TalkChess.com

strange behaviour of Spark 0.4 in this position against Rybka 4 at Leiden 2010 today
[D] 2k2r1r/1pqb2p1/p3p1p1/3pn1N1/8/1NP3Q1/PP4PP/R4RK1 w - - 0 19

the best move here is Rxf8 and position is equal. After Spark's move 19 Nd4 Rybka gains the upperhand with fine move sequence 19 - Rf6 20. Re1 - Nd3!!

The strange thing here is that Spark isn't able to find Rxf8. After depht 21 he gives black - 0,15 better and still will not switch to beter move Rxf8. When you give 19 Rxf1 he evals this as 0,00

Kind regards Bram

Hi,

Is there a website of some kind where I can download the games and/or see the results? This is for the computer chess.

A quick search on Google did not find anything.

Regards,
Vlad.

Totally off topic, but I just have to say this:

Vlad, I always read your name as Vlad Stalemate. Totally awesome!

Which gui are you using? On my system under chessbase and arena spark 0.4 finds the move instantly

New game,
2k2r1r/1pqb2p1/p3p1p1/3pn1N1/8/1NP3Q1/PP4PP/R4RK1 w - - 0 1

Analysis by spark-0.4:

1.h3
-/+ (-0.82) Depth: 1 00:00:00
1.h3
-/+ (-0.82) Depth: 1 00:00:00
1.Nd4
=/+ (-0.69) Depth: 1 00:00:00
1.Nd4 Qd6 2.Rxf8+ Rxf8
=/+ (-0.38) Depth: 2/1 00:00:00
1.Nd4 Qd6 2.Rxf8+ Rxf8
=/+ (-0.38) Depth: 3/3 00:00:00
1.Nd4 Qd6 2.Rxf8+ Rxf8 3.Re1
= (-0.14) Depth: 4/4 00:00:00
1.Nd4 Rf6 2.Rxf6 gxf6 3.Ngxe6 Bxe6 4.Nxe6
= (0.02) Depth: 5/8 00:00:00
1.Rxf8+ Rxf8 2.Nd4 Rh8 3.h3 Nc4 4.Qxc7+ Kxc7 5.Ngxe6+ Kd6
= (0.08) Depth: 6/10 00:00:00 15kN
1.Rxf8+ Rxf8 2.Re1 Qb6+ 3.Nd4 Nc4 4.Ngxe6 Bxe6 5.Rxe6 Qxb2 6.Rxg6
= (0.16) Depth: 7/12 00:00:00 39kN
1.Rxf8+ Rxf8 2.Re1 Qb6+ 3.Nd4 Nc4 4.b3 e5 5.bxc4 exd4
= (-0.04) Depth: 8/12 00:00:00 54kN
1.Rxf8+ Rxf8 2.Nd4 Re8 3.Re1 Nc4 4.Qxc7+ Kxc7 5.Ndxe6+ Kd6 6.Nf7+ Ke7 7.Nfg5 Kd6
= (0.00) Depth: 9/16 00:00:00 130kN
1.Rxf8+ Rxf8 2.Nd4 Re8 3.Re1 Nc4 4.Qxc7+ Kxc7 5.Ndxe6+ Kd6 6.Nf7+ Ke7 7.Nfg5 Kd6
= (0.00) Depth: 10/18 00:00:00 213kN
1.Rxf8+ Rxf8 2.Nd4 Re8 3.Re1 Nc4 4.Qxc7+ Kxc7 5.b3 e5 6.Ndf3 Nd6 7.Rxe5 Ne4 8.Rxe8 Bxe8
= (0.10) Depth: 11/19 00:00:00 364kN
1.Rxf8+ Rxf8 2.Nd4 Re8 3.Re1 Nc4 4.Qxc7+ Kxc7 5.b3 Na3 6.Ndxe6+ Kb6 7.Kf2 Nb5 8.c4 dxc4 9.bxc4
= (0.12) Depth: 12/22 00:00:00 589kN
1.Rxf8+ Rxf8 2.Nd4 Re8 3.Re1 Nc4 4.Qxc7+ Kxc7 5.b3 Na3 6.Ndxe6+ Kb6 7.Kf2 Nb5 8.c4 dxc4 9.bxc4
= (0.12) Depth: 13/25 00:00:00 933kN
1.Rxf8+ Rxf8 2.Nd4 Re8 3.Re1 Nc4 4.Qxc7+ Kxc7 5.b3 Na3 6.Ndxe6+ Kb6 7.Re3 Nb5 8.Nf4 Rxe3 9.Nxd5+ Kc5 10.Nxe3 Nxc3
= (0.04) Depth: 14/26 00:00:00 1737kN
1.Rxf8+ Rxf8 2.Nd4 Re8 3.Re1 Nc4 4.Qxc7+ Kxc7 5.b3 Na3 6.Ndxe6+ Kb6 7.Re3 Nb5 8.Nf4 Rxe3 9.Nxd5+ Kc5 10.Nxe3 Nxc3
= (0.04) Depth: 15/26 00:00:01 3848kN
1.Rxf8+ Rxf8 2.Nd4 Re8 3.Re1 Nc4 4.Qxc7+ Kxc7 5.b3 Nb2 6.Ndxe6+ Bxe6 7.Rxe6 Rxe6 8.Nxe6+ Kd6 9.Nd8 Kc7 10.Nf7 Nd1 11.c4 dxc4 12.bxc4
= (0.09) Depth: 16/27 00:00:01 6713kN
1.Rxf8+ Rxf8 2.Nd4 Re8 3.Re1 Nc4 4.Qxc7+ Kxc7 5.b3 Nb2 6.Ndxe6+ Bxe6 7.Rxe6 Rxe6 8.Nxe6+ Kd6 9.Nd8 Kc7 10.Ne6+
= (0.00) Depth: 17/30 00:00:04 12794kN
1.Rxf8+ Rxf8 2.Re1 Qb6+ 3.Nd4 Nc4 4.Ngxe6 Bxe6 5.Rxe6 Qxb2 6.h3 Nd2 7.Qxg6 Qxa2 8.Qxg7 Qb1+ 9.Kh2 Nf1+ 10.Kg1 Nd2+
= (0.00) Depth: 18/36 00:00:08 40750kN
1.Rxf8+ Rxf8 2.Re1 Qb6+ 3.Nd4 Nc4 4.Ngxe6 Bxe6 5.Rxe6 Qxb2 6.h3 Nd2 7.Qxg6 Qxa2 8.Qxg7 Qb1+ 9.Kh2 Nf1+ 10.Kg1 Nd2+
= (0.00) Depth: 19/38 00:00:12 68558kN

I am using F12 gui. When starting analysis in this position Spark is giving different moves in stead of each time the same move when restarting with a restart of Fritz.

vladstamate wrote:Hi,

Is there a website of some kind where I can download the games and/or see the results? This is for the computer chess.

A quick search on Google did not find anything.

Regards,
Vlad.

I downloaded the game from the playchess server

regards Bram

beram wrote:I am using F12 gui. When starting analysis in this position Spark is giving different moves in stead of each time the same move when restarting with a restart of Fritz.

Spark moves should be perfectly reproducably provided that
- you start each search with an empty hashtable
- you use a single cpu/thread only

Anyway, in Leiden I am using the current development version of
Spark, not the recently released 0.4 version. That could also explain
why it played the wrong move here (although this version appears to
be considerably stronger overall). I will examine what happened here
after the tournament, thank you for noticing.