Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / (Game 13 14 or 15) Rybka Cluster vs Highendman = 0-1
1 2 3 Previous Next  
- - By Gaмßito (****) [cr] Date 2009-07-11 20:07
[Event "60m + 1s, unrated"]
[Site "?"]
[Date "2009.07.11"]
[Round "?"]
[White "Rechenschieber, Rybka 3g55 cluster A"]
[Black "Highendman"]
[Result "0-1"]
[ECO "B90"]
[Annotator ",Gambito"]
[PlyCount "105"]
[EventDate "2009.07.11"]
[EventType "rapid"]
[TimeControl "3600+1"]

1. e4 {  0.05/1 0  } c5 {  5  } 2. Nf3 {  0.05/1 0  } d6 {  2  } 3. d4 {
0.05/1 0  } cxd4 {  1  } 4. Nxd4 {  0.05/1 0  } Nf6 {  2  } 5. Nc3 {  0.05/1 0
} a6 {  14  } 6. Be3 {  0.05/1 0  } e5 {  7  } 7. Nb3 {  0.05/1 0  } Be6 {  9
} 8. f3 {  0.05/1 0  } h5 {  44  } 9. Qd2 {  0.05/1 0  } Nbd7 {  26  } 10.
Nd5 {  0.05/1 0  } Bxd5 {  27  } 11. exd5 {  0.05/1 0  } g6 {  29  } 12. Be2 {
0.05/1 0  } Bg7 {  23  } 13. h3 {  0.35/19 125  } h4 {  23 (Qc7)  } 14. O-O {
0.41/20 149  } O-O {  3  } 15. Rfe1 {  0.41/19 131  } Rc8 {  10 (Qc7)  } 16.
Rac1 {  0.41/19 132  } Nh5 {  38 (e4)  } 17. Na5 {  0.38/  19 100  } Qc7 {  79
(Rb8)  } 18. c4 {  0.20/20 104  } Ng3 {  8  } 19. c5 {  0.17/20 119  } dxc5 {
46  } 20. b4 {  0.18/19 5  } Qb6 {  121 (Nf5)  } 21. bxc5 {  0.10/19 59  }
Rxc5 {  38  } 22. Nc4 {  0.11/19 22  } Qd8 {  14 (Qa7)  } 23. Bxc5 {  -0.12/18
117  } Nxc5 {  8  } 24. Ne3 {  -0.25/19  121  } Ne6 {  10 (Qb6)  } 25. Bd3 {
0.00/17 83  } Nf4 {  53  } 26. Kh2 {  -0.11/18 41  } Qd6 {  139 (Qb8)  } 27.
Be4 {  0.00/19 70  } Rd8 {  52  } 28. Nc4 {  0.00/19 17  } Qf6 {  334 (Qf8)  }
29. Na5 {  0.00/18 63  } Rd7 {  37  } 30. Rb1 {  -0.25/18 71  } Qg5 {  130  }
31. Nxb7 {  -0.53/  18 0  } f5 {  9  } 32. Nc5 {  -0.59/17 19  } fxe4 {  39  }
33. fxe4 {  -0.85/17 32  } Rd8 {  19  } 34. Nxa6 {  -0.83/18 57  } Bf8 {  3  }
35. Nb4 {  -1.46/18 106  } Rb8 {  39 (Bd6)  } 36. a3 {  -0.99/17 18  } Bc5 {
23 (Ra8)  } 37. Rb3 {  -1.32/17 22  } Kg7 {  32 (Nxh3)  } 38. Rf3 {  -2.
87/19 37  } Rxb4 {  33  } 39. axb4 {  -2.95/19 0  } Bxb4 {  14  } 40. Qf2 {
-3.26/19 0  } Bxe1 {  23  } 41. Qxe1 {  -3.38/19 0  } Nxe4 {  39  } 42. Qg1 {
-3.84/20 0  } Kh6 {  27  } 43. d6 {  -3.  84/20 0  } Nxd6 {  12  } 44. g4 {
-3.47/20 6  } hxg3+ {  196  } 45. Rxg3 {  -4.63/19 38  } Qd8 {  37  } 46. Rb3
{  -4.87/18 0  } Nf5 {  45 (Ne4)  } 47. Qf2 {  -5.41/15 26  } Qd1 {  64  } 48.
Ra3 {  -5.55/14 0  } Qd6 {  39  } 49. Qa2 {  -6.76/14 0  } e4 {  27  } 50. Ra6
{  -10.80/14 0  } Qd3 {  32  } 51. Ra3 {  -14.79/14 0  } Qd4 {  9  } 52. Ra8 {
-6.09/12 11  } Nh4 {  26 (e3)  } 53. Ra3 {  288.06/11 37 Rechenschieber,
Rybka 3g55 cluster A resigns (Lag: Av=0.52s,  max=1.7s)  }  0-1

8/8/6pk/8/3qpn1n/R6P/Q6K/8 b - - 0 53


That's an amazing game! Congratulations! :-)

Regards,
Gaмßito.
Parent - - By Leto (***) [us] Date 2009-07-11 20:10
I watched it live, congrats Highendman.  A kibitzer by the name of Super_GM said "earlier Cluster should play Bg5 and take on h4, would give good chances to win." 
Parent - - By Highendman (****) Date 2009-07-11 20:21
thanks all. that kibitzer was wrong ;) I may post some analysis later.
so at the moment I have one win with white, one win with black and 11 or 12 draws (lost count).
Parent - By Leto (***) [us] Date 2009-07-11 20:23
An impressive record against the cluster
Parent - - By Fulcrum2000 (****) [nl] Date 2009-07-11 20:25
Excellent, and this time not a clear book error from the cluster as far as I can tell. Good job Highendman!. You will be fighting for the podium next freestyle.
Parent - - By Bouddha (****) [ch] Date 2009-07-11 21:03
Hmmm.

This line with 0-0 is wellknown on playchess.com and results are not good for white.

the h5 line has been played a lot the last months and there are lots of progress made. Other lines seems more promising even so black currently holds the ballance
Parent - - By Fulcrum2000 (****) [nl] Date 2009-07-11 21:16
It's maybe not the most optimal line, but no way we can speak about a book error here.
Parent - By Bouddha (****) [ch] Date 2009-07-11 21:40
Possibly you are right.

But usually even when Rybka after the opening evaluate the position as = if after thousands of games, the results are much more in black favor, I consider this more important than the current engine evaluation.

In the h5 sicilian, 0-0 really dont give good results and it seems that white King is going the wrong side of the board closer the the black h pawn that will create holes in white position.

regards
Parent - - By turbojuice1122 (Gold) [us] Date 2009-07-11 22:00

> the h5 line has been played a lot the last months and there are lots of progress made.


That's interesting--I had independently found the benefits of this line using IDeA, and it does seem to be the best move in the position.
Parent - - By Bouddha (****) [ch] Date 2009-07-11 22:24
You mean h5 for black is not the best ?

many of us managed to claimb to over 2800 on playchess using this line.

But I would not be surprised if IDEA doesn't think h5 is best. You must help a little the engine just after h5.

But I would agree that Be7 is certainly close to h5 or on par on an absolute.
But Be7 has been worked a lot on playchess before h5.  => h5 is "newer"
Parent - By Vempele (Silver) [fi] Date 2009-07-11 22:25

>>and [h5] does seem to be the best move in the position.
> You mean h5 for black is not the best ?


Note the lack of a 'not' after the 'does'.
Parent - - By Master Om (Bronze) [in] Date 2009-07-12 03:52
You mean h5 for black is not the best ?
Well it's playable.
I think CR played the opening badly. There is no way to attack king when 2 knights are there to support the king.
In human games it is well known fact that Knights+Knights+Queen endgames are superior to Queen +Rooks +Rooks with good out posts for knights. This has been prevailed here. Nice endgame work by Hghendman. Two Exchange Sacrifices and good night out post with last minutes tactics make this game a good one probably the best Confident win against CR. Still i think CR book is bad. According to analysis Black has more wining cahances here in the opening until and unless whites achieves this position.
r2q1rk1/1p1n1pb1/p2p1np1/3Pp2p/6P1/1N2BP2/PPPQB2P/2KR3R b - g3 0 14

only way to tackle h5 is to castle opposite and break the fortress with pawn storm and dont give black a chance to make any counter play on queen side. So o-o-o helps in this strategy. For me 13.h3 is the crap move.  What i have found in last 2 months in Playchess and my analysis for my next Om Master Book is that Rybka evals this position roughly equal ...but when the black bishops get exchanged at g5 after Bg7-f6-g5 in conjuction to h4 and Nh5 Rybka evals suddenly change when the knight sits in g3 ...just like white having a pain in the knee. So i have used Shredder 11,Naum 4 and DF 10 in analysis when there is a fortress case.Another thing of losing the match for CR is it is not meant for playing tactically. More or less its tendency is to play positionaly. But here it was agrressive but solid defence prevailed.
Parent - - By Bouddha (****) [ch] Date 2009-07-12 06:20
I dont know what you did read in my posts but what you have written is exactly the same as what I found !

I sead that 0-0 for white is not the best and imho give black an edge even so Rybka evals this as =
Parent - By turbojuice1122 (Gold) [us] Date 2009-07-12 11:26
Much ado about nothing.  ...h5 is the best move in the position, I think, as I already said. :-)
Parent - By Master Om (Bronze) [in] Date 2009-07-12 12:41
I accept your statements +1
Parent - By turbojuice1122 (Gold) [us] Date 2009-07-12 11:26
Trace back to the statement.  IDeA thinks that ...h5 is the best move.  What's more, if it's not played on the eighth move, it wants to play it whenever reasonably possible.
Parent - - By Sciolto (***) [nl] Date 2009-07-11 22:54 Edited 2009-07-11 23:04
What is a clear book error by your definition ?
In my humble opinion h3? followed by 0-0 is a clear book error.
Obviously Rybka was out of book too soon and played two really bad moves.
Parent - - By Uri Blass (*****) [il] Date 2009-07-11 23:10
I do not know but the cluster's evaluation seems strange to me
The cluster showed 0.38/19 evaluation for 17.Na5
Rybka3 is clearly more realistic and shows 0.03/18 and 0.02/19

Maybe the cluster could perform better if it had the superior evaluation of rybka3.
The cluster was surprised by 17...Qc7 but it is exactly the move that rybka3 expects.

analysis on slow hardware by rybka3.

Rechenschieber,R - Highendman, 60m + 1s, unrated rapid 2009
2rq1rk1/1p1n1pb1/p2p2p1/3Pp2n/7p/1N2BP1P/PPPQB1P1/2R1R1K1 w - - 0 1


Analysis by Rybka 3 1-cpu 32-bit :

17.Qd2-b4
  ²  (0.45)   Depth: 2   00:00:00
17.Qd2-b4
  ²  (0.58)   Depth: 3   00:00:00
17.Qd2-b4
  ²  (0.45)   Depth: 4   00:00:00
17.Qd2-b4 Qd8-c7
  ²  (0.56)   Depth: 5   00:00:00
17.Qd2-b4 Qd8-c7 18.Qb4xh4
  ²  (0.46)   Depth: 6   00:00:00  2kN
17.Qd2-b4 Nd7-f6 18.c2-c4 Qd8-d7
  ²  (0.48)   Depth: 7   00:00:00  4kN
17.Qd2-b4 Nd7-f6 18.c2-c4 b7-b5 19.Nb3-a5
  ²  (0.49)   Depth: 8   00:00:00  9kN
17.Qd2-b4 Nd7-f6 18.c2-c4 b7-b5 19.Nb3-a5 Nh5-f4
  ²  (0.51)   Depth: 9   00:00:00  16kN
17.Qd2-b4 Nd7-f6 18.c2-c4 b7-b5 19.Qb4-a3 Rc8-a8 20.Nb3-a5
  ²  (0.50)   Depth: 10   00:00:01  39kN
17.Qd2-b4 Nd7-f6 18.c2-c4 b7-b5 19.Nb3-a5 Nh5-f4 20.Na5-c6 Qd8-c7
  ²  (0.52)   Depth: 11   00:00:04  104kN
17.Qd2-b4 Nd7-f6 18.c2-c4 Qd8-d7 19.Nb3-a5 Rc8-c7 20.Be3-g5 Nh5-f4 21.Be2-f1 b7-b5
  ²  (0.50)   Depth: 12   00:00:07  175kN
17.Qd2-b4 Bg7-f6 18.Be2-f1 Qd8-c7 19.c2-c3 Nh5-g3 20.Bf1-d3 Bf6-e7 21.Rc1-d1 Rf8-e8 22.Qb4-a5 Qc7xa5
  =  (0.24)   Depth: 13   00:00:46  1303kN
17.Qd2-b4 Bg7-f6 18.Be2-f1 Qd8-c7 19.c2-c3 Nh5-g3 20.Bf1-d3 Bf6-e7 21.Rc1-d1 Rf8-e8 22.Nb3-a5 Nd7-c5 23.Bd3-c2 Be7-f6
  =  (0.22)   Depth: 14   00:01:09  2058kN
17.Qd2-b4 Bg7-f6 18.Be2-f1 Qd8-c7 19.c2-c3 Nh5-g3 20.Bf1-d3 Bf6-d8 21.a2-a4 f7-f5 22.a4-a5 Bd8-f6 23.Rc1-d1 Rf8-e8
  =  (0.23)   Depth: 15   00:01:55  3403kN
17.Qd2-b4 Bg7-f6 18.Be2-f1 Qd8-c7 19.c2-c3 Nh5-g3 20.Bf1-d3 Bf6-d8 21.a2-a4 f7-f5 22.a4-a5 Bd8-f6 23.Bd3-c2 Rc8-e8 24.Nb3-d2
  =  (0.21)   Depth: 16   00:03:18  6057kN
17.Qd2-b4 Bg7-f6 18.Be2-f1 Qd8-c7 19.c2-c3 Nh5-g3 20.Bf1-d3 Bf6-d8 21.a2-a4 f7-f5 22.a4-a5 Bd8-f6 23.Bd3-c2 Rc8-e8 24.Nb3-d2
  =  (0.21)   Depth: 17   00:05:08  9600kN
17.Nb3-a5 Qd8-c7 18.Qd2-b4
  =  (0.03)   Depth: 18   00:31:57  56407kN
17.Nb3-a5 Qd8-c7 18.c2-c4 Nh5-g3 19.Be2-d3 Rf8-e8 20.b2-b4 Nd7-f6
  =  (0.02)   Depth: 19   01:09:56  114285kN

(so k,  12.07.2009)
Parent - - By Highendman (****) Date 2009-07-11 23:21
The Cluster was playing with high contempt - probably 50. That's why it scores itself higher than 0 contempt.
That's probably why it also went for the (overly?) aggressive pawn push on the queen side, and didn't realize the super initiative black would get after giving up the exchange.

As an aside: I remembered Vas saying in a post a while ago that he has no 'knight outpost' logic in his engine -> that he tried it and couldn't make it worthwhile.
So with that in mind, I tried in all my games to reach positions with knight outposts, figuring maybe somehow the engine under evaluates them ;)
Parent - By Patricio (***) [ar] Date 2009-07-12 03:27
If you could take advantage of that you are really great.

Congratulations.

Best regards,

                       Patricio.
Parent - - By Uri Blass (*****) [il] Date 2009-07-12 04:55
I see no reason for high contempt for the cluster.
It is simply handicapping the cluster.

high contempt is something that you use against significantly weaker opponent and it was clearly not the case here.

Uri
Parent - By Capa (***) [us] Date 2009-07-12 10:49
I assume the high contempt is to avoid the plethora of draws.  Perhaps it is better to draw than to lose. ;)
Parent - By Bouddha (****) [ch] Date 2009-07-12 06:23
+1 for the contempt.

but it doesn't explain why it prefers h3 move, which is imho not good
Parent - By Vasik Rajlich (Silver) [pl] Date 2009-07-12 06:24
The knight on d4 gets a really big bonus, though. :)

It was a nice exchange sacrifice. Thematic, yes, but it can also backfire. To prove it in a game like this is very impressive.

Vas
Parent - - By Lukas Cimiotti (Bronze) [de] Date 2009-07-12 07:15
Yes, contempt was 50. This game answers one of the questions I had: can contempt hurt? The answer is yes.
Also it's very dangerous to allow only 12 book moves and then play B90 - which is so popular on playchess ;)

Congrats - a very nice win
Parent - By Bouddha (****) [ch] Date 2009-07-12 08:31
agreed on all points !
Parent - - By M ANSARI (*****) [kw] Date 2009-07-12 22:29
OK this answers a lot of the questions.  h3 was just a horrible move and with B90 you really can't just leave the engine alone since the initial book moves follow a strategy that is man made and not engine made ... I think even leaving the cluster without book would be better since it would have at least be following its unique strategy.  In my book I have 0-0 and 0-0-0 and c4 as possible alternatives to h3 with h3 getting zero wins and zero draws and 3 losses.  Still even after h4 by black and the obvious opening for highway for the black knights to have an outpost at g3 I would prefer white but white has to immediately pin the Knight and attack h3 pawn ... who knows maybe h4 is not optimal.  As it was, delaying the pin made pinning the knight impossible because the bishop could be trapped.  If Knight was immediately pinned and eventually black tries center pawn advance the white bishop pair would become very active and again I would prefer white.  Instead cluster played strangely by going after material on the queenside, readily gave up the bishop pair in an open position and basically self destructed.  I was surprised since on my octa my R3 could see that tactically taking material on queenside would backfire and my guess is that high contempt is maybe the reason.

Anyway all these games are great to pinpoint problems and I am not so sure contempt of 50 is a good idea at slow time controls ... it is ok to try out, but in a different opening or at least deeper into an opening.  Playchess books (especially B90) have been analyzed extremely deeply and taking them on really negates the advantage of the Cluster.  I personally have spent hundreds of computer hours on some moves in B90 and am sure many here have done the same.  Ofcourse I would not want the Cluster to use a killer book as personally I am interested in seeing how it evaluation has changed.  So far evaluation seems very good ... but I would still love to see how an added extra .3 to .4 centipawns to its evaluation of a bishop pair in an open position would be like.
Parent - - By yanquis1972 (****) [us] Date 2009-07-13 03:37
i wouldnt mind seeing a bookless cluster game or several.  i've wondered about how advanced software + killer hardware would handle the openings unguiged, this seems like a fine time to find out.
Parent - By Ray (****) Date 2009-07-13 12:06

> i wouldnt mind seeing a bookless cluster game or several


I would love to see someone challenge the cluster at chess960 !!!!!!!!!!!
Parent - - By Vasik Rajlich (Silver) [pl] Date 2009-07-13 05:23

> Playchess books (especially B90) have been analyzed extremely deeply and taking them on really negates the advantage of the Cluster.  I personally have spent hundreds of computer hours on some moves in B90 and am sure many here have done the same.


Indeed, in some positions, forcing the cluster to try to find known theory is obviously counterproductive. The theory is based not just on hundreds of hours of analysis but also on humans digesting tons of game results. In addition, the human centaurs know the terrain well and can pick the moves which give them the best chances to get the kind of game they want, much more so than in unknown positions.

There is a lot more which could be written about this topic. We are investigating it and thinking about it.

Re. contempt: contempt does produce some desirable anti-human/anti-centaur behavior (fe. keeping pieces on the board), but it has some major side-effects (fe. causing the cluster to sidestep variations which end in repetitions). We need a more direct anti-centaur approach here.

We've learned a lot from this series of games against Highendman. In August we'll come back with solutions to these problems.

Vas
Parent - - By Highendman (****) Date 2009-07-13 07:15
Vas I'm happy to be destroyed by a new and improved version - or not ;)
I'm going to be away all of August (Galapagos islands!) but would really look forward to more games before and after that. The prospect of an anti-centaur contempt is exciting.
Parent - - By M ANSARI (*****) [kw] Date 2009-07-13 14:28
Galapagos !!!  I have always wanted to go there.  Take plenty of pix and post them here.
Parent - - By Highendman (****) Date 2009-07-13 15:06
Will do. 28 nights in the Galapagos and Amazons and across Ecuador. I told my wife I hope it's the online store Amazon, but she confirmed it isn't ;)
Parent - By Vasik Rajlich (Silver) [pl] Date 2009-07-14 06:02
I think it's named after the store, though. :)

Vas

ps. Have a good trip. We'll be waiting for you when you return!
Parent - - By Bouddha (****) [ch] Date 2009-07-12 06:21
+1

h3 a move that Rybka 3 often plays in this positions.

but h3 is imho not good
Parent - - By Nelson Hernandez (Gold) [us] Date 2009-07-12 12:08
13.h3??  My fifth-ranked move in that position.  Here's the problem with Contempt: there are times when avoiding a draw is the exactly wrong thing to do!  You have to have dynamic contempt based on what is actually going on fer chrissakes.

I'd have played 13.O-O-O almost for sure: 13...Qc7 14.Kb1 O-O 15.g4 Rfc8 16.Rc1 a5 17.g5 Ne8 and now white would have some interesting choices.  Other variations lead to probable draws.
Parent - - By Uri Blass (*****) [il] Date 2009-07-12 12:18
I guess that contempt=50 at time control of 60+1 is even bad against unassisted rybka on good hardware.
I wonder if people tested cluster rybka against rybka3 not at fast blitz to find the best contempt value.

Uri
Parent - By Lukas Cimiotti (Bronze) [de] Date 2009-07-12 13:08

>I wonder if people tested cluster rybka against rybka3 not at fast blitz to find the best contempt value.


You need hundreds of games per setting to get reliable results. This takes years at a time control of 60+1. So the simple answer is no.

Regards,
Lukas
Parent - By Christoph Fieberg (*) [de] Date 2009-07-12 13:22
In my book 10.Nd5 is not the best (43%, 14 games, av 2550), better moves in the book are 10.a4 (52%, 30 games, av 2533) or 10.Le2 (62%, 13 games, av 2551).
The most popular move 10.0-0-0 is the weakest (39%, 63 games, av 2519).
Parent - - By Highendman (****) Date 2009-07-11 20:28
Thanks for posting the game Gambito. If you want, you could edit the title - it's actually game 13 or 14 (or 15? I lost count) against the cluster by me, not game 9. The rest were all draws :(
Parent - - By Uly (Gold) [mx] Date 2009-07-11 20:34 Edited 2009-07-11 20:38
Congrats on the win! Do you think that your winning chances increase the more games you play? (while the cluster can't adapt to you.)

> it's actually game 13 or 14 (or 15? I lost count)


Title changed, any better idea?
Parent - - By Lukas Cimiotti (Bronze) [de] Date 2009-07-12 07:18

>while the cluster can't adapt to you


The cluster can adapt to him :)
I won't play against him with high contempt anymore - and I won't cut the book to 12 moves. And maybe I should also only play 1.d4 against him ;)
Parent - By Highendman (****) Date 2009-07-12 08:28
then you'd be like karpov who abandoned e4 as white against garry. i feel honoured already even though i gave myself this comparison. lol.
Parent - By Bouddha (****) [ch] Date 2009-07-12 08:33
Well, you can still play 1.e4 but in the h5 B90, castle King side is imho not the way to go.

the current situation on playchess with 0-0-0 is unclear but it seems that black has good counterchances.

Maybe the Bg5 variation should be played, but necessite a lot of preparation also. (B97 and others....)
Parent - By Nelson Hernandez (Gold) [us] Date 2009-07-12 12:48
This post is like a third victory for Highendman.  The Cluster Team has been taken out of it's normal game!  It's strategy has been overthrown!

Here's a tip, Highendman.  If, someday (we hope) there is another Freestyle with significant prize money (sigh), you must play anonymously.  You don't want the Rybka team adjusting it's play settings just against you personally.  And don't think they wouldn't!
Parent - - By Master Om (Bronze) [in] Date 2009-07-13 03:15
Does Rybka Cluster play 960?. Y don't have a 960 challenge against Highendman as he must have the Rybka 960 ,Sjeng ,Fritz etc to play.
Parent - - By Lukas Cimiotti (Bronze) [de] Date 2009-07-13 04:56

>Does Rybka Cluster play 960?


No, only normal chess
Parent - By Highendman (****) Date 2009-07-13 07:18
Also, I'm not 'into' 960 - not because of lack of book, but because some of the aesthetic feel, for me, is gone. It just seems like a mess and I have no clue how to form a plan other than wiping all pieces off the board with a virtual hand, to get them back to their proper starting position. A limitation for me, surely, but such is life.
Parent - By InspectorGadget (*****) [za] Date 2009-07-13 11:26
Highendman is starting to be our hero here. Do we need a poll as to what would happen in a 10-game match between the cluster and Highendman?
Parent - - By Gaмßito (****) [cr] Date 2009-07-11 20:38
Thanks for posting the game Gambito. If you want, you could edit the title - it's actually game 13 or 14 (or 15? I lost count) against the cluster by me, not game 9. The rest were all draws :-(

Are you sure? :-)

I only have 9 games. I saw yesterday Lukas's file, and you only had 8 games.

But it's possible that I could be wrong. Could you please provide all your games?

Regards,
Gaмßito.
Up Topic Rybka Support & Discussion / Rybka Discussion / (Game 13 14 or 15) Rybka Cluster vs Highendman = 0-1
1 2 3 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill