Page 1 of 2

SF mates Leela in 25 moves at CCC

Posted: Fri Mar 15, 2019 1:41 pm
by Jouni
[pgn][Event "CCC 6: The Winter Classic Final (10|10)"] [Site "?"] [Date "2019.03.14"] [Round "1"] [White "Stockfish"] [Black "Lc0"] [Result "1-0"] [ECO "B04"] [GameDuration "00:12:55"] [GameEndTime "2019-03-14T17:32:38.175 PDT"] [GameStartTime "2019-03-14T17:19:42.757 PDT"] [Opening "Alekhine's defense"] [PlyCount "49"] [TimeControl "600+10"] [Variation "Modern, Larsen Variation"] 1. e4 Nf6 2. e5 Nd5 3. d4 d6 4. Nf3 dxe5 5. Nxe5 Nd7 6. Nxf7 Kxf7 7. Qh5+ Ke6 8. c4 N5f6 9. d5+ Kd6 10. Qf7 Ne5 11. Bf4 c5 12. Nc3 a6 13. b4 b6 14. Bd3 g6 15. bxc5+ bxc5 16. Rb1 Rg8 17. Rb7 Qa5 18. O-O Qxc3 19. Rfb1 Qa5 20. Bxe5+ Kxe5 21. Rxe7+ Bxe7 22. Qxe7+ Be6 23. Qxe6+ Kd4 24. Qe3+ Kc3 25. Rb3# 1-0[/pgn]

11. -c5 is losing move?

BTW SF speed in CCC 6 is stunning: 200 meganodes/s in endings with 7 piece syzygy :o . How is it possible with 2 TB SDD!?

Re: SF mates Leela in 25 moves at CCC

Posted: Fri Mar 15, 2019 2:21 pm
by xr_a_y
How many book moves were used ?

Re: SF mates Leela in 25 moves at CCC

Posted: Fri Mar 15, 2019 2:47 pm
by pferd
I have seen a couple of those quick losses in my private matches, too. Leela runs with standard settings + 5-men syzygy. Stockfish uses 4threads +2048mb Hash.

I picked up Leela's worst loss:

[pgn][Event "My Tournament"] [Site "?"] [Date "2019.03.12"] [Round "2"] [White "Stockfish 120319 64 BMI2"] [Black "Lc0 v0.22.0-dev 41500"] [Result "1-0"] [ECO "D15"] [GameDuration "00:03:43"] [GameEndTime "2019-03-12T18:31:34.538 CET"] [GameStartTime "2019-03-12T18:27:51.021 CET"] [Opening "QGD Slav"] [PlyCount "39"] [TimeControl "300+3"] [Variation "4.Nc3"] 1.d4 {book} 1...d5 {book} 2.c4 {book} 2...c6 {book} 3.Nf3 {book} 3...Nf6 {book} 4.Nc3 {book} 4...e6 {book} 5.Bg5 {book} 5...dxc4 {book} 6.e4 {book} 6...b5 {book} 7.e5 {book} 7...h6 {-0.83/16 4.4s} 8.Bh4 {+0.22/25 10s} 8...g5 {-0.78/17 4.9s} 9.Nxg5 {0.00/28 8.8s} 9...hxg5 {-0.77/17 5.8s} 10.Bxg5 {0.00/28 4.5s} 10...Nbd7 {-0.75/14 7.2s} 11.exf6 {+0.28/27 10s} 11...Bb7 {-0.75/14 11s} 12.g3 {+0.33/26 7.9s} 12...c5 {-0.79/14 4.6s} 13.d5 {+0.60/23 4.3s} 13...Qc7 {-0.62/17 14s} 14.Qe2 {+0.61/29 56s} 14...Qe5 {-0.48/18 5.8s} 15.Bg2 {+0.90/23 4.8s} 15...Qxe2+ {+0.04/17 22s} 16.Kxe2 {+1.25/25 2.9s} 16...b4 {+0.26/19 2.1s} 17.Nb5 {+2.03/24 5.4s} 17...Kd8 {+0.16/20 4.4s} 18.dxe6 {+4.53/26 4.0s} 18...Bxg2 {-0.13/17 7.6s} 19.e7+ {+4.65/26 3.9s} 19...Ke8 {-0.11/18 7.1s} 20.Nd6# {+M1/127 0.003s, White mates} 1-0 [/pgn]

Re: SF mates Leela in 25 moves at CCC

Posted: Fri Mar 15, 2019 3:21 pm
by jdart
I think 5. .. Nd7 is just bad, and this may be an example of CCC choosing an unbalanced opening. The game stats from Chessbase actually don't look too horrible, but with best play White is better. The game follows known theory at least through 13. b4. Black had a few chances to deviate: for example 8. .. N7f6 was possible, but the alternatives look bad, too.

--Jon

Re: SF mates Leela in 25 moves at CCC

Posted: Fri Mar 15, 2019 4:03 pm
by Eduard
Nice game, thanks! :P

In 3 variants mode, Stockfish does not want to play 5 ... Nd7. It must be a Leela move. I do not think it was a book move. Leela is not good tactically, then something like that can happen. 8-)

Btw, I like such sacrifices on f7!

Re: SF mates Leela in 25 moves at CCC

Posted: Fri Mar 15, 2019 4:08 pm
by Jesse Gersenson
It is not exactly a 2 TB ssd. Rather, it is 10x 2TB drives arranged in "raid 10".

Re: SF mates Leela in 25 moves at CCC

Posted: Fri Mar 15, 2019 4:11 pm
by jp
xr_a_y wrote: Fri Mar 15, 2019 2:21 pm How many book moves were used ?
Three. 4.Nf3 on are their own moves.

Re: SF mates Leela in 25 moves at CCC

Posted: Fri Mar 15, 2019 4:12 pm
by Jesse Gersenson

Re: SF mates Leela in 25 moves at CCC

Posted: Fri Mar 15, 2019 4:16 pm
by jp
jdart wrote: Fri Mar 15, 2019 3:21 pm I think 5. .. Nd7 is just bad, and this may be an example of CCC choosing an unbalanced opening. The game stats from Chessbase actually don't look too horrible, but with best play White is better. The game follows known theory at least through 13. b4. Black had a few chances to deviate: for example 8. .. N7f6 was possible, but the alternatives look bad, too.
--Jon
Move 4 on are their own moves. In the reverse game, Lc chose 4. c5. The game ended 1/2-1/2.

Re: SF mates Leela in 25 moves at CCC

Posted: Fri Mar 15, 2019 7:47 pm
by supersharp77
Jouni wrote: Fri Mar 15, 2019 1:41 pm [pgn][Event "CCC 6: The Winter Classic Final (10|10)"] [Site "?"] [Date "2019.03.14"] [Round "1"] [White "Stockfish"] [Black "Lc0"] [Result "1-0"] [ECO "B04"] [GameDuration "00:12:55"] [GameEndTime "2019-03-14T17:32:38.175 PDT"] [GameStartTime "2019-03-14T17:19:42.757 PDT"] [Opening "Alekhine's defense"] [PlyCount "49"] [TimeControl "600+10"] [Variation "Modern, Larsen Variation"] 1. e4 Nf6 2. e5 Nd5 3. d4 d6 4. Nf3 dxe5 5. Nxe5 Nd7 6. Nxf7 Kxf7 7. Qh5+ Ke6 8. c4 N5f6 9. d5+ Kd6 10. Qf7 Ne5 11. Bf4 c5 12. Nc3 a6 13. b4 b6 14. Bd3 g6 15. bxc5+ bxc5 16. Rb1 Rg8 17. Rb7 Qa5 18. O-O Qxc3 19. Rfb1 Qa5 20. Bxe5+ Kxe5 21. Rxe7+ Bxe7 22. Qxe7+ Be6 23. Qxe6+ Kd4 24. Qe3+ Kc3 25. Rb3# 1-0[/pgn]

11. -c5 is losing move?

BTW SF speed in CCC 6 is stunning: 200 meganodes/s in endings with 7 piece syzygy :o . How is it possible with 2 TB SDD!?
Move #17 by White....Rb7 !! A masterstroke! Thats why we love Chess!! AR :D :wink: