0 0} Nf6 {B/0 0} 3. e5 {B/0 0} Nd5 {B/0 0} 4. d4 {B/0 0} cxd4 {B/0 0} 5. Qxd4
{B/0 0} e6 {B/0 0} 6. Nf3 {B/0 0} Nc6 {B/0 0} 7. Qe4 {B/0 0} d6 {B/0 0} 8. Nbd2
{B/0 0} dxe5 {B/0 0} 9. Nxe5 {B/0 0} Nxe5 {B/0 0} 10. Qxe5 {B/0 0} Qd6 {B/0 0}
11. Bb5+ {B/0 0} Bd7 {B/0 0} 12. Bxd7+ {B/0 0} Qxd7 {B/0 0} 13. O-O {B/0 0} Qc7
{B/0 0} 14. Qe2 {B/0 0} Bd6 {B/0 0} 15. Nf3 {B/0 0} Rc8 {B/0 0} 16. g3 {B/0 0}
O-O {B/0 0} 17. Nd4 {B/0 0} Qc5 {B/0 0} 18. Be3 {B/0 0} Nxe3 {B/0 0} 19. Qxe3 {
B/0 0} Rfd8 {B/0 0} 20. Rfd1 {B/0 0} Be7 {B/0 0} 21. Rd3 {-0.07/13 9} Qb6 {
(Ôd5) B/0 0} 22. b3 {-0.03/14 7} Rd6 {(Ëc5) B/0 0} 23. Rad1 {0.00/13 2} Rcd8 {
B/0 0} 24. Qe4 {0.00/14 7} g6 {B/0 0} 25. Nf3 {0.00/14 6} Rxd3 {B/0 0} 26. Rxd3
{0.00/15 2} Rxd3 {B/0 0} 27. Qxd3 {0.00/16 4} Qd6 {(Ôa5) B/0 0} 28. Qe4 {
0.01/15 6} Qd5 {(Ôa6) B/0 0} 29. Qxd5 {0.07/16 2} exd5 {B/0 0} 30. Nd4 {
0.04/18 7} Bc5 {B/0 0} 31. Nc2 {0.05/17 1} b5 {(Kðg7) 0.11/17 10} 32. Kf1 {
0.08/15 2} Kf8 {(f5) 0.10/19 8} 33. Nb4 {(Ke2) 0.37/18 6} Bxb4 {-0.18/17 8} 34.
cxb4 {0.27/21 3} Ke7 {-0.12/19 0} 35. f4 {(Ke2) 0.00/20 23} Kf6 {
(Kðd6) -0.37/18 15} 36. Ke2 {-0.19/19 6} g5 {(Kðf5) -0.52/19 1} 37. Ke3 {
0.15/19 13} gxf4+ {-0.41/19 8} 38. Kxf4 {-2.10/24 28} h5 {-1.00/22 0} 39. h3 {
-2.10/23 4} Ke6 {-1.19/21 5} 40. g4 {(a3) -2.42/23 2} hxg4 {-1.51/19 7} 41.
hxg4 {-5.08/25 20} Kf6 {-1.54/23 0} 42. a3 {-5.35/23 24} d4 {-1.98/21 0} 43. a4
{-5.67/21 18} d3 {-6.89/21 0} 44. Ke3 {-5.77/20 9} Kg5 {-7.74/24 0} 45. Kxd3 {
-5.72/19 2} Kxg4 {-7.94/21 3} 46. Ke3 {-6.57/19 8} a6 {-7.94/21 0} 47. axb5 {
(Kf2) -8.10/15 10} axb5 {-8.40/13 9} 48. Kf2 {(Kd3) -10.84/15 0} f6 {
(Kðf4) Nvv777Fritz,Rybka 2.3.2a mp ñäàåòñÿ -8.87/14 9} 0-1

Nvv777Fritz,R - Wolkenlos,T, Wertungspartie, 5m + 0s 2008

5k2/p4p1p/6p1/1pbp4/8/1PP3P1/P1N2P1P/5K2 w - - 0 1
  

Rybka 2.3.2a MP moves here 33. Nb4?? and lost of course! 

Rybka on my Core 2 Duo 2.66 GHz, from the PGN Notation, with klick on black move 32:

Analysis by Rybka 2.3.2 mp 32-bit :

33.Sb4 Lxb4 34.cxb4 Ke7 35.Ke2 Ke6 36.Kd3 g5 37.Kd4 g4 38.a4 Kd6 39.Ke3 a6
  =  (0.16)   Tiefe: 22   00:00:28  7273kN
33.Sb4 Lxb4 34.cxb4 Ke7 35.Ke2 Ke6 36.Kd3 g5 37.Kd4 g4 38.Kc5 f6 39.Kxb5 Kd6
  =  (0.00)   Tiefe: 23   00:01:03  16100kN
33.Ke2 Ke7 34.f4 h5 35.h3 Lg1 36.Sb4 Kd6 37.Sd3 Ke6 38.Se5 Lh2 39.Sc6 Lxg3
  =  (0.08)   Tiefe: 23   00:03:41  50066kN

(Nemeth, Privat 11.03.2008)
Up Topic Rybka Support & Discussion / Rybka Discussion / Endgame knowledge of Rybka

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill