test positions for weak playing engines only
Moderators: hgm, Rebel, chrisw
-
- Posts: 7220
- Joined: Mon May 27, 2013 10:31 am
Re: test positions for weak playing engines only
Until time is up or engine decides move is good enough.
-
- Posts: 558
- Joined: Sat Mar 25, 2006 8:27 pm
Re: test positions for weak playing engines only
Typically, you pick a time limit, say 10 seconds per position. Then see whether the last pv your program output has selected one of the best moves.flok wrote:Hi,
This is probably a dumb question but I want to be sure that I'm doing the right thing.
Let's say I have an EPD-string like this:
8/7p/5k2/5p2/p1p2P2/Pr1pPK2/1P1R3P/8 b - - bm Rxb2; id "WAC.002";
How should I test this?
Do I start at search-depth and then keep increasing until either time is up or a certain depth is reached? Or do I keep searching until I find that Rxb2? I'm asking because sometimes the move is found at, say, depth 3 but at depth 4 something else is chosen.
regards
Afterward, you can go back and see if they picked the right move for the right reason (e.g. seeing the outcome of a winning capture sequence). Or you can focus on positions that you take a long time to find that other programs find quickly.
-
- Posts: 7220
- Joined: Mon May 27, 2013 10:31 am
Re: test positions for weak playing engines only
Don't like it that test positions have only best move and not a PV. If my engine cannot find it how do I know best move is correct and what might have been the cause.
-
- Posts: 7220
- Joined: Mon May 27, 2013 10:31 am
Re: test positions for weak playing engines only
Maybe a work around is to add positions position + best move for all moves of a PV. In this way input format doesn't need to be modified. So you add easier test positions for the same problem.
-
- Posts: 859
- Joined: Mon Aug 10, 2009 10:05 pm
- Location: Italy
- Full name: Stefano Gemma
Re: test positions for weak playing engines only
Satana (that is one of the weakest engines in the world) finds Qf3 after a few instants (ply 6) and Qb3 after a few seconds (ply 8). It keeps this move as the best one at ply 10... and then I've stopped. It seems to me that ply 8 is the minimum to find this move. What ply does your software reach?Henk wrote:https://chessprogramming.wikispaces.com/Test-Positions
[d] rn1qkb1r/pp2pppp/5n2/3p1b2/3P4/2N1P3/PP3PPP/R1BQKBNR w KQkq - 0 1
I tried the first one of "CCR one hour test" but Skipper thinks Nf3 is better than Qb3.
Author of Drago, Raffaela, Freccia, Satana, Sabrina.
http://www.linformatica.com
http://www.linformatica.com
-
- Posts: 7220
- Joined: Mon May 27, 2013 10:31 am
Re: test positions for weak playing engines only
Code: Select all
Depth Value Time(seconds) Nodes
13 0.07 10.00 432989 g1f3
12 0.03 4.75 201217 f1b5
11 0.00 3.52 148977 d1b3
10 0.00 1.91 78964 d1b3
9 0.01 0.72 28456 d1b3
8 0.01 0.29 11112 d1b3
7 0.01 0.27 10053 d1b3
6 0.01 0.18 6522 d1b3
5 0.01 0.10 2977 d1b3
4 0.01 0.09 2714 d1b3
3 0.01 0.05 1176 d1b3
2 0.01 0.04 791 d1b3
1 0.01 0.02 83 d1b3
-
- Posts: 7220
- Joined: Mon May 27, 2013 10:31 am
Re: test positions for weak playing engines only
Stockfish 6
Code: Select all
26 +0.17 43.7M 0:32.85 Qb3 e6 Qxb7 Nbd7 Nf3 Be7 Be2 a5 O-O O-O Qc6 Rb8 h3 Ne4 Nxe4 Bxe4 Qa4 Qc7 Bb5 Nf6 Ne5 Qb6 Nc6 Qxb5 Nxe7+ Kh8 Qxb5 Rxb5 Re1 h6 Bd2 Rxb2
26 +0.16 33.0M 0:24.67 Qb3 e6?
25 +0.22 25.6M 0:19.19 Qb3 e6 Qxb7 Nbd7 Nf3 Be7 Qa6 O-O Be2 Qb8 h3 Rc8 Bb5 Rc7 O-O Ne4 Bd2 Nd6 Bxd7 Rxd7 Ne5 Rb7 b3 Qc8 Nc6 Bg5 f4 Rc7 Qxc8+ Raxc8 Nxa7
24 +0.19 19.7M 0:14.72 Qb3 e6 Qxb7 Nbd7 Nf3 Be7 Qa6 O-O Be2 Qb8 O-O Rc8 h3 Ne4 Nxe4 Bxe4 Bb5 Qb6 Qxb6 Nxb6 Ne5 Bf6 f3 Bxe5 dxe5 a6
24 +0.20 16.4M 0:12.27 Qb3!
24 +0.14 14.4M 0:10.91 Qb3 e6?
23 +0.20 10.8M 0:08.14 Qb3 e6 Qxb7 Nbd7 Nf3 Be7 Be2 a5 Qb3 O-O O-O Ne4 Nxe4 Bxe4 Qa4 Rb8 h3 Qc7 Bb5 Nf6 b3 Bb4 Bd2 Qb6 Bxb4
23 +0.15 9.6M 0:07.24 Qb3 e6?
22 +0.21 6.7M 0:05.02 Qb3 e6 Qxb7 Nbd7 Nf3 Be7 Be2 O-O O-O a6 Qc6 Qc8 Qa4 Nb6 Qd1 h6 Bd2 Nc4 Bxc4 dxc4 Qe2 Ne4 Nxe4
22 +0.20 6.0M 0:04.50 Qb3!
22 +0.14 5.7M 0:04.27 Qb3 e6?
21 +0.20 4.1M 0:03.06 Qb3 e6 Qxb7 Nbd7 Nf3 Be7 Be2 O-O O-O a5 Qb3 Ne4 Nxe4 Bxe4 Qa4 Nb6 Qb5 Bxf3 Bxf3 Nc4 Be2 Rc8
20 +0.10 3.1M 0:02.25 Qb3 e6 Qxb7 Nbd7 Nf3 Be7 Be2 O-O O-O a5 Qb3 Ne4 Nxe4 Bxe4 Bd2 Rb8 Qc3 Bb4 Qc1 Rc8 Qd1 Rc2 Rb1
19 +0.13 2.3M 0:01.72 Qb3 e6 Qxb7 Nbd7 Nf3 Be7 Be2 O-O O-O a5 Qb3 Ne4 Nxe4 Bxe4 h3 Rb8 Qa4 Rb4 Qd1 Rb6 b3
18 +0.17 1.4M 0:01.02 Qb3 e6 Qxb7 Nbd7 Qa6 Bb4 Nf3 O-O Bd2 Rb8 Rc1 Qc8 a3 Bxc3 Bxc3 Ne4 Qxc8 Rfxc8 h3
17 +0.17 1.2M 0:00.89 Qb3 e6 Qxb7 Nbd7 Qa6 Bb4 Nf3 O-O Bd2 Rb8 Rc1 Qc8 a3 Bxc3 Bxc3 Ne4 Qxc8 Rfxc8 h3 Nxc3
16 +0.17 1.1M 0:00.83 Qb3 e6 Qxb7 Nbd7 Qa6 Bb4 Nf3 O-O Bd2 Rb8 a3 Be7 Bb5 Qc8 Qa4 Qb7 O-O Nb6
15 +0.17 963113 0:00.70 Qb3 e6 Qxb7 Nbd7 Qa6 Bb4 Nf3 O-O Bd2 Rb8 Rc1 Qc8 Bb5 Qxa6 Bxa6
14 +0.25 482005 0:00.34 Nf3 Nc6 Be2 e6 Nh4 Bg6 O-O a6 Bd2 Bd6 h3 Rc8 a3 O-O Nxg6 hxg6 Rc1 b5 Bd3 Qc7
13 +0.22 215874 0:00.16 Nf3 e6 Nh4 Bg6 Be2 Nc6 O-O a6 Bd2 Bd6 h3 Rc8 Nxg6 hxg6 a3 O-O
12 +0.16 144827 0:00.10 Qb3 e6 Qxb7 Nbd7 Qa6 Bb4 Nf3 O-O Bb5 Rb8 O-O Qb6
11 +0.13 86570 0:00.06 Nf3 e6 Nh4 Bg6 Be2 Nc6 O-O Bd6 Bd2 a6 h3 O-O Nxg6 hxg6
10 +0.32 36546 0:00.02 Nf3 e6 Nh4 Bg6 Be2 Nc6 O-O Bd6 Bd2 O-O Nxg6
9 +0.29 12245 0:00.00 Nf3 e6 Ne5 Nbd7 g4 Nxe5 gxf5 Nc6 fxe6
8 +0.29 9337 0:00.00 Nf3 e6 Ne5 Nbd7 g4 Nxe5 gxf5 Nc6
7 +0.41 6146 0:00.00 Nf3 Nc6 Qa4 e6 Ne5 Be7 Nxc6
6 +1.49 1155 0:00.00 g4 Nxg4 h3 Nxf2 Bb5+ Nc6 Kxf2
5 +1.49 918 0:00.00 g4 Nxg4 h3 Nxf2 Bb5+ Nc6
4 +1.49 795 0:00.00 g4 Nxg4 h3 Nxf2 Bb5+ Nc6 Kxf2
3 +0.32 312 0:00.00 Nf3 e6 Bb5+ Nc6 Ne5
2 +0.97 115 0:00.00 Nf3 b6 Bb5+ Nbd7
1 +0.49 45 0:00.00 Nf3
-
- Posts: 1563
- Joined: Thu Jul 16, 2009 10:47 am
- Location: Almere, The Netherlands
Re: test positions for weak playing engines only
Nightmare sees it at depth 12, and sometimes at depth 11, it depends upon the randomness of the SMP.
Today I found another bug which I introduced some time ago by refactoring.
It was SMP related, I always tune my engine single core, that's why I didn't notice it.
You're not the only one plagued by bugs.
Code: Select all
info depth 1 score cp 60 time 0 nodes 257 nps 257 tbhits 0 pv g1f3
info depth 2 score cp 10 time 1 nodes 1003 nps 1003000 tbhits 0 pv f1d3 d8d7
info depth 3 score cp 15 time 2 nodes 2983 nps 1491500 tbhits 0 pv g1f3 b8c6 f1d3
info depth 4 score cp 5 time 4 nodes 5015 nps 1253750 tbhits 0 pv g1f3 b8c6 f1d3 d8d7
info depth 5 score cp 42 time 9 nodes 13602 nps 1511333 tbhits 0 pv f1d3 f5d3 d1d3 b8c6 g1f3
info depth 6 score cp 23 time 21 nodes 40288 nps 1918476 tbhits 0 pv f1b5 b8c6 g1f3 a7a6 b5c6 b7c6
info depth 7 score cp 53 time 29 nodes 82048 nps 2829241 tbhits 0 pv f1b5 b8c6 g1f3 d8d6 e1g1 f6e4 f3e5
info depth 8 score cp 58 time 47 nodes 185162 nps 3939617 tbhits 0 pv f1d3 f5d3 d1d3 e7e6 g1f3 b8c6 e1g1 f8d6
info depth 9 score cp 61 time 68 nodes 336656 nps 4950823 tbhits 0 pv g1f3 e7e6 d1b3 b8c6 b3b7 a8c8 f1b5 d8d7 b5c6
info depth 10 score cp 53 time 94 nodes 567809 nps 6040521 tbhits 0 pv g1f3 e7e6 d1b3 b8d7 b3b7 f8d6 f1b5 e8g8 e1g1 g8h8
info depth 11 score cp 57 time 131 nodes 907178 nps 6925022 tbhits 0 pv g1f3 e7e6 d1b3 b8d7 b3b7 f8d6 f1b5 e8g8 e1g1 d8b6 b5c6
info depth 12 score cp 29 time 374 nodes 3872182 nps 10353427 tbhits 0 pv d1b3 d8c7 g1e2 e7e6 e2g3 f5g6
info depth 13 score cp 26 time 733 nodes 8722731 nps 11900042 tbhits 0 pv d1b3 d8c7 g1f3 b8c6 f1e2 e7e6 e1g1 a7a6
info depth 14 score cp 37 time 921 nodes 11012733 nps 11957364 tbhits 0 pv d1b3 d8c7 g1f3 b8c6 c3d5 f6d5 b3d5 e7e6 d5b3 f8b4 c1d2 b4d2
info depth 15 score cp 23 time 1487 nodes 18706992 nps 12580357 tbhits 0 pv d1b3 d8c7 c3d5 f6d5 b3d5 e7e6 f1b5 b8c6 d5c4 f8d6 b5c6 c7c6
info depth 16 score cp 33 time 1969 nodes 25289482 nps 12843820 tbhits 0 pv d1b3 d8c7 c3d5 f6d5 b3d5 e7e6 f1b5 b8c6 d5c4 a8c8 g1f3 f8d6
info depth 17 score cp 44 time 3241 nodes 42990064 nps 13264444 tbhits 0 pv d1b3 d8c7 c3d5 f6d5 b3d5 e7e6 d5b3 b8c6 g1f3 f5g4 f3d2 f8b4
info depth 18 score cp 33 time 6517 nodes 86849078 nps 13326542 tbhits 0 pv d1b3 d8c7 c3d5 f6d5 b3d5 e7e6 d5b3 b8c6 g1f3 f5e4 c1d2 e4f3
info depth 19 score cp 23 time 15124 nodes 208191514 nps 13765638 tbhits 0 pv d1b3 e7e6 b3b7 b8d7 g1f3 f8d6 b7a6 d8e7 f1d3 f5d3 a6d3 e8g8
It was SMP related, I always tune my engine single core, that's why I didn't notice it.
You're not the only one plagued by bugs.
-
- Posts: 7220
- Joined: Mon May 27, 2013 10:31 am
Re: test positions for weak playing engines only
Skipper can't solve the second of Bratko:
[d] 3r1k2/4npp1/1ppr3p/p6P/P2PPPP1/1NR5/5K2/2R5 w - -
Stockfish 6:
Skipper:
[d] 3r1k2/4npp1/1ppr3p/p6P/P2PPPP1/1NR5/5K2/2R5 w - -
Stockfish 6:
Code: Select all
28 +1.10 42.3M 0:22.42 d5 cxd5 e5!
28 +0.91 33.2M 0:17.55 d5 cxd5 e5 R6d7 Nd4 Rb8 f5 Rdd8 Kf3 Re8 Rb3 f6 e6 Rec8 Rxc8+ Nxc8 Nc6?
28 +0.97 32.0M 0:16.92 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 Rc7 Rbc8 e6 Rxc7 Rxc7 Rc8?
27 +1.03 28.0M 0:14.83 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 Kd3 Rdc8 Rc7 Rxc7 Rxc7 Rc8 Rxc8+ Nxc8 Nb5 g6 hxg6 fxg6 Kd4 Ne7 Nc3 h5 gxh5 gxh5 Nxd5 Nxd5 Kxd5 Kf7 f5 Ke7
27 +0.98 26.3M 0:13.92 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8?
27 +1.04 25.3M 0:13.45 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 Kd3 Rdc8 Rc7 Rxc7 Rxc7 Rc8 Rxc8+ Nxc8 Nb5 g6 hxg6 fxg6 Nd6 Nxd6?
26 +1.10 24.7M 0:13.12 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 Kd3 Rdc8 Rc7 Rxc7 Rxc7 Rc8 Rxc8+
26 +1.05 23.2M 0:12.28 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 Kd3 Rdc8 Rxc8+ Rxc8 Rxc8+ Nxc8 Nb5 g6 Nc3 gxh5 gxh5 f6 exf6 Na7 Kd4 Kf7 Kxd5 Kxf6 Ke4 Kf7 Kd5?
25 +1.11 20.1M 0:10.67 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 Kd3 Rbc8 Rxc8 Rxc8 Rxc8+ Nxc8 Nb5 Ne7 Kd4 g6 Nc3 gxh5 gxh5 Ke8!
24 +1.05 12.6M 0:06.63 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 Kd3 Rbc8 Rxc8 Rxc8 Rxc8+ Nxc8 Nb5 Ne7 Kd4 g6 Nc3 gxh5 gxh5 Ke8 Nxd5 Nxd5 Kxd5 Kd7
24 +1.04 11.2M 0:05.86 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 Kd3 Kg8 Rc7 Rbc8 R1c2 Kf8 Rxc8 Rxc8 Rxc8+ Nxc8 Nb5 g6 hxg6 fxg6 Kd4 Ne7!
23 +0.98 7.9M 0:04.13 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 R3c2 Rbc8 Rxc8 Rxc8 Rxc8+ Nxc8 Nb5 Ke7 Kd4 Kd7 Kxd5 Ne7+ Ke4 f6 Nd6 fxe5 fxe5 g5 Nf7 Ke6 Nxh6 Nc6 Nf5 Nxe5
22 +0.92 6.7M 0:03.52 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 R3c2 Kg8 f5 Re8 Rc7 Kf8 Rb1 Rbc8 Rxc8 Nxc8 Nc6 Ne7 Nxe7 Rxe7 Kd4 Rc7 Kxd5 Rc5+ Kd4
22 +0.88 6.4M 0:03.33 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 R3c2 Rbc8 Nb5 Rxc2 Rxc2 d4+ Nxd4 Nd5+ Ke4 Nb4 Rd2 Kg8 f5?
22 +1.01 5.9M 0:03.06 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 R3c2 Rdc8 Nb5 Rxc2 Rxc2 Ke8 Rc7 Kd8 Kd4 Rc8 Rxc8+ Kxc8 Nd6+ Kd7!
21 +0.94 3.9M 0:02.03 d5 cxd5 e5 R6d7 Nd4 Rb8 Ke3 Rdd8 R3c2 Rdc8 Nb5 Rxc2 Rxc2 Ke8 Rc7 Kd8 Kd4 Rc8 Rxc8+ Kxc8 Nd6+ Kd7 Nxf7 Ke6 Nd8+ Kd7 Nb7 Kc6
20 +0.85 3.3M 0:01.70 d5 cxd5 e5 R6d7 Nd4 Re8 Ke3 f6 e6 Rdd8 Rc7 Rc8 Rxc8 Rxc8 Rxc8+ Nxc8 Nf5 Ne7 Nxe7 Kxe7 f5 b5 axb5
19 +0.91 1.9M 0:01.00 d5 cxd5 e5 R6d7 Nd4 Rb7 Ke3 Rbb8 R3c2 Rdc8 Rxc8+ Rxc8 Rxc8+ Nxc8 Nc6 Ne7 Nxe7
18 +0.87 1.3M 0:00.69 d5 cxd5 e5 R6d7 Nd4 Re8 Ke3 f6 Rb1 Rc8 Rxc8+ Nxc8 Rc1 Ne7 e6 Rd8 Rb1 Rc8 Rxb6 Rc3+ Kf2 Rc4 Rb8+ Rc8 Rxc8+ Nxc8
17 +0.87 1.3M 0:00.66 d5 cxd5 e5 R6d7 Nd4 Re8 Ke3 f6 Rb1 Rc8 Rxc8+ Nxc8 Rc1 Ne7 e6 Rd8 Rb1 Rc8 Rxb6 Rc3+ Kf2 Rc4 Rb8+ Rc8 Rxc8+ Nxc8
16 +0.93 664926 0:00.36 d5 cxd5 e5 R6d7 Nd4 Re8 Ke3 Rdd8 Rc7 Rc8 Rxc8 Rxc8 Rxc8+ Nxc8 Nc6 Ne7 Nxe7 Kxe7 Kd4 Ke6 f5+ Ke7 Kxd5
15 +0.87 290966 0:00.17 d5 cxd5 e5 R6d7 Nd4 Re8 Kf3 Rdd8 Rb1 Rb8 Rc7 Rec8 Rxc8+ Rxc8 Rxb6 Kg8 Nb5 Rc4 Rb8+ Kh7 Rf8
14 +0.80 196303 0:00.13 d5 cxd5 e5 R6d7 Nd4 Re8 Rb1 Rb7 Kf3 Rc8 Rxc8+ Nxc8 Rc1 Ne7 Ke3 Rb8 Rb1 f6 exf6 gxf6
13 +0.84 132881 0:00.10 d5 cxd5 e5 R6d7 Nd4 Re8 Rb1 Rb7 Ke3 Rbb8 Rc7 Rec8 Rxc8+ Rxc8 Rxb6 Rc3+ Kd2 Rc4 Rb8+ Nc8 Ne2
12 +0.89 58965 0:00.05 d5 cxd5 e5 R6d7 Nd4 Rb7 f5 Re8 Kf3 Rbb8 Rb1 Kg8 Kf4 Kh7 e6 fxe6 fxe6
11 +0.80 28566 0:00.03 d5 cxd5 e5 R6d7 Nd4 Rb7 f5 Re8 f6 gxf6 exf6 Ng8 Rc6 Rbb8
10 +0.91 17386 0:00.02 d5 cxd5 e5 R6d7 Nd4 Rb7 f5 Re8 Rb3 Rc8 Rxc8+ Nxc8 Rc3
9 +0.98 9410 0:00.02 d5 cxd5 e5 R6d7 Nd4 Kg8 f5 Rb7 f6
8 +0.90 6111 0:00.00 d5 cxd5 e5 R6d7 Nd4 Kg8 Rb3 Rb8
7 +0.86 4910 0:00.00 d5 cxd5 e5 R6d7 Nd4 Kg8 Rb1
6 +0.83 4187 0:00.00 d5 cxd5 e5 R6d7 Nd4 Kg8
5 +0.51 2094 0:00.00 g5 R6d7 Rg1 hxg5 fxg5
4 +0.53 1344 0:00.00 g5 R6d7 Ke2 Kg8
3 +0.51 964 0:00.00 g5 R6d7 gxh6 gxh6
2 +0.55 469 0:00.00 g5 R6d7
1 +0.85 69 0:00.00 e5 Rxd4 Nxd4 Rxd4
0 #
Code: Select all
Depth Value Time(seconds) Nodes
18 -0.09 9.62 559265 f2e3
17 -0.08 5.97 341768 f2e3
16 -0.09 3.92 232516 f2e3
15 -0.08 2.43 140494 f2e3
14 -0.09 1.84 107010 f2e3
13 -0.08 1.54 87297 f2e3
12 -0.09 0.83 42127 f2e3
11 -0.08 0.63 31872 f2e3
10 -0.09 0.30 14984 f2e3
9 -0.08 0.17 8466 f2e3
8 -0.09 0.14 6654 f2e3
7 -0.08 0.10 4296 f2e3
6 -0.08 0.07 2625 f2e3
5 -0.08 0.06 2028 f2e3
4 -0.09 0.04 904 f2e3
3 -0.09 0.02 278 f2e3
2 -0.09 0.02 206 f2e3
1 -0.09 0.02 76 f2e3
-
- Posts: 1563
- Joined: Thu Jul 16, 2009 10:47 am
- Location: Almere, The Netherlands
Re: test positions for weak playing engines only
It is a long time ago that I tried Bratko, Nightmare has no problems in finding this one.
It probably is an evaluation thingy that Skipper can't find it, you temporary sacrifice a pawn to get a positional advantage.
This doesn't have to be a bug, it can be some missing features in your evaluation function.
Code: Select all
FEN: 3r1k2/4npp1/1ppr3p/p6P/P2PPPP1/1NR5/5K2/2R5 w - - 0 1
Nightmare MP:
1 00:00 54 54 +0.50 Kf2-g2
2 00:00 217 217 +0.44 Kf2-g2 Kf8-g8
3 00:00 1k 1k +0.40 Kf2-g2 Rd6-e6 Rc1-e1
4 00:00 4k 2,073k +0.30 g4-g5 Rd6-e6 Rc1-e1 h6xg5
5 00:00 10k 2,609k +0.36 Kf2-g3 Kf8-g8 Rc1-c2 Rd6-e6 Rc2-e2
6 00:00 27k 3,050k +0.46 d4-d5 c6xd5 e4-e5 Rd6-d7 Nb3-d4 Kf8-g8
7 00:00 40k 3,663k +0.58 d4-d5 c6xd5 e4-e5 Rd6-d7 Nb3-d4 Rd8-e8 Kf2-g2
8 00:00 52k 3,969k +0.54 d4-d5 c6xd5 e4-e5 Rd6-d7 Nb3-d4 Rd8-e8 Rc3-c7 Rd7xc7
9 00:00 75k 4,391k +0.61 d4-d5 c6xd5 e4-e5 Rd6-d7 Nb3-d4 Rd8-e8 Kf2-g3 Rd7-b7 Kg3-g2
10 00:00 106k 5,292k +0.56 d4-d5 c6xd5 e4-e5 Rd6-d7 Nb3-d4 Rd8-e8 Kf2-g3 Rd7-b7 Rc3-c7 Rb7xc7
11 00:00 167k 6,411k +0.56 d4-d5 c6xd5 e4-e5 Rd6-d7 Nb3-d4 Rd8-e8 Kf2-g3 Re8-a8 Rc1-c2 Ra8-e8 Rc3-c7
12 00:00 274k 7,599k +0.56 d4-d5 c6xd5 e4-e5 Rd6-d7 Nb3-d4 Rd8-e8 Kf2-g3 Rd7-a7 e5-e6 Kf8-g8 f4-f5 Ra7-a8
13 00:00 572k 9,703k +0.55 d4-d5 c6xd5 e4-e5 Rd6-d7 Nb3-d4 Rd7-b7 Kf2-f3 Kf8-g8 Rc3-c7 Rd8-d7 Rc7xb7 Rd7xb7
14 00:00 1,946k 13,513k +0.54 d4-d5 c6xd5 e4-e5 Rd6-d7 Nb3-d4 Rd8-e8 Kf2-f3 Rd7-d8 Rc3-c7 Rd8-c8 Rc1-c2 Rc8xc7
15 00:00 2,760k 13,800k +0.60 d4-d5 c6xd5 e4-e5 Rd6-d7 Nb3-d4 Rd8-a8 Kf2-g3 Rd7-d8 Rc3-b3 Ra8-c8 Rc1xc8 Ne7xc8
16 00:00 4,852k 14,660k +0.59 d4-d5 c6xd5 e4-e5 Rd6-d7 Nb3-d4 Rd8-e8 Nd4-b5 Re8-a8 Nb5-d6 Kf8-g8 Rc3-c7 Ra8-a7
17 00:00 8,713k 15,368k +0.70 d4-d5 c6xd5 e4-e5 Rd6-d7 Nb3-d4 Rd8-e8 Nd4-b5 Re8-a8 Nb5-d6 Kf8-g8 Rc3-c7 Ra8-a7
This doesn't have to be a bug, it can be some missing features in your evaluation function.