Some of the top programs find this almost instantly (Rybka, Toga, Doch) while others do not. What makes some programs find this so fast?
[D]2r2k2/r3bp2/1n2p1p1/p1p1P1Pp/PpP1NP1P/1P1RBK2/8/3R4 w - - bm Nxc5;
Test position Nxc5
Moderators: hgm, Rebel, chrisw
-
AdminX
- Posts: 6340
- Joined: Mon Mar 13, 2006 2:34 pm
- Location: Acworth, GA
Re: Test position Nxc5
[Event "?"]
[Site "?"]
[Date "?"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[FEN "2r2k2/r3bp2/1n2p1p1/p1p1P1Pp/PpP1NP1P/1P1RBK2/8/3R4 w - - 0 1"]
[SetUp "1"]
1. Nxc5 Rxc5 2. Rd8+ Kg7 3. Rb8 Rd7 4. Rxd7 Nxd7 5. Rb7 Rc6 6. Rxd7 Kf8 7. Ke4
Ke8 8. Rd1 Rc7 9. Rc1 Rc6 10. Kf3 Kd7 11. Rd1+ Ke8 12. Bd4 Rc7 13. Rd2 Rd7 14.
Ke4 Bc5 15. Rd3 Bxd4 16. Rxd4 {(0:00:11) 15977kN}{[%eval 173,19,Doch]} *
[Site "?"]
[Date "?"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[FEN "2r2k2/r3bp2/1n2p1p1/p1p1P1Pp/PpP1NP1P/1P1RBK2/8/3R4 w - - 0 1"]
[SetUp "1"]
1. Nxc5 Rxc5 2. Rd8+ Kg7 3. Rb8 Rd7 4. Rxd7 Nxd7 5. Rb7 Rc6 6. Rxd7 Kf8 7. Ke4
Ke8 8. Rd1 Rc7 9. Rc1 Rc6 10. Kf3 Kd7 11. Rd1+ Ke8 12. Bd4 Rc7 13. Rd2 Rd7 14.
Ke4 Bc5 15. Rd3 Bxd4 16. Rxd4 {(0:00:11) 15977kN}{[%eval 173,19,Doch]} *
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers
__________________________________________________________________
Ted Summers
-
zullil
- Posts: 6442
- Joined: Tue Jan 09, 2007 12:31 am
- Location: PA USA
- Full name: Louis Zulli
Re: Test position Nxc5
Crafty-23.1 finds Nxc5, but not quickly.
Code: Select all
Crafty v23.1 (8 cpus)
White(1): book off
book file disabled.
White(1): ponder off
pondering disabled.
White(1): setboard 2r2k2/r3bp2/1n2p1p1/p1p1P1Pp/PpP1NP1P/1P1RBK2/8/3R4 w - - bm Nxc5
White(1): st 300
search time set to 300.00.
White(1): display stats
display statistics at end of each search.
White(1): display
+---+---+---+---+---+---+---+---+
8 | | . |<R>| . | |<K>| | . |
+---+---+---+---+---+---+---+---+
7 |<R>| | . | |<B>|<P>| . | |
+---+---+---+---+---+---+---+---+
6 | |<N>| | . |<P>| . |<P>| . |
+---+---+---+---+---+---+---+---+
5 |<P>| |<P>| |-P-| |-P-|<P>|
+---+---+---+---+---+---+---+---+
4 |-P-|<P>|-P-| . |-N-|-P-| |-P-|
+---+---+---+---+---+---+---+---+
3 | . |-P-| . |-R-|-B-|-K-| . | |
+---+---+---+---+---+---+---+---+
2 | | . | | . | | . | | . |
+---+---+---+---+---+---+---+---+
1 | . | | . |-R-| . | | . | |
+---+---+---+---+---+---+---+---+
a b c d e f g h
White(1): go
time limit 5:00 (+0.00) (5:00)
depth time score variation (1)
starting thread 1
starting thread 2
starting thread 3
starting thread 4
starting thread 5
starting thread 6
starting thread 7
10 0.03 1.36 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Nf6 Ra8 5. Bf2 Rh8
10-> 0.03 1.36 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Nf6 Ra8 5. Bf2 Rh8
11 0.04 1.30 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Nf6 Rc6 5. Bf2 Rc8 6. Kf3
11-> 0.05 1.30 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Nf6 Rc6 5. Bf2 Rc8 6. Kf3
12 0.06 1.37 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Nf6 Rc8 5. Bf2 Raa8 6. Kf3 Rh8
12-> 0.07 1.37 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Nf6 Rc8 5. Bf2 Raa8 6. Kf3 Rh8
13 0.09 1.30 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Rc6 5. Bf2 Ra8 6. Nf6 Rc7 7.
Be3
13-> 0.12 1.30 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Rc6 5. Bf2 Ra8 6. Nf6 Rc7 7.
Be3
14 0.18 1.27 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rh8 6. Nf6 Ra7 7.
Bg3 Raa8
14-> 0.24 1.27 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rh8 6. Nf6 Ra7 7.
Bg3 Raa8
15 0.28 1.27 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rh8 6. Nf6 Ra8 7.
Be3 Rf8 8. Ne4
15-> 0.36 1.27 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rh8 6. Nf6 Ra8 7.
Be3 Rf8 8. Ne4
16 0.45 1.29 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Rh8 7.
Be3 Ra8 8. Kg2 Rh8
16-> 0.51 1.29 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Rh8 7.
Be3 Ra8 8. Kg2 Rh8
17 0.60 1.29 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Rh8 7.
Be3 Ra8 8. Ne4 Rh8 9. R3d2
17-> 0.81 1.29 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Rh8 7.
Be3 Ra8 8. Ne4 Rh8 9. R3d2 (s=4)
18 0.99 1.30 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rb8 9. Rd3 Rh8 (s=3)
18-> 1.14 1.30 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rb8 9. Rd3 Rh8
19 1.73 1.29 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rb8 9. Rd3 Rh8 10.
Kf3
19-> 1.95 1.29 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rb8 9. Rd3 Rh8 10.
Kf3
20 3.01 1.27 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rc6 9. Be3 Rc7 10.
Ne4 Rcc8
20-> 4.90 1.27 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rc6 9. Be3 Rc7 10.
Ne4 Rcc8
21 5.73 1.28 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rhc8 9. Rd3 Rh8 10.
Be3 Rcc8 11. Kf3
21-> 6.67 1.28 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rhc8 9. Rd3 Rh8 10.
Be3 Rcc8 11. Kf3 (s=2)
22 8.32 1.29 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rhc8 9. Rd3 Rh8 10.
Be3 Rb8 11. Kg2 Rh8
22-> 10.48 1.29 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rhc8 9. Rd3 Rh8 10.
Be3 Rb8 11. Kg2 Rh8
23 12.80 1.29 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rhc8 9. Rd3 Rh8 10.
Be3 Rb8 11. Kg2 Rh8 12. Kf3
23-> 16.48 1.29 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rhc8 9. Rd3 Rh8 10.
Be3 Rb8 11. Kg2 Rh8 12. Kf3 (s=3)
24 23.45 1.27 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rhc8 9. Rd3 Rh8 10.
Be3 Rb7 11. R3d2 Rbb8 12. Ne4 Rbc8
(s=2)
24-> 32.68 1.27 1. R3d2 Rc6 2. Kg2 Kg7 3. Rd3 Rcc7
4. Kf3 Ra8 5. Bf2 Rac8 6. Nf6 Ra8 7.
R3d2 Rh8 8. Kg3 Rhc8 9. Rd3 Rh8 10.
Be3 Rb7 11. R3d2 Rbb8 12. Ne4 Rbc8
25 1:36 1.28 1. R3d2 Kg7 2. Nd6 Rc6 3. Ne8+ Kf8
4. Nf6 Kg7 5. Bf2 Rc8 6. Ne4 Raa8 7.
Nf6 Ra7 <HT>
25-> 1:55 1.28 1. R3d2 Kg7 2. Nd6 Rc6 3. Ne8+ Kf8
4. Nf6 Kg7 5. Bf2 Rc8 6. Ne4 Raa8 7.
Nf6 Ra7 <HT>
26 2:06 1.28 1. R3d2 Kg7 2. Nd6 Rc6 3. Ne8+ Kf8
4. Nf6 Kg7 5. Bf2 Rc8 6. Ne4 Raa8 7.
Nf6 Ra7 <HT>
26-> 2:50 1.28 1. R3d2 Kg7 2. Nd6 Rc6 3. Ne8+ Kf8
4. Nf6 Kg7 5. Bf2 Rc8 6. Ne4 Raa8 7.
Nf6 Ra7 <HT> (s=3)
27 3:03 1.28 1. R3d2 Kg7 2. Nd6 Rc6 3. Ne8+ Kf8
4. Nf6 Kg7 5. Bf2 Rc8 6. Ne4 Raa8 7.
Nf6 Ra7 <HT> (s=2)
27 3:44 +1 1. Nxc5!
27 4:22 2.10 1. Nxc5 Rxc5 2. Rd8+ Kg7 3. Rb8 Nd7
4. Re8 Rcc7 5. Rxe7 Nxe5+ 6. fxe5 Rxe7
7. Bxa7 Rxa7 8. Rd8 Ra6 9. Ke4 Ra7
10. Kd4 Rc7 11. c5 Kh7 12. Ra8 Rd7+
13. Kc4 Rd5 14. Rxa5 Rxe5 15. Kxb4
Re4+ 16. Kc3 Rxh4
27-> 4:52 2.10 1. Nxc5 Rxc5 2. Rd8+ Kg7 3. Rb8 Nd7
4. Re8 Rcc7 5. Rxe7 Nxe5+ 6. fxe5 Rxe7
7. Bxa7 Rxa7 8. Rd8 Ra6 9. Ke4 Ra7
10. Kd4 Rc7 11. c5 Kh7 12. Ra8 Rd7+
13. Kc4 Rd5 14. Rxa5 Rxe5 15. Kxb4
Re4+ 16. Kc3 Rxh4 (s=2)
time=5:00 mat=0 n=6743460998 fh=92% nps=22.5M
extensions=263.7M qchecks=212.4M reduced=1003.6M pruned=2288.9M
predicted=0 evals=2565.0M 50move=0 EGTBprobes=0 hits=0
SMP-> splits=1032798 aborts=92928 data=62/512 elap=5:00
terminating SMP processes.
White(1): Nxc5
time used: 5:00
-
michiguel
- Posts: 6401
- Joined: Thu Mar 09, 2006 8:30 pm
- Location: Chicago, Illinois, USA
Re: Test position Nxc5
Excellent position. It is not tactical, it is mainly positional. The engines that understand how to transition to a won rook endgame (based on king activity, rook activity and passed pawn) will see this quickly. Gaviota does not understand this.Howard E wrote:Some of the top programs find this almost instantly (Rybka, Toga, Doch) while others do not. What makes some programs find this so fast?
[D]2r2k2/r3bp2/1n2p1p1/p1p1P1Pp/PpP1NP1P/1P1RBK2/8/3R4 w - - bm Nxc5;
The key here, after a while of searching the position after Nxc5 Gaviota sees
Code: Select all
1486955638 16: 2151.6 -0.96 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7 4.Re8 Rcc7
5.Bxa7 Rxa7 6.Rxe7 Nxe5+ 7.Ke4 Rxe7
8.Kxe5 Rc7 9.Rd2 Kf8 10.Kd6
Code: Select all
212333481 14 319.8 +1.34 1.Ke2 Rac7 2.Bf2 Ke8 3.R3d2 Rb8 4.Be3
Rbc8 5.Kf3 Kf8 6.Rc2 Ra8 7.Rg2 Ke8Miguel
-
Eelco de Groot
- Posts: 4565
- Joined: Sun Mar 12, 2006 2:40 am
- Full name:
Re: Test position Nxc5
I don't know about "instantly", could that be SMP search luck? On my computer at least Toga needs almost a minute with one thread. I just tried a random Toga version of mine.
Rainbow Serpent could only solve the position with throwing in a lot of Ancalagon theory, the Null window searches in Stockfish and Rainbow Serpent just are not going to find it, there is too much pruning somewhere, maybe Null move pruning, which I use just iteratively so that might be it. But if I use fairly short PV-searches as well in root_search instead of, or in combination with, the standard Null window search, at least I avoid any Zugzwang or other heavy pruning, and now I have a Rainbow Serpent that finds it but it is a lot like Ancalagon again
Very complex root_search()
So assuming this is not just search luck by grafting etc. I just come back to the fact that the different types of searches can sometimes be combined for overcoming blind spots, and this is the main reason, in my eyes, for using a PV search that avoids 'dangerous' pruning techniques, it loses efficiency but gains accuracy which is what you need most in complex board situations.
This related to a discussion on the Programming and Technical Discussion Forum where I have not yet replied because the matter is not so clear to myself either - But I do have some examples, and maybe this position is one. But Miguel Ballicora interprets the problem quite differently, which probably just shows there are no easy answers, also not to Howard Exners's question, why some programs find this and can avoid the blind spots of typical Null window searches here, maybe helped by their eval...
Eelco
[D]2r2k2/r3bp2/1n2p1p1/p1p1P1Pp/PpP1NP1P/1P1RBK2/8/3R4 w - -
Engine: Rainbow Serpent 1.5 Cubic Build 074 (256 MB)
by Romstad, Costalba, Kiiski, De Groot
1.00 0:00 +1.69 1.R3d2 (88) 0
2.00 0:00 +1.57 1.Nf6 Kg7 2.R3d2 Rh8 3.Kg2 (887) 2
2.00 0:00 +1.65 1.R3d2 Kg7 2.Kg2 Rcc7 (8.658) 20
2.00 0:00 +1.77 1.Nd6 Bxd6 2.Rxd6 Rb7 3.R1d2 (119.168) 190
3.00 0:00 +1.29 1.Nd6 Bxd6 2.exd6 Ke8 3.Ke4 Kd7
4.R3d2 Raa8 5.Re2 (135.135) 205
3.00 0:00 +1.57 1.R3d2 Kg7 2.Nf6 Rh8 3.Kg2 (143.360) 213
3.04 0:01 +1.61 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Ra8 5.Kg3 Rd8 6.f5 exf5
7.Nxf5+ (431.724) 383
4.01 0:02 +1.65 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Rd8 5.f5 exf5 6.Nxf5+ (991.673) 488
5.01 0:02 +1.65 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Rd8 5.f5 exf5 6.Nxf5+ (1.447.189) 517
6.01 0:03 +1.65 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Rd8 5.f5 exf5 6.Nxf5+ (1.854.061) 536
7.01 0:06 +1.57 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Ra8 5.Ne4 Rcc8 6.Rb2 Rc7 7.Nd6 Nd7
8.Kg3 (3.671.808) 577
8.01 0:11 +1.57 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Ra8 5.Ne4 Rcc8 6.Rb2 Rc6 7.Nd6 Bxd6
8.Rxd6 Rxd6 9.exd6 Nd7 (7.119.114) 608
9.01 0:23 +1.49 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Ra8 5.Ne4 Rcc8 6.Rb2 Rc6 7.Kg3 Rh8
8.Rbd2 Ra8 9.Rh2 Rc7 10.Nd6 Rd8
11.Rg2 Rc6 12.Rgd2 Ra8 13.Ne4 Rc7
14.Kf3 (14.782.201) 624
9.20 0:47 +1.77++ {8-)
}1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Na8
4.Rd6 Bxd6 5.exd6 (30.530.901) 648
10.01 0:48 +1.89 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Bxa7 Rxa7 6.Rxe7 Nxe5+
7.Ke4 Rxe7 8.fxe5 Re8 9.Rd6 Ra8
10.c5 Kf8 11.c6 Rc8 12.Ke3 (31.322.406) 649
11.01 1:11 +1.89 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Bxa7 Rxa7 6.Rxe7 Nxe5+
7.Ke4 Rxe7 8.fxe5 Re8 9.Rd6 Ra8
10.c5 Kf8 11.c6 Rc8 12.Ke3 (47.368.240) 662
12.01 1:32 +2.10++ 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Bxa7 Rxa7 6.Rxe7 Nxe5+
7.fxe5 Rxe7 8.Rd8 Rb7 9.Ke4 Ra7
10.Kd4 Ra6 11.c5 Rc6 12.Ra8 (63.010.915) 680
13.01 1:54 +2.90++ 1.Nxc5 (78.566.577) 684
14.01 2:05 +2.86 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Bxa7 Rxa7 6.Rxe7 Nxe5+
7.fxe5 Rxe7 8.Rd8 Rb7 9.Ke3 Rc7
10.Ke4 Ra7 11.Kd4 Rc7 12.c5 Kh7
13.Ra8 Rd7+ 14.Ke4 (86.790.080) 688
15.01 2:24 +2.86 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Bxa7 Rxa7 6.Rxe7 Nxe5+
7.fxe5 Rxe7 8.Rd8 Rb7 9.Ke3 Rc7
10.Ke4 Ra7 11.Kd4 Rc7 12.c5 Kh7
13.Ra8 Rd7+ 14.Ke4 (103.633.241) 716
16.01 3:12 +2.78 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Rd7
4.Rxd7 Nxd7 5.Rb7 Kf8 6.Rxd7 Rc6
7.Ke4 Ke8 8.Rd3 Kf8 9.Rd2 Ke8 10.Rd1 Kf8
11.Bd4 Ke8 12.Kd3 Bd8 13.c5 Kd7
14.Kc4 (143.319.244) 743
best move: Ne4xc5 time: 5:05.969 min n/s: 792.475 nodes: 242.460.320
Rainbow Serpent could only solve the position with throwing in a lot of Ancalagon theory, the Null window searches in Stockfish and Rainbow Serpent just are not going to find it, there is too much pruning somewhere, maybe Null move pruning, which I use just iteratively so that might be it. But if I use fairly short PV-searches as well in root_search instead of, or in combination with, the standard Null window search, at least I avoid any Zugzwang or other heavy pruning, and now I have a Rainbow Serpent that finds it but it is a lot like Ancalagon again
Very complex root_search()So assuming this is not just search luck by grafting etc. I just come back to the fact that the different types of searches can sometimes be combined for overcoming blind spots, and this is the main reason, in my eyes, for using a PV search that avoids 'dangerous' pruning techniques, it loses efficiency but gains accuracy which is what you need most in complex board situations.
This related to a discussion on the Programming and Technical Discussion Forum where I have not yet replied because the matter is not so clear to myself either - But I do have some examples, and maybe this position is one. But Miguel Ballicora interprets the problem quite differently, which probably just shows there are no easy answers, also not to Howard Exners's question, why some programs find this and can avoid the blind spots of typical Null window searches here, maybe helped by their eval...
Eelco
[D]2r2k2/r3bp2/1n2p1p1/p1p1P1Pp/PpP1NP1P/1P1RBK2/8/3R4 w - -
Engine: Rainbow Serpent 1.5 Cubic Build 074 (256 MB)
by Romstad, Costalba, Kiiski, De Groot
1.00 0:00 +1.69 1.R3d2 (88) 0
2.00 0:00 +1.57 1.Nf6 Kg7 2.R3d2 Rh8 3.Kg2 (887) 2
2.00 0:00 +1.65 1.R3d2 Kg7 2.Kg2 Rcc7 (8.658) 20
2.00 0:00 +1.77 1.Nd6 Bxd6 2.Rxd6 Rb7 3.R1d2 (119.168) 190
3.00 0:00 +1.29 1.Nd6 Bxd6 2.exd6 Ke8 3.Ke4 Kd7
4.R3d2 Raa8 5.Re2 (135.135) 205
3.00 0:00 +1.57 1.R3d2 Kg7 2.Nf6 Rh8 3.Kg2 (143.360) 213
3.04 0:01 +1.61 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Ra8 5.Kg3 Rd8 6.f5 exf5
7.Nxf5+ (431.724) 383
4.01 0:02 +1.65 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Rd8 5.f5 exf5 6.Nxf5+ (991.673) 488
5.01 0:02 +1.65 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Rd8 5.f5 exf5 6.Nxf5+ (1.447.189) 517
6.01 0:03 +1.65 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Rd8 5.f5 exf5 6.Nxf5+ (1.854.061) 536
7.01 0:06 +1.57 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Ra8 5.Ne4 Rcc8 6.Rb2 Rc7 7.Nd6 Nd7
8.Kg3 (3.671.808) 577
8.01 0:11 +1.57 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Ra8 5.Ne4 Rcc8 6.Rb2 Rc6 7.Nd6 Bxd6
8.Rxd6 Rxd6 9.exd6 Nd7 (7.119.114) 608
9.01 0:23 +1.49 1.Kg3 Kg7 2.R3d2 Rac7 3.Nd6 Rh8
4.Kf3 Ra8 5.Ne4 Rcc8 6.Rb2 Rc6 7.Kg3 Rh8
8.Rbd2 Ra8 9.Rh2 Rc7 10.Nd6 Rd8
11.Rg2 Rc6 12.Rgd2 Ra8 13.Ne4 Rc7
14.Kf3 (14.782.201) 624
9.20 0:47 +1.77++ {8-)
}1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Na8 4.Rd6 Bxd6 5.exd6 (30.530.901) 648
10.01 0:48 +1.89 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Bxa7 Rxa7 6.Rxe7 Nxe5+
7.Ke4 Rxe7 8.fxe5 Re8 9.Rd6 Ra8
10.c5 Kf8 11.c6 Rc8 12.Ke3 (31.322.406) 649
11.01 1:11 +1.89 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Bxa7 Rxa7 6.Rxe7 Nxe5+
7.Ke4 Rxe7 8.fxe5 Re8 9.Rd6 Ra8
10.c5 Kf8 11.c6 Rc8 12.Ke3 (47.368.240) 662
12.01 1:32 +2.10++ 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Bxa7 Rxa7 6.Rxe7 Nxe5+
7.fxe5 Rxe7 8.Rd8 Rb7 9.Ke4 Ra7
10.Kd4 Ra6 11.c5 Rc6 12.Ra8 (63.010.915) 680
13.01 1:54 +2.90++ 1.Nxc5 (78.566.577) 684
14.01 2:05 +2.86 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Bxa7 Rxa7 6.Rxe7 Nxe5+
7.fxe5 Rxe7 8.Rd8 Rb7 9.Ke3 Rc7
10.Ke4 Ra7 11.Kd4 Rc7 12.c5 Kh7
13.Ra8 Rd7+ 14.Ke4 (86.790.080) 688
15.01 2:24 +2.86 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Bxa7 Rxa7 6.Rxe7 Nxe5+
7.fxe5 Rxe7 8.Rd8 Rb7 9.Ke3 Rc7
10.Ke4 Ra7 11.Kd4 Rc7 12.c5 Kh7
13.Ra8 Rd7+ 14.Ke4 (103.633.241) 716
16.01 3:12 +2.78 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Rd7
4.Rxd7 Nxd7 5.Rb7 Kf8 6.Rxd7 Rc6
7.Ke4 Ke8 8.Rd3 Kf8 9.Rd2 Ke8 10.Rd1 Kf8
11.Bd4 Ke8 12.Kd3 Bd8 13.c5 Kd7
14.Kc4 (143.319.244) 743
best move: Ne4xc5 time: 5:05.969 min n/s: 792.475 nodes: 242.460.320
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan
-
Howard E
- Posts: 261
- Joined: Wed Mar 08, 2006 8:49 pm
Re: Test position Nxc5
I thought I was using 1 core for Toga but was mistaken . Turns out it was the 8 thread version.
Toga141se-8cpu:
10/18 00:00 97.312 0 +0.84 Nd6 Bxd6 exd6 Ke8 Bc1 Kd7 Bb2 Kc6 Kf2 Rd8
10/20 00:00 107.429 0 +0.91 Kf2 Rac7 Kg2 Ra8 Kf3 Rb8 Ke2 Kg7
11/23 00:00 143.780 0 +0.89 Nd6 Rc6 Nb5 Ra8 Kg3 Kg8 Kg2 Kg7 R3d2 Rh8
12/21 00:00 191.406 0 +0.88 Nd6 Rc6 Nb5 Ra8 Kg2 Kg7 Nd6 Kg8 Ne4 Rc7 Bf2
13/28 00:00 393.344 0 +0.86 Nd6 Rc6 Kg2 Rd7 Ne4 Ra7 Bf2 Rc8 Re1 Rc6 Be3 Kg7 Red1
14/22 00:00 610.753 0 +0.85 Ke2 Rc6 Kd2 Kg7 Kc1 Ra8 Kc2 Rac8 Kb2 R6c7 Bf2 Ra8 R3d2 Rh8
15/29 00:01 791.575 0 +0.84 Ke2 Rc6 Kd2 Ra8 Kc2 Kg7 Kc1 Rc7 Kb2 Rc6 Kc2 Rac8 Bd2 Rd8 Rxd8 Bxd8
15/29 00:01 872.220 0 +0.85 Ke2 Rc6 Ke1 Ra8 Kd2 Kg7 Kc1 Rc7 Kb2 Kg8 Bg1 Kg7
16/29 00:01 1.266.877 7.610.840 +0.83 Ke2 Rc6 Ke1 Ra8 Kd2 Kg7 Kc1 Rc7 Kc2 Rcc8 R3d2 Rc7 Rd6 Bxd6 exd6 Rcc8 Bxc5 Ra6 Bd4+ Kf8
17/26 00:02 1.966.779 7.568.886 +0.82 Nd6 Rc6 Ke2 Ra8 Kd2 Kg8 Bf2 Kg7 Kc1 Rc7 Ne4 Rc6 Be3 Rf8 Kb2
17/39 00:03 2.994.423 7.760.000 +1.21 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Bxa7 Rxa7 Rxe7 Nxe5+ fxe5 Rxe7 Rd8 Rc7 Ke4 Ra7 c5 Rc7 Kd4 Kh7 Kc4 Kg7
17/39 00:03 3.080.353 7.760.000 +1.43 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Rxe7 Nxe5+ fxe5 Rxe7 Bxa7 Rxa7 Rd8 Rc7 Ke4 Rc6 Rd7 Rc5 Kd4 Rc6 Ra7 Rc8 Rxa5
18/34 00:03 3.823.008 7.751.958 +1.49 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Rxe7 Nxe5+ fxe5 Rxe7 Bxa7 Rxa7 Rd8 Rb7 Ke4 Rc7 Kd4 Ra7 c5 Rc7 Ra8 Rd7+ Kc4 Rd2 Rxa5
19/39 00:05 5.663.460 7.984.000 +1.47 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Rxe7 Nxe5+ fxe5 Rxe7 Bxa7 Rxa7 Rd8 Rb7 Ke4 Rc7 Kd4 Ra7 c5 Ra6 Rd7 Kf8 Rc7 Ke8
20/37 00:08 8.543.826 8.180.000 +1.47 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Rxe7 Nxe5+ fxe5 Rxe7 Bxa7 Rxa7 Rd8 Rb7 Ke4 Rc7 Kd4 Ra7 c5 Ra6 Rd7 Kf8 Rc7 Ke8 Ke4 Ra8
21/39 00:13 13.812.088 8.221.538 +1.52 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Rxe7 Kf8 Bxa7 Kxe7 Bf2 Rc6 Ke4 Kd8 Bd4 Ke7 Be3 Ra6 Bf2 Rc6 Bd4 Rc8
21/46 00:13 13.964.497 8.221.538 +1.53 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Rxe7 Kf8 Bxa7 Kxe7 Bf2 Rc6 Ke4 Ra6 c5 Rc6 Rc1 Rc7 Bd4 Rc6 Kd3 Ke8 Ke4
22/47 00:24 25.687.059 8.236.667 +1.84 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Rd7 Rxd7 Nxd7 Rb7 Kf8 Rxd7 Rc6 Ke4 Ke8 Ra7 Bd8 c5 Bc7 Kd3 Kd7 Kc4 Kd8 Bd4 Kc8 Kd3 Kd7
Toga141se-8cpu:
10/18 00:00 97.312 0 +0.84 Nd6 Bxd6 exd6 Ke8 Bc1 Kd7 Bb2 Kc6 Kf2 Rd8
10/20 00:00 107.429 0 +0.91 Kf2 Rac7 Kg2 Ra8 Kf3 Rb8 Ke2 Kg7
11/23 00:00 143.780 0 +0.89 Nd6 Rc6 Nb5 Ra8 Kg3 Kg8 Kg2 Kg7 R3d2 Rh8
12/21 00:00 191.406 0 +0.88 Nd6 Rc6 Nb5 Ra8 Kg2 Kg7 Nd6 Kg8 Ne4 Rc7 Bf2
13/28 00:00 393.344 0 +0.86 Nd6 Rc6 Kg2 Rd7 Ne4 Ra7 Bf2 Rc8 Re1 Rc6 Be3 Kg7 Red1
14/22 00:00 610.753 0 +0.85 Ke2 Rc6 Kd2 Kg7 Kc1 Ra8 Kc2 Rac8 Kb2 R6c7 Bf2 Ra8 R3d2 Rh8
15/29 00:01 791.575 0 +0.84 Ke2 Rc6 Kd2 Ra8 Kc2 Kg7 Kc1 Rc7 Kb2 Rc6 Kc2 Rac8 Bd2 Rd8 Rxd8 Bxd8
15/29 00:01 872.220 0 +0.85 Ke2 Rc6 Ke1 Ra8 Kd2 Kg7 Kc1 Rc7 Kb2 Kg8 Bg1 Kg7
16/29 00:01 1.266.877 7.610.840 +0.83 Ke2 Rc6 Ke1 Ra8 Kd2 Kg7 Kc1 Rc7 Kc2 Rcc8 R3d2 Rc7 Rd6 Bxd6 exd6 Rcc8 Bxc5 Ra6 Bd4+ Kf8
17/26 00:02 1.966.779 7.568.886 +0.82 Nd6 Rc6 Ke2 Ra8 Kd2 Kg8 Bf2 Kg7 Kc1 Rc7 Ne4 Rc6 Be3 Rf8 Kb2
17/39 00:03 2.994.423 7.760.000 +1.21 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Bxa7 Rxa7 Rxe7 Nxe5+ fxe5 Rxe7 Rd8 Rc7 Ke4 Ra7 c5 Rc7 Kd4 Kh7 Kc4 Kg7
17/39 00:03 3.080.353 7.760.000 +1.43 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Rxe7 Nxe5+ fxe5 Rxe7 Bxa7 Rxa7 Rd8 Rc7 Ke4 Rc6 Rd7 Rc5 Kd4 Rc6 Ra7 Rc8 Rxa5
18/34 00:03 3.823.008 7.751.958 +1.49 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Rxe7 Nxe5+ fxe5 Rxe7 Bxa7 Rxa7 Rd8 Rb7 Ke4 Rc7 Kd4 Ra7 c5 Rc7 Ra8 Rd7+ Kc4 Rd2 Rxa5
19/39 00:05 5.663.460 7.984.000 +1.47 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Rxe7 Nxe5+ fxe5 Rxe7 Bxa7 Rxa7 Rd8 Rb7 Ke4 Rc7 Kd4 Ra7 c5 Ra6 Rd7 Kf8 Rc7 Ke8
20/37 00:08 8.543.826 8.180.000 +1.47 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Rxe7 Nxe5+ fxe5 Rxe7 Bxa7 Rxa7 Rd8 Rb7 Ke4 Rc7 Kd4 Ra7 c5 Ra6 Rd7 Kf8 Rc7 Ke8 Ke4 Ra8
21/39 00:13 13.812.088 8.221.538 +1.52 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Rxe7 Kf8 Bxa7 Kxe7 Bf2 Rc6 Ke4 Kd8 Bd4 Ke7 Be3 Ra6 Bf2 Rc6 Bd4 Rc8
21/46 00:13 13.964.497 8.221.538 +1.53 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Rxe7 Kf8 Bxa7 Kxe7 Bf2 Rc6 Ke4 Ra6 c5 Rc6 Rc1 Rc7 Bd4 Rc6 Kd3 Ke8 Ke4
22/47 00:24 25.687.059 8.236.667 +1.84 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Rd7 Rxd7 Nxd7 Rb7 Kf8 Rxd7 Rc6 Ke4 Ke8 Ra7 Bd8 c5 Bc7 Kd3 Kd7 Kc4 Kd8 Bd4 Kc8 Kd3 Kd7
-
Eelco de Groot
- Posts: 4565
- Joined: Sun Mar 12, 2006 2:40 am
- Full name:
Re: Test position Nxc5
That seems about what you'd expect Howard for a very fast Core i7. Impressive it is! The total nodes Toga reports is just for a single thread if I'm not mistaken, we could not quite figure out how to improve that on the TDDB 
and your NPS is then just over 8 million after a while, more than ten times as fast as a slow compile on my Athlon with one core only. I ran it in parallel first so the speed was even lower, but now with 100% CPU but a slow compile I get, with 635.241 nodes per second :
2r2k2/r3bp2/1n2p1p1/p1p1P1Pp/PpP1NP1P/1P1RBK2/8/3R4 w - -
Engine: Toga 1.4 Checkov Beta 4 (128 MB)
by Fabien Letouzey, Thomas Gaksch, Eelco de Groot
1/05 0:00 +0.71 1.Ra1 Kg7 (49)
1/05 0:00 +0.73 1.Re1 Kg7 (193)
1/05 0:00 +0.91 1.R3d2 Kg7 (437)
1/11 0:00 +1.03 1.Kg2 Kg7 (2.087)
2/14 0:00 +0.76 1.Kg2 Kg7 2.Bf2 (3.492)
2/14 0:00 +1.02 1.Nd6 Bxd6 2.exd6 Rd7 (3.604)
3/14 0:00 +1.00 1.Nd6 Bxd6 2.exd6 Rd7 3.Re1 Kg7 (6.088)
4/14 0:00 +0.90 1.Nd6 Bxd6 2.Rxd6 Rb7 3.Kg2 Kg7 (10.580)
4/14 0:00 +1.06 1.Kg2 Kg8 2.Bf2 Kg7 (11.587)
5/16 0:00 +1.02 1.Kg2 Kg7 2.Nd6 Rc6 3.Bf2 Rd7 4.Ne8+ Kf8
5.Rxd7 Nxd7 6.Rxd7 Kxe8 (28.655)
6/27 0:00 +0.80 1.Kg2 Kg7 2.Nd6 Rc6 3.Bf2 Bxd6
4.exd6 Rd7 5.R3d2 (64.528)
6/27 0:00 +0.94 1.Nd6 Rc6 2.Nb5 Ra8 3.Kg2 Kg7 (72.530)
7/27 0:00 +0.98 1.Nd6 Rc6 2.Nb5 Ra8 3.Kg2 Kg7 4.Bf2 Rh8 (133.330)
8/28 0:00 +0.76 1.Nd6 Rc6 2.Kg2 Ra8 3.Ne4 Kg7 4.Bg1 Rc7
5.Bf2 (278.992)
8/28 0:00 +0.88 1.Kg2 Rc6 2.Kg3 Ra8 3.Bc1 Kg7 4.Bb2 Rc7 (353.681)
9/29 0:01 +0.82 1.Kg2 Raa8 2.Kg3 Rc7 3.R1d2 Rac8
4.Bf2 Rb8 5.Rd1 Kg7 (1.145.326) 588
10/40 0:04 +0.82 1.Kg2 Raa8 2.Kg3 Kg7 3.Kf2 Rc6 4.Kf1 Kg8
5.Bf2 Kg7 6.Kg2 (2.427.546) 602
11/40 0:11 +0.81 1.Kg2 Raa8 2.Bc1 Rc6 3.Kf2 Rc7 4.Be3 Kg7
5.Kg2 Rc6 6.Kf1 Rc7 (6.778.506) 612
11/40 0:14 +0.92 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Bxa7 Rxa7 6.Rxe7 Nxe5+
7.Ke4 Rxe7 8.fxe5 Rc7 9.Rd8 Rc6 (9.059.087) 612
12/40 0:17 +0.89 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Rxe7 Nxe5+ 6.fxe5 Rxe7
7.Bxa7 Rxa7 8.Ke4 Rc7 9.Rd8 Kh7
10.Kd4 Kg7 (10.494.005) 615
13/40 0:26 +1.23 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Rxe7 Nxe5+ 6.fxe5 Rxe7
7.Bxa7 Rxa7 8.Ke4 Rc7 9.Rd6 Kf8
10.Kd4 Ke7 11.c5 Rc8 (16.547.260) 620
14/40 0:51 +1.34 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Rxe7 Nxe5+ 6.fxe5 Rxe7
7.Bxa7 Rxa7 8.Ke3 Kf8 9.Rd6 Ra8
10.c5 Rc8 11.c6 Rc7 12.Ke4 Ke7 (32.781.858) 633
15/55 1:56 +1.81 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Rxe7 Kf8 6.Bxa7 Kxe7
7.Bf2 Rc6 8.Ke4 Nb8 9.Bd4 Na6 10.Bg1 Rc8
11.Be3 (74.593.349) 640
16/55 4:13 +1.84 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Rxe7 Kf8 6.Bxa7 Kxe7
7.Bf2 Rc6 8.Ke4 Nb8 9.c5 Na6 10.Kd4 Nxc5
11.Kc4 Nxb3+ 12.Kxb3 (160.869.138) 635
best move: Ne4xc5 time: 4:16.484 min n/s: 635.241 CPU 100.0% n/s(1CPU): 635.241 nodes: 162.870.000
(Toga Checkov's color is dark red of course, Toga Blueberry had berry blue)
Fruit's strength here is I believe the smooth transition to endgames, knowing when to exchange and when not. Rybka more or less perfected that and is very strong in the late middle game, probably also because it knows when to exchange and what pieces. It makes sense then for the Belarus clones to be called after Robert Fischer, when we read this "blurb" on the Chessbase site for one of their recent Mikalchishin CDs, "Power of Exchange":
The great Botvinnik even said that chess is just multiple exchanges. He explained Fischer’s technique of play: He had never bad pieces - he always exchanged them for his opponent’s strong pieces.
So there is my proof that Little Robber is The Little Fish But some people just do not want to see what is staring them in the face. Only human.
Eelco
and your NPS is then just over 8 million after a while, more than ten times as fast as a slow compile on my Athlon with one core only. I ran it in parallel first so the speed was even lower, but now with 100% CPU but a slow compile I get, with 635.241 nodes per second :2r2k2/r3bp2/1n2p1p1/p1p1P1Pp/PpP1NP1P/1P1RBK2/8/3R4 w - -
Engine: Toga 1.4 Checkov Beta 4 (128 MB)
by Fabien Letouzey, Thomas Gaksch, Eelco de Groot
1/05 0:00 +0.71 1.Ra1 Kg7 (49)
1/05 0:00 +0.73 1.Re1 Kg7 (193)
1/05 0:00 +0.91 1.R3d2 Kg7 (437)
1/11 0:00 +1.03 1.Kg2 Kg7 (2.087)
2/14 0:00 +0.76 1.Kg2 Kg7 2.Bf2 (3.492)
2/14 0:00 +1.02 1.Nd6 Bxd6 2.exd6 Rd7 (3.604)
3/14 0:00 +1.00 1.Nd6 Bxd6 2.exd6 Rd7 3.Re1 Kg7 (6.088)
4/14 0:00 +0.90 1.Nd6 Bxd6 2.Rxd6 Rb7 3.Kg2 Kg7 (10.580)
4/14 0:00 +1.06 1.Kg2 Kg8 2.Bf2 Kg7 (11.587)
5/16 0:00 +1.02 1.Kg2 Kg7 2.Nd6 Rc6 3.Bf2 Rd7 4.Ne8+ Kf8
5.Rxd7 Nxd7 6.Rxd7 Kxe8 (28.655)
6/27 0:00 +0.80 1.Kg2 Kg7 2.Nd6 Rc6 3.Bf2 Bxd6
4.exd6 Rd7 5.R3d2 (64.528)
6/27 0:00 +0.94 1.Nd6 Rc6 2.Nb5 Ra8 3.Kg2 Kg7 (72.530)
7/27 0:00 +0.98 1.Nd6 Rc6 2.Nb5 Ra8 3.Kg2 Kg7 4.Bf2 Rh8 (133.330)
8/28 0:00 +0.76 1.Nd6 Rc6 2.Kg2 Ra8 3.Ne4 Kg7 4.Bg1 Rc7
5.Bf2 (278.992)
8/28 0:00 +0.88 1.Kg2 Rc6 2.Kg3 Ra8 3.Bc1 Kg7 4.Bb2 Rc7 (353.681)
9/29 0:01 +0.82 1.Kg2 Raa8 2.Kg3 Rc7 3.R1d2 Rac8
4.Bf2 Rb8 5.Rd1 Kg7 (1.145.326) 588
10/40 0:04 +0.82 1.Kg2 Raa8 2.Kg3 Kg7 3.Kf2 Rc6 4.Kf1 Kg8
5.Bf2 Kg7 6.Kg2 (2.427.546) 602
11/40 0:11 +0.81 1.Kg2 Raa8 2.Bc1 Rc6 3.Kf2 Rc7 4.Be3 Kg7
5.Kg2 Rc6 6.Kf1 Rc7 (6.778.506) 612
11/40 0:14 +0.92 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Bxa7 Rxa7 6.Rxe7 Nxe5+
7.Ke4 Rxe7 8.fxe5 Rc7 9.Rd8 Rc6 (9.059.087) 612
12/40 0:17 +0.89 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Rxe7 Nxe5+ 6.fxe5 Rxe7
7.Bxa7 Rxa7 8.Ke4 Rc7 9.Rd8 Kh7
10.Kd4 Kg7 (10.494.005) 615
13/40 0:26 +1.23 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Rxe7 Nxe5+ 6.fxe5 Rxe7
7.Bxa7 Rxa7 8.Ke4 Rc7 9.Rd6 Kf8
10.Kd4 Ke7 11.c5 Rc8 (16.547.260) 620
14/40 0:51 +1.34 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Rxe7 Nxe5+ 6.fxe5 Rxe7
7.Bxa7 Rxa7 8.Ke3 Kf8 9.Rd6 Ra8
10.c5 Rc8 11.c6 Rc7 12.Ke4 Ke7 (32.781.858) 633
15/55 1:56 +1.81 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Rxe7 Kf8 6.Bxa7 Kxe7
7.Bf2 Rc6 8.Ke4 Nb8 9.Bd4 Na6 10.Bg1 Rc8
11.Be3 (74.593.349) 640
16/55 4:13 +1.84 1.Nxc5 Rxc5 2.Rd8+ Kg7 3.Rb8 Nd7
4.Re8 Rcc7 5.Rxe7 Kf8 6.Bxa7 Kxe7
7.Bf2 Rc6 8.Ke4 Nb8 9.c5 Na6 10.Kd4 Nxc5
11.Kc4 Nxb3+ 12.Kxb3 (160.869.138) 635
best move: Ne4xc5 time: 4:16.484 min n/s: 635.241 CPU 100.0% n/s(1CPU): 635.241 nodes: 162.870.000
(Toga Checkov's color is dark red of course, Toga Blueberry had berry blue)
Fruit's strength here is I believe the smooth transition to endgames, knowing when to exchange and when not. Rybka more or less perfected that and is very strong in the late middle game, probably also because it knows when to exchange and what pieces. It makes sense then for the Belarus clones to be called after Robert Fischer, when we read this "blurb" on the Chessbase site for one of their recent Mikalchishin CDs, "Power of Exchange":
The great Botvinnik even said that chess is just multiple exchanges. He explained Fischer’s technique of play: He had never bad pieces - he always exchanged them for his opponent’s strong pieces.
So there is my proof that Little Robber is The Little Fish
But some people just do not want to see what is staring them in the face. Only human.Eelco
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan
-
Howard E
- Posts: 261
- Joined: Wed Mar 08, 2006 8:49 pm
Re: Test position Nxc5
Prodeo1.6.will not play Nxc5 for similar reasons as Gaviota. I noticed other engines that do not choose Nxc5 take a number of input moves before they "see" the value in that move.
-
pijl
Re: Test position Nxc5
The Baron 3.12.2 on a slow computer (T5500, 1.67Ghz), single core needs less than 30 seconds to choose Nxc5:
Code: Select all
ply score time nodes pv
6. 0.976 0:00.15 53879 e4d6 c8c6 d3d2 a7d7 f3g3 f8g7 d6e8 g7f8 d2d7 b6d7 d1d7 f8e8
7& 0.928 0:00.20 72755 e4d6 c8c6 d3d2 a7d7 f3g3 d7d8 d2d3
7. 0.928 0:00.50 204589 e4d6 c8c6 d3d2 a7d7 f3g3 d7d8 d2d3
8& 0.880 0:00.60 251000 e4d6 c8c6 d3d2 a7d7 f3g3 d7d8 d2d3 f8g7
8. 0.880 0:01.43 650089 e4d6 c8c6 d3d2 a7d7 f3g3 d7d8 d2d3 f8g7
9& 0.912 0:01.70 804575 e4d6 c8c6 d6b5 a7a8 d3d2 f8g7 b5d6 a8d8 d6f5 e6f5 d2d8 e7d8 d1d8
9. 0.912 0:02.58 1289238 e4d6 c8c6 d6b5 a7a8 d3d2 f8g7 b5d6 a8d8 d6f5 e6f5 d2d8 e7d8 d1d8
10& 0.896 0:02.93 1489563 e4d6 c8c6 d6b5 a7a8 d3d2 a8b8 b5d6 b8d8 f3g3 f8g7
10. 0.896 0:05.64 3120878 e4d6 c8c6 d6b5 a7a8 d3d2 a8b8 b5d6 b8d8 f3g3 f8g7
11& 0.896 0:06.10 3419698 e4d6 c8c6 d6b5 a7a8 d3d2 a8b8 b5d6 f8g7 d2d3 c6c7 f3g3
11. 0.896 0:10.07 5779060 e4d6 c8c6 d6b5 a7a8 d3d2 a8b8 b5d6 f8g7 d2d3 c6c7 f3g3
12& 0.880 0:12.32 7153820 e4d6 c8c6 d6b5 a7a8 d3d2 a8b8 b5d6 f8g7 f3g3 b8a8 d2d3 a8d8
12& 1.280 0:28.55 16665209 e4c5 c8c5 d3d8 f8g7 d8b8 b6d7 b8e8 c5c7 e8e7 d7e5 f4e5 c7e7 e3a7 e7a7 c4c5
12. 1.280 0:29.40 17212705 e4c5 c8c5 d3d8 f8g7 d8b8 b6d7 b8e8 c5c7 e8e7 d7e5 f4e5 c7e7 e3a7 e7a7 c4c5
13& 1.104 0:31.93 18845571 e4c5 c8c5 d3d8 f8g7 d8b8 b6d7 b8e8 c5c7 e3a7 c7a7 e8e7 d7e5 f3e4 a7e7 e4e5 e7c7 d1d8 c7c5 e5d6
13. 1.104 0:35.10 20990453 e4c5 c8c5 d3d8 f8g7 d8b8 b6d7 b8e8 c5c7 e3a7 c7a7 e8e7 d7e5 f3e4 a7e7 e4e5 e7c7 d1d8 c7c5 e5d6
14& 1.104 0:39.67 23811747 e4c5 c8c5 d3d8 f8g7 d8b8 b6d7 b8e8 c5c7 e3a7 c7a7 e8e7 d7e5 f3e4 a7e7 e4e5 e7c7 d1d8 c7c5 e5d6
14. 1.104 0:46.64 27942035 e4c5 c8c5 d3d8 f8g7 d8b8 b6d7 b8e8 c5c7 e3a7 c7a7 e8e7 d7e5 f3e4 a7e7 e4e5 e7c7 d1d8 c7c5 e5d6
15& 1.184 0:58.06 34497696 e4c5 c8c5 d3d8 f8g7 d8b8 b6d7 b8e8 c5c7 e3a7 c7a7 e8e7 d7e5 f3e4 a7e7 f4e5 e7c7 d1d8 c7c5 d8d7
15. 1.184 1:12.32 43406426 e4c5 c8c5 d3d8 f8g7 d8b8 b6d7 b8e8 c5c7 e3a7 c7a7 e8e7 d7e5 f3e4 a7e7 f4e5 e7c7 d1d8 c7c5 d8d7
16& 1.200 1:35.00 57352757 e4c5 c8c5 d3d8 f8g7 d8b8 b6d7 b8e8 c5c7 e3a7 c7a7 e8e7 d7e5 f3e4 a7e7 f4e5 e7c7 d1d8 c7c6 e4d4 c6c7
16. 1.200 1:59.46 73944961 e4c5 c8c5 d3d8 f8g7 d8b8 b6d7 b8e8 c5c7 e3a7 c7a7 e8e7 d7e5 f3e4 a7e7 f4e5 e7c7 d1d8 c7c6 e4d4 c6c7-
Michel
- Posts: 2272
- Joined: Mon Sep 29, 2008 1:50 am
Re: Test position Nxc5
Gnuchess 5.07.17testb finds it in 34 seconds on a core duo.
Code: Select all
memory 68
force
setboard 2r2k2/r3bp2/1n2p1p1/p1p1P1Pp/PpP1NP1P/1P1RBK2/8/3R4 w - - 0 1
analyze
1 155 0 45 R1d2
2 153 0 486 Kg3 Kg7
2 153 0 528 Kg3 Kg7
3 164 0 632 Kg3 Kg7 R3d2
3 164 0 885 Kg3 Kg7 R3d2
4 149 0 1246 Kg3 Kg7 R3d2 Raa8
4 181 0 2318 Nd6 Bxd6 exd6 Rd7 R3d2
4 181 0 2564 Nd6 Bxd6 exd6 Rd7 R3d2
5 167 0 3730 Nd6 Rc6 Nb5 Ra8 Kg3
5 167 1 6358 Nd6 Rc6 Nb5 Ra8 Kg3
6 143 1 9410 Nd6 Rc6 R3d2 Ra8 Rh2
6 143 4 20699 Nd6 Rc6 R3d2 Ra8 Rh2
7 160 5 25120 Nd6 Rc6 Kg3 Ra8 R3d2 Kg7 Nb5
7 160 6 31321 Nd6 Rc6 Kg3 Ra8 R3d2 Kg7 Nb5
8 142 7 40078 Nd6 Rc6 Kg3 Ra8 R1d2 Kg7 Ne4
8 142 15 98699 Nd6 Rc6 Kg3 Ra8 R1d2 Kg7 Ne4
9 134 22 178530 Nd6 Rc6 Kf2 Kg7 R3d2 Ra8 Nb7 Kg8 Kg3
9 135 31 278165 R1d2 Rc6 Bf2 Ra8 Nd6 Kg7 Be1 Kg8
9 135 57 594850 R1d2 Rc6 Bf2 Ra8 Nd6 Kg7 Be1 Kg8
10 130 71 748186 R1d2 Rc6 Nf6 Ra8 Ke4 Bxf6 gxf6 Kg8 Rd6 Nd5
10 131 102 1114167 Nf6 Rac7 Bxc5 Bxc5 Rd6
10 134 145 1605933 Nd6 Rc6 R3d2 Ra8 Kg3 Kg7 Rh2 Rh8 Rhd2 Rc7
10 134 156 1732117 Nd6 Rc6 R3d2 Ra8 Kg3 Kg7 Rh2 Rh8 Rhd2 Rc7
11 134 172 1912942 Nd6 Rc6 Bc1 Ra8 Bb2 Kg7 R3d2 Rf8 Kg3 Rh8 Rd3
11 134 246 2825935 Nd6 Rc6 Bc1 Ra8 Bb2 Kg7 R3d2 Rf8 Kg3 Rh8 Rd3
12 127 280 3214765 Nd6 Rc6 R3d2 Ra8 Kg3 Rd8 Kf2 Nc8 Nxc8 Rcxc8 Rd7
12 135 369 4286223 R3d2 Kg7 Nd6 Rc6 Ne8+ Kh7 Nf6+ Bxf6 gxf6 Rc8 Rd6 Rb7 Ke4 Rbb8 R6d3
12 135 516 6121729 R3d2 Kg7 Nd6 Rc6 Ne8+ Kh7 Nf6+ Bxf6 gxf6 Rc8 Rd6 Rb7 Ke4 Rbb8 R6d3
13 127 593 7064921 R3d2 Kg7 Nd6 Rc6 Ne8+ Kh8 Kf2 Ra8 Nd6 Kg8 Kg2 Kg7 Kg3
13 128 967 11677541 R1d2 Kg7 Nd6 Rc6 Ne8+ Kh8 Nf6 Ra8 Ke4 Rf8 Bg1 Bxf6 gxf6 Kg8 Bf2
13 128 1054 12713820 R1d2 Kg7 Nd6 Rc6 Ne8+ Kh8 Nf6 Ra8 Ke4 Rf8 Bg1 Bxf6 gxf6 Kg8 Bf2
14 127 1201 14452541 R1d2 Kg7 Nd6 Rc6 Ne8+ Kh8 Kg3 Ra8 Nd6 Kg7 Bg1 Rh8 Rd1 Rc7 Bf2
14 128 1504 17970019 Bg1 Rc6 Nf6 Ra8 Nd7+ Ke8 Nxb6 Rxb6 Kg3 Ra7 R3d2 Rc7 Rf1 Bd8
14 129 2011 24100456 Bf2 Rc6 Nf6 Ra8 Nd7+ Nxd7 Rxd7 Ke8 Kg3 Rd8 Kh3
14 129 2234 27041949 Bf2 Rc6 Nf6 Ra8 Nd7+ Nxd7 Rxd7 Ke8 Kg3 Rd8 Kh3
15 133 2646 32392984 Bf2 Kg7 Nd6 Rc6 Ne8+ Kh8 Nf6 Ra8 Nd7 Ra6 Nxb6 Raxb6 Ke4 Rd6
15 223 3447 43060009 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Bxa7 Rxa7 Rxe7 Nxe5+ Ke4 Rxe7 fxe5 Rc7 Rf1
15 223 3690 46300211 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Bxa7 Rxa7 Rxe7 Nxe5+ Ke4 Rxe7 fxe5
16 199 4021 50337636 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Bxa7 Rxa7 Rxe7 Nxe5+ Ke4 Rxe7 fxe5 Rc7
16 199 5067 64514524 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Bxa7 Rxa7 Rxe7
17 207 5833 73970807 Nxc5 Rxc5 Rd8+ Kg7 Rb8 Nd7 Re8 Rcc7 Bxa7 Rxa7 Rxe7 Nxe5+ Ke4 Rxe7 fxe5