Syzygy implementations of top engines

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

User avatar
Laskos
Posts: 10948
Joined: Wed Jul 26, 2006 10:21 pm
Full name: Kai Laskos

Syzygy implementations of top engines

Post by Laskos »

I forgot to post a small experiment from one month ago, I will post it now.
I have an EPD file of 225 hard 5-men Wins. The ideal play exhibited by Texel 1.07 + Syzygy-5-men + Gaviota-5-men shows average length to Mate of 47.80 moves. I will now post a table of "efficiency" or "length to Mate" of top engines enabled with Syzygy-5-men. The time control is 1 second per move on these 225 hard 5-men Wins.

Code: Select all

Engine                      Moves to Mate
------------------------------------------
Ideal play                      47.80

Houdini 6.03                    59.43
Stockfish 9 (Ronald patch)      63.71
Komodo 11.2.2                   64.57
Stockfish 8                     68.25
------------------------------------------
Moreover, Komodo 11.2.2 enabled with 5-men Syzygy fails to convert two 5-men Wins (out of 225 positions tested):

Code: Select all

[Event "?"]
[Site "?"]
[Date "2018.03.14"]
[Round "19"]
[White "Komodo_Syzygy"]
[Black "Texel_Syzygy_Gaviota"]
[Result "1/2-1/2"]
[PlyCount "54"]
[SetUp "1"]
[TimeControl "1/move"]
[FEN "2q5/8/k7/4Q3/4K3/8/6P1/8 w - - 0 1"]

1. Qf6+ {+280.01/16 1.0s} 1. ... Ka7 {-293.00/13 1.0s} 2. Qe7+ {+280.01/15
1.0s} 2. ... Kb8 {-292.74/18 1.0s} 3. Qe5+ {+280.01/18 1.0s} 3. ... Kb7 {
-292.78/15 1.00s} 4. Kf3 {+280.01/18 1.0s} 4. ... Qf8+ {-292.74/19 1.00s} 
5. Ke4 {+280.01/5 0s} 5. ... Qc8 {-292.78/14 1.0s} 6. Qb2+ {+280.01/5 0s} 
6. ... Ka8 {-293.24/20 1.00s} 7. Qa3+ {+280.01/18 1.0s} 7. ... Kb7 {
-293.24/18 1.0s} 8. Qf3 {+280.01/18 1.0s} 8. ... Ka7 {-293.30/13 1.00s} 9.
 Qf7+ {+280.01/17 1.0s} 9. ... Ka8 {-293.32/18 1.00s} 10. Qf3 {+280.01/18 
1.0s} 10. ... Ka7 {-293.30/19 1.0s} 11. Qf7+ {+280.01/5 0.001s} 11. ... 
Ka8 {-293.32/18 1.0s} 12. g3 {+280.01/5 0.001s} 12. ... Qc2+ {-293.34/18 
1.00s} 13. Kf3 {+280.01/5 0s} 13. ... Qc6+ {-293.36/19 1.0s} 14. Kg4 {
+280.01/5 0s} 14. ... Qe4+ {-293.38/19 1.0s} 15. Qf4 {+280.01/18 1.0s} 
15. ... Qg6+ {-293.40/18 1.00s} 16. Kf3 {+280.01/19 1.0s} 16. ... Qd3+ {
-293.42/15 1.0s} 17. Qe3 {+280.01/19 1.0s} 17. ... Qf5+ {-293.40/15 1.00s}
18. Kg2 {+280.01/19 1.0s} 18. ... Qg6 {-293.38/18 1.0s} 19. Kf3 {+280.01/5
0s} 19. ... Qf5+ {-293.40/18 1.0s} 20. Qf4 {+280.01/5 0.001s} 20. ... Qd3+
{-293.42/21 1.0s} 21. Kf2 {+280.01/5 0s} 21. ... Qc2+ {-293.44/20 1.00s} 
22. Ke3 {+280.01/20 1.0s} 22. ... Qc5+ {-293.46/20 1.00s} 23. Kf3 {
+280.01/20 1.0s} 23. ... Qc3+ {-293.46/15 1.00s} 24. Kg2 {+280.01/19 1.0s}
24. ... Qc2+ {-293.40/14 1.0s} 25. Kh3 {+280.01/24 1.0s} 25. ... Qh7+ {
-293.38/28 1.00s} 26. Kg2 {+280.01/19 1.00s} 26. ... Qc2+ {-293.40/34 
1.00s} 27. Kf3 {+280.01/5 0.001s} 27. ... Qd3+ {0.00/100 0.005s, Draw by 
3-fold repetition} 1/2-1/2






[Event "?"]
[Site "?"]
[Date "2018.03.14"]
[Round "67"]
[White "Komodo_Syzygy"]
[Black "Texel_Syzygy_Gaviota"]
[Result "1/2-1/2"]
[PlyCount "53"]
[SetUp "1"]
[TimeControl "1/move"]
[FEN "8/8/6Q1/4Kp2/8/5r2/5k2/8 w - - 0 1"]

1. Qh6 {+280.01/24 1.0s} 1. ... Kg2 {-M96/97 0.005s} 2. Qg5+ {+280.01/27 
1.0s} 2. ... Kf2 {-M94/95 0.002s} 3. Qh5 {+280.01/26 1.00s} 3. ... Kg2 {
-M92/93 0.003s} 4. Kd4 {+280.01/26 1.00s} 4. ... Kg3 {-M90/91 0.005s} 5. 
Qh1 {+280.01/26 1.0s} 5. ... Kf2 {-M88/89 0.003s} 6. Qh2+ {+280.01/24 
1.00s} 6. ... Kf1 {-M86/87 0.002s} 7. Kc4 {+280.01/24 1.00s} 7. ... Ke1 {
-M84/85 0.002s} 8. Qg2 {+280.01/23 1.00s} 8. ... Rf2 {-M82/83 0.002s} 9. 
Qh3 {+280.01/21 1.00s} 9. ... Kd2 {-M80/81 0.002s} 10. Qd3+ {+280.01/25 
1.0s} 10. ... Ke1 {-M78/79 0.002s} 11. Kc3 {+280.01/24 1.00s} 11. ... f4 {
-M78/79 0.003s} 12. Qb5 {+280.01/26 1.00s} 12. ... Re2 {-M78/79 0.002s} 
13. Qg5 {+280.01/25 1.00s} 13. ... Rf2 {-M76/77 0.003s} 14. Qg4 {
+280.01/23 1.00s} 14. ... Rf1 {-M76/77 0.001s} 15. Qh4+ {+280.01/19 1.0s} 
15. ... Ke2 {-M78/79 0.001s} 16. Qh5+ {+280.01/22 1.00s} 16. ... Kf2 {
-M76/77 0.002s} 17. Kd2 {+280.01/21 1.00s} 17. ... Kg3 {-M74/75 0.003s} 
18. Qg6+ {+280.01/22 1.00s} 18. ... Kf3 {-M72/73 0.003s} 19. Qd3+ {
+280.01/21 1.0s} 19. ... Kg2 {-M70/71 0.001s} 20. Qc4 {+280.01/21 1.00s} 
20. ... Rf2+ {-M68/69 0.003s} 21. Ke1 {+280.01/23 1.0s} 21. ... Kg3 {
-M66/67 0.003s} 22. Qg8+ {+280.01/24 1.00s} 22. ... Kf3 {-M64/65 0.001s} 
23. Qd5+ {+280.01/23 1.0s} 23. ... Kg3 {-M66/67 0.002s} 24. Qg8+ {
+280.01/23 1.00s} 24. ... Kf3 {-M64/65 0.001s} 25. Qb3+ {+280.01/5 0.001s}
25. ... Kg2 {-M62/63 0.001s} 26. Qd5+ {+280.01/24 1.00s} 26. ... Kg3 {
-M66/67 0.001s} 27. Qg8+ {+280.01/5 0.001s, Draw by 3-fold repetition} 
1/2-1/2
[pgn][Event "?"]
[Site "?"]
[Date "2018.03.14"]
[Round "19"]
[White "Komodo_Syzygy"]
[Black "Texel_Syzygy_Gaviota"]
[Result "1/2-1/2"]
[PlyCount "54"]
[SetUp "1"]
[TimeControl "1/move"]
[FEN "2q5/8/k7/4Q3/4K3/8/6P1/8 w - - 0 1"]

1. Qf6+ {+280.01/16 1.0s} 1. ... Ka7 {-293.00/13 1.0s} 2. Qe7+ {+280.01/15
1.0s} 2. ... Kb8 {-292.74/18 1.0s} 3. Qe5+ {+280.01/18 1.0s} 3. ... Kb7 {
-292.78/15 1.00s} 4. Kf3 {+280.01/18 1.0s} 4. ... Qf8+ {-292.74/19 1.00s}
5. Ke4 {+280.01/5 0s} 5. ... Qc8 {-292.78/14 1.0s} 6. Qb2+ {+280.01/5 0s}
6. ... Ka8 {-293.24/20 1.00s} 7. Qa3+ {+280.01/18 1.0s} 7. ... Kb7 {
-293.24/18 1.0s} 8. Qf3 {+280.01/18 1.0s} 8. ... Ka7 {-293.30/13 1.00s} 9.
Qf7+ {+280.01/17 1.0s} 9. ... Ka8 {-293.32/18 1.00s} 10. Qf3 {+280.01/18
1.0s} 10. ... Ka7 {-293.30/19 1.0s} 11. Qf7+ {+280.01/5 0.001s} 11. ...
Ka8 {-293.32/18 1.0s} 12. g3 {+280.01/5 0.001s} 12. ... Qc2+ {-293.34/18
1.00s} 13. Kf3 {+280.01/5 0s} 13. ... Qc6+ {-293.36/19 1.0s} 14. Kg4 {
+280.01/5 0s} 14. ... Qe4+ {-293.38/19 1.0s} 15. Qf4 {+280.01/18 1.0s}
15. ... Qg6+ {-293.40/18 1.00s} 16. Kf3 {+280.01/19 1.0s} 16. ... Qd3+ {
-293.42/15 1.0s} 17. Qe3 {+280.01/19 1.0s} 17. ... Qf5+ {-293.40/15 1.00s}
18. Kg2 {+280.01/19 1.0s} 18. ... Qg6 {-293.38/18 1.0s} 19. Kf3 {+280.01/5
0s} 19. ... Qf5+ {-293.40/18 1.0s} 20. Qf4 {+280.01/5 0.001s} 20. ... Qd3+
{-293.42/21 1.0s} 21. Kf2 {+280.01/5 0s} 21. ... Qc2+ {-293.44/20 1.00s}
22. Ke3 {+280.01/20 1.0s} 22. ... Qc5+ {-293.46/20 1.00s} 23. Kf3 {
+280.01/20 1.0s} 23. ... Qc3+ {-293.46/15 1.00s} 24. Kg2 {+280.01/19 1.0s}
24. ... Qc2+ {-293.40/14 1.0s} 25. Kh3 {+280.01/24 1.0s} 25. ... Qh7+ {
-293.38/28 1.00s} 26. Kg2 {+280.01/19 1.00s} 26. ... Qc2+ {-293.40/34
1.00s} 27. Kf3 {+280.01/5 0.001s} 27. ... Qd3+ {0.00/100 0.005s, Draw by
3-fold repetition} 1/2-1/2[/pgn]

[pgn][Event "?"]
[Site "?"]
[Date "2018.03.14"]
[Round "67"]
[White "Komodo_Syzygy"]
[Black "Texel_Syzygy_Gaviota"]
[Result "1/2-1/2"]
[PlyCount "53"]
[SetUp "1"]
[TimeControl "1/move"]
[FEN "8/8/6Q1/4Kp2/8/5r2/5k2/8 w - - 0 1"]

1. Qh6 {+280.01/24 1.0s} 1. ... Kg2 {-M96/97 0.005s} 2. Qg5+ {+280.01/27
1.0s} 2. ... Kf2 {-M94/95 0.002s} 3. Qh5 {+280.01/26 1.00s} 3. ... Kg2 {
-M92/93 0.003s} 4. Kd4 {+280.01/26 1.00s} 4. ... Kg3 {-M90/91 0.005s} 5.
Qh1 {+280.01/26 1.0s} 5. ... Kf2 {-M88/89 0.003s} 6. Qh2+ {+280.01/24
1.00s} 6. ... Kf1 {-M86/87 0.002s} 7. Kc4 {+280.01/24 1.00s} 7. ... Ke1 {
-M84/85 0.002s} 8. Qg2 {+280.01/23 1.00s} 8. ... Rf2 {-M82/83 0.002s} 9.
Qh3 {+280.01/21 1.00s} 9. ... Kd2 {-M80/81 0.002s} 10. Qd3+ {+280.01/25
1.0s} 10. ... Ke1 {-M78/79 0.002s} 11. Kc3 {+280.01/24 1.00s} 11. ... f4 {
-M78/79 0.003s} 12. Qb5 {+280.01/26 1.00s} 12. ... Re2 {-M78/79 0.002s}
13. Qg5 {+280.01/25 1.00s} 13. ... Rf2 {-M76/77 0.003s} 14. Qg4 {
+280.01/23 1.00s} 14. ... Rf1 {-M76/77 0.001s} 15. Qh4+ {+280.01/19 1.0s}
15. ... Ke2 {-M78/79 0.001s} 16. Qh5+ {+280.01/22 1.00s} 16. ... Kf2 {
-M76/77 0.002s} 17. Kd2 {+280.01/21 1.00s} 17. ... Kg3 {-M74/75 0.003s}
18. Qg6+ {+280.01/22 1.00s} 18. ... Kf3 {-M72/73 0.003s} 19. Qd3+ {
+280.01/21 1.0s} 19. ... Kg2 {-M70/71 0.001s} 20. Qc4 {+280.01/21 1.00s}
20. ... Rf2+ {-M68/69 0.003s} 21. Ke1 {+280.01/23 1.0s} 21. ... Kg3 {
-M66/67 0.003s} 22. Qg8+ {+280.01/24 1.00s} 22. ... Kf3 {-M64/65 0.001s}
23. Qd5+ {+280.01/23 1.0s} 23. ... Kg3 {-M66/67 0.002s} 24. Qg8+ {
+280.01/23 1.00s} 24. ... Kf3 {-M64/65 0.001s} 25. Qb3+ {+280.01/5 0.001s}
25. ... Kg2 {-M62/63 0.001s} 26. Qd5+ {+280.01/24 1.00s} 26. ... Kg3 {
-M66/67 0.001s} 27. Qg8+ {+280.01/5 0.001s, Draw by 3-fold repetition}
1/2-1/2[/pgn]
Top
mjlef
Posts: 1494
Joined: Thu Mar 30, 2006 2:08 pm

Re: Syzygy implementations of top engines

Post by mjlef »

Laskos wrote:I forgot to post a small experiment from one month ago, I will post it now.
I have an EPD file of 225 hard 5-men Wins. The ideal play exhibited by Texel 1.07 + Syzygy-5-men + Gaviota-5-men shows average length to Mate of 47.80 moves. I will now post a table of "efficiency" or "length to Mate" of top engines enabled with Syzygy-5-men. The time control is 1 second per move on these 225 hard 5-men Wins.

Code: Select all

Engine                      Moves to Mate
------------------------------------------
Ideal play                      47.80

Houdini 6.03                    59.43
Stockfish 9 (Ronald patch)      63.71
Komodo 11.2.2                   64.57
Stockfish 8                     68.25
------------------------------------------
Moreover, Komodo 11.2.2 enabled with 5-men Syzygy fails to convert two 5-men Wins (out of 225 positions tested):

Code: Select all

[Event "?"]
[Site "?"]
[Date "2018.03.14"]
[Round "19"]
[White "Komodo_Syzygy"]
[Black "Texel_Syzygy_Gaviota"]
[Result "1/2-1/2"]
[PlyCount "54"]
[SetUp "1"]
[TimeControl "1/move"]
[FEN "2q5/8/k7/4Q3/4K3/8/6P1/8 w - - 0 1"]

1. Qf6+ {+280.01/16 1.0s} 1. ... Ka7 {-293.00/13 1.0s} 2. Qe7+ {+280.01/15
1.0s} 2. ... Kb8 {-292.74/18 1.0s} 3. Qe5+ {+280.01/18 1.0s} 3. ... Kb7 {
-292.78/15 1.00s} 4. Kf3 {+280.01/18 1.0s} 4. ... Qf8+ {-292.74/19 1.00s} 
5. Ke4 {+280.01/5 0s} 5. ... Qc8 {-292.78/14 1.0s} 6. Qb2+ {+280.01/5 0s} 
6. ... Ka8 {-293.24/20 1.00s} 7. Qa3+ {+280.01/18 1.0s} 7. ... Kb7 {
-293.24/18 1.0s} 8. Qf3 {+280.01/18 1.0s} 8. ... Ka7 {-293.30/13 1.00s} 9.
 Qf7+ {+280.01/17 1.0s} 9. ... Ka8 {-293.32/18 1.00s} 10. Qf3 {+280.01/18 
1.0s} 10. ... Ka7 {-293.30/19 1.0s} 11. Qf7+ {+280.01/5 0.001s} 11. ... 
Ka8 {-293.32/18 1.0s} 12. g3 {+280.01/5 0.001s} 12. ... Qc2+ {-293.34/18 
1.00s} 13. Kf3 {+280.01/5 0s} 13. ... Qc6+ {-293.36/19 1.0s} 14. Kg4 {
+280.01/5 0s} 14. ... Qe4+ {-293.38/19 1.0s} 15. Qf4 {+280.01/18 1.0s} 
15. ... Qg6+ {-293.40/18 1.00s} 16. Kf3 {+280.01/19 1.0s} 16. ... Qd3+ {
-293.42/15 1.0s} 17. Qe3 {+280.01/19 1.0s} 17. ... Qf5+ {-293.40/15 1.00s}
18. Kg2 {+280.01/19 1.0s} 18. ... Qg6 {-293.38/18 1.0s} 19. Kf3 {+280.01/5
0s} 19. ... Qf5+ {-293.40/18 1.0s} 20. Qf4 {+280.01/5 0.001s} 20. ... Qd3+
{-293.42/21 1.0s} 21. Kf2 {+280.01/5 0s} 21. ... Qc2+ {-293.44/20 1.00s} 
22. Ke3 {+280.01/20 1.0s} 22. ... Qc5+ {-293.46/20 1.00s} 23. Kf3 {
+280.01/20 1.0s} 23. ... Qc3+ {-293.46/15 1.00s} 24. Kg2 {+280.01/19 1.0s}
24. ... Qc2+ {-293.40/14 1.0s} 25. Kh3 {+280.01/24 1.0s} 25. ... Qh7+ {
-293.38/28 1.00s} 26. Kg2 {+280.01/19 1.00s} 26. ... Qc2+ {-293.40/34 
1.00s} 27. Kf3 {+280.01/5 0.001s} 27. ... Qd3+ {0.00/100 0.005s, Draw by 
3-fold repetition} 1/2-1/2






[Event "?"]
[Site "?"]
[Date "2018.03.14"]
[Round "67"]
[White "Komodo_Syzygy"]
[Black "Texel_Syzygy_Gaviota"]
[Result "1/2-1/2"]
[PlyCount "53"]
[SetUp "1"]
[TimeControl "1/move"]
[FEN "8/8/6Q1/4Kp2/8/5r2/5k2/8 w - - 0 1"]

1. Qh6 {+280.01/24 1.0s} 1. ... Kg2 {-M96/97 0.005s} 2. Qg5+ {+280.01/27 
1.0s} 2. ... Kf2 {-M94/95 0.002s} 3. Qh5 {+280.01/26 1.00s} 3. ... Kg2 {
-M92/93 0.003s} 4. Kd4 {+280.01/26 1.00s} 4. ... Kg3 {-M90/91 0.005s} 5. 
Qh1 {+280.01/26 1.0s} 5. ... Kf2 {-M88/89 0.003s} 6. Qh2+ {+280.01/24 
1.00s} 6. ... Kf1 {-M86/87 0.002s} 7. Kc4 {+280.01/24 1.00s} 7. ... Ke1 {
-M84/85 0.002s} 8. Qg2 {+280.01/23 1.00s} 8. ... Rf2 {-M82/83 0.002s} 9. 
Qh3 {+280.01/21 1.00s} 9. ... Kd2 {-M80/81 0.002s} 10. Qd3+ {+280.01/25 
1.0s} 10. ... Ke1 {-M78/79 0.002s} 11. Kc3 {+280.01/24 1.00s} 11. ... f4 {
-M78/79 0.003s} 12. Qb5 {+280.01/26 1.00s} 12. ... Re2 {-M78/79 0.002s} 
13. Qg5 {+280.01/25 1.00s} 13. ... Rf2 {-M76/77 0.003s} 14. Qg4 {
+280.01/23 1.00s} 14. ... Rf1 {-M76/77 0.001s} 15. Qh4+ {+280.01/19 1.0s} 
15. ... Ke2 {-M78/79 0.001s} 16. Qh5+ {+280.01/22 1.00s} 16. ... Kf2 {
-M76/77 0.002s} 17. Kd2 {+280.01/21 1.00s} 17. ... Kg3 {-M74/75 0.003s} 
18. Qg6+ {+280.01/22 1.00s} 18. ... Kf3 {-M72/73 0.003s} 19. Qd3+ {
+280.01/21 1.0s} 19. ... Kg2 {-M70/71 0.001s} 20. Qc4 {+280.01/21 1.00s} 
20. ... Rf2+ {-M68/69 0.003s} 21. Ke1 {+280.01/23 1.0s} 21. ... Kg3 {
-M66/67 0.003s} 22. Qg8+ {+280.01/24 1.00s} 22. ... Kf3 {-M64/65 0.001s} 
23. Qd5+ {+280.01/23 1.0s} 23. ... Kg3 {-M66/67 0.002s} 24. Qg8+ {
+280.01/23 1.00s} 24. ... Kf3 {-M64/65 0.001s} 25. Qb3+ {+280.01/5 0.001s}
25. ... Kg2 {-M62/63 0.001s} 26. Qd5+ {+280.01/24 1.00s} 26. ... Kg3 {
-M66/67 0.001s} 27. Qg8+ {+280.01/5 0.001s, Draw by 3-fold repetition} 
1/2-1/2
[pgn][Event "?"]
[Site "?"]
[Date "2018.03.14"]
[Round "19"]
[White "Komodo_Syzygy"]
[Black "Texel_Syzygy_Gaviota"]
[Result "1/2-1/2"]
[PlyCount "54"]
[SetUp "1"]
[TimeControl "1/move"]
[FEN "2q5/8/k7/4Q3/4K3/8/6P1/8 w - - 0 1"]

1. Qf6+ {+280.01/16 1.0s} 1. ... Ka7 {-293.00/13 1.0s} 2. Qe7+ {+280.01/15
1.0s} 2. ... Kb8 {-292.74/18 1.0s} 3. Qe5+ {+280.01/18 1.0s} 3. ... Kb7 {
-292.78/15 1.00s} 4. Kf3 {+280.01/18 1.0s} 4. ... Qf8+ {-292.74/19 1.00s}
5. Ke4 {+280.01/5 0s} 5. ... Qc8 {-292.78/14 1.0s} 6. Qb2+ {+280.01/5 0s}
6. ... Ka8 {-293.24/20 1.00s} 7. Qa3+ {+280.01/18 1.0s} 7. ... Kb7 {
-293.24/18 1.0s} 8. Qf3 {+280.01/18 1.0s} 8. ... Ka7 {-293.30/13 1.00s} 9.
Qf7+ {+280.01/17 1.0s} 9. ... Ka8 {-293.32/18 1.00s} 10. Qf3 {+280.01/18
1.0s} 10. ... Ka7 {-293.30/19 1.0s} 11. Qf7+ {+280.01/5 0.001s} 11. ...
Ka8 {-293.32/18 1.0s} 12. g3 {+280.01/5 0.001s} 12. ... Qc2+ {-293.34/18
1.00s} 13. Kf3 {+280.01/5 0s} 13. ... Qc6+ {-293.36/19 1.0s} 14. Kg4 {
+280.01/5 0s} 14. ... Qe4+ {-293.38/19 1.0s} 15. Qf4 {+280.01/18 1.0s}
15. ... Qg6+ {-293.40/18 1.00s} 16. Kf3 {+280.01/19 1.0s} 16. ... Qd3+ {
-293.42/15 1.0s} 17. Qe3 {+280.01/19 1.0s} 17. ... Qf5+ {-293.40/15 1.00s}
18. Kg2 {+280.01/19 1.0s} 18. ... Qg6 {-293.38/18 1.0s} 19. Kf3 {+280.01/5
0s} 19. ... Qf5+ {-293.40/18 1.0s} 20. Qf4 {+280.01/5 0.001s} 20. ... Qd3+
{-293.42/21 1.0s} 21. Kf2 {+280.01/5 0s} 21. ... Qc2+ {-293.44/20 1.00s}
22. Ke3 {+280.01/20 1.0s} 22. ... Qc5+ {-293.46/20 1.00s} 23. Kf3 {
+280.01/20 1.0s} 23. ... Qc3+ {-293.46/15 1.00s} 24. Kg2 {+280.01/19 1.0s}
24. ... Qc2+ {-293.40/14 1.0s} 25. Kh3 {+280.01/24 1.0s} 25. ... Qh7+ {
-293.38/28 1.00s} 26. Kg2 {+280.01/19 1.00s} 26. ... Qc2+ {-293.40/34
1.00s} 27. Kf3 {+280.01/5 0.001s} 27. ... Qd3+ {0.00/100 0.005s, Draw by
3-fold repetition} 1/2-1/2[/pgn]

[pgn][Event "?"]
[Site "?"]
[Date "2018.03.14"]
[Round "67"]
[White "Komodo_Syzygy"]
[Black "Texel_Syzygy_Gaviota"]
[Result "1/2-1/2"]
[PlyCount "53"]
[SetUp "1"]
[TimeControl "1/move"]
[FEN "8/8/6Q1/4Kp2/8/5r2/5k2/8 w - - 0 1"]

1. Qh6 {+280.01/24 1.0s} 1. ... Kg2 {-M96/97 0.005s} 2. Qg5+ {+280.01/27
1.0s} 2. ... Kf2 {-M94/95 0.002s} 3. Qh5 {+280.01/26 1.00s} 3. ... Kg2 {
-M92/93 0.003s} 4. Kd4 {+280.01/26 1.00s} 4. ... Kg3 {-M90/91 0.005s} 5.
Qh1 {+280.01/26 1.0s} 5. ... Kf2 {-M88/89 0.003s} 6. Qh2+ {+280.01/24
1.00s} 6. ... Kf1 {-M86/87 0.002s} 7. Kc4 {+280.01/24 1.00s} 7. ... Ke1 {
-M84/85 0.002s} 8. Qg2 {+280.01/23 1.00s} 8. ... Rf2 {-M82/83 0.002s} 9.
Qh3 {+280.01/21 1.00s} 9. ... Kd2 {-M80/81 0.002s} 10. Qd3+ {+280.01/25
1.0s} 10. ... Ke1 {-M78/79 0.002s} 11. Kc3 {+280.01/24 1.00s} 11. ... f4 {
-M78/79 0.003s} 12. Qb5 {+280.01/26 1.00s} 12. ... Re2 {-M78/79 0.002s}
13. Qg5 {+280.01/25 1.00s} 13. ... Rf2 {-M76/77 0.003s} 14. Qg4 {
+280.01/23 1.00s} 14. ... Rf1 {-M76/77 0.001s} 15. Qh4+ {+280.01/19 1.0s}
15. ... Ke2 {-M78/79 0.001s} 16. Qh5+ {+280.01/22 1.00s} 16. ... Kf2 {
-M76/77 0.002s} 17. Kd2 {+280.01/21 1.00s} 17. ... Kg3 {-M74/75 0.003s}
18. Qg6+ {+280.01/22 1.00s} 18. ... Kf3 {-M72/73 0.003s} 19. Qd3+ {
+280.01/21 1.0s} 19. ... Kg2 {-M70/71 0.001s} 20. Qc4 {+280.01/21 1.00s}
20. ... Rf2+ {-M68/69 0.003s} 21. Ke1 {+280.01/23 1.0s} 21. ... Kg3 {
-M66/67 0.003s} 22. Qg8+ {+280.01/24 1.00s} 22. ... Kf3 {-M64/65 0.001s}
23. Qd5+ {+280.01/23 1.0s} 23. ... Kg3 {-M66/67 0.002s} 24. Qg8+ {
+280.01/23 1.00s} 24. ... Kf3 {-M64/65 0.001s} 25. Qb3+ {+280.01/5 0.001s}
25. ... Kg2 {-M62/63 0.001s} 26. Qd5+ {+280.01/24 1.00s} 26. ... Kg3 {
-M66/67 0.001s} 27. Qg8+ {+280.01/5 0.001s, Draw by 3-fold repetition}
1/2-1/2[/pgn]
I cannot reproduce this on my laptop. And the score Komodo shows for a syzygy mate value is about 318.00 and not the 280.01 values shown. I assume you had Syzygy set right and all the files are present. Smart Syzygy off, right? We will investigate.
Top
User avatar
Laskos
Posts: 10948
Joined: Wed Jul 26, 2006 10:21 pm
Full name: Kai Laskos

Re: Syzygy implementations of top engines

Post by Laskos »

mjlef wrote:
I cannot reproduce this on my laptop. And the score Komodo shows for a syzygy mate value is about 318.00 and not the 280.01 values shown. I assume you had Syzygy set right and all the files are present. Smart Syzygy off, right? We will investigate.
I checked my Syzygy 5-men files, all seems right, and other engines never exhibited something wrong with them.

The Cutechess-Cli parameters are (others are default):

Code: Select all

 { 
    "command" : "komodo_1122", 
    "name" : "Komodo_Syzygy", 
    "protocol": "uci",
    "options" : [ 
      { 
        "name" : "Hash", 
        "value" : 64 
      }, 
      { 
        "name" : "Threads", 
        "value" : 1 
      }, 
      { 
        "name" : "Contempt", 
        "value" : 0
      }, 
      { 
        "name" : "Overhead ms", 
        "value" : 30
      }, 
      { 
        "name" : "Variety", 
        "value" : 0
      },
      {
        "name" : "SyzygyPath",
        "value" : "H:/syzygy5"
      }
 }
Top
Vizvezdenec
Posts: 52
Joined: Fri Jan 12, 2018 1:30 am

Re: Syzygy implementations of top engines

Post by Vizvezdenec »

SF is known to uselessly shuffle in TB positions till it sees the only winning moves which it makes instantly then.
But since it doesn't lose elo it devs don't really care about it that much.
Top
User avatar
Laskos
Posts: 10948
Joined: Wed Jul 26, 2006 10:21 pm
Full name: Kai Laskos

Re: Syzygy implementations of top engines

Post by Laskos »

Vizvezdenec wrote:SF is known to uselessly shuffle in TB positions till it sees the only winning moves which it makes instantly then.
But since it doesn't lose elo it devs don't really care about it that much.
Ronald's patch in SF9 was meant to avoid this sort of shuffling, and it does shorten the length to Mate. But Houdini 6.03 still comes better than SF9.
Top
syzygy
Posts: 5557
Joined: Tue Feb 28, 2012 11:56 pm

Re: Syzygy implementations of top engines

Post by syzygy »

Laskos wrote:
Vizvezdenec wrote:SF is known to uselessly shuffle in TB positions till it sees the only winning moves which it makes instantly then.
But since it doesn't lose elo it devs don't really care about it that much.
Ronald's patch in SF9 was meant to avoid this sort of shuffling, and it does shorten the length to Mate. But Houdini 6.03 still comes better than SF9.
No, that patch only affects SF's behaviour after it has found a TB win in the search but before the TB win is reached on the board.

The alleged useless shuffling refers to rare TB positions that are too difficult for SF's regular TB-less search to figure out how to make progress.

I don't know if H6 handles those positions "better" (without switching to pure DTZ play which will generate it's own set of user complaints).
Top
Jouni
Posts: 3278
Joined: Wed Mar 08, 2006 8:15 pm

Re: Syzygy implementations of top engines

Post by Jouni »

Because most tests have something like +6,50 adjucation conversion of win is not necessary at all! BTW in all of my tests SF plays endings better than Houdini (syzygy or not).
Jouni
Top
User avatar
Laskos
Posts: 10948
Joined: Wed Jul 26, 2006 10:21 pm
Full name: Kai Laskos

Re: Syzygy implementations of top engines

Post by Laskos »

Jouni wrote:Because most tests have something like +6,50 adjucation conversion of win is not necessary at all! BTW in all of my tests SF plays endings better than Houdini (syzygy or not).
Here was the Syzygy implementation with TBs at root. Aside Komodo, which seems to miss some wins, there is no Elo loss, but sometimes one would like to see the shortest Mate.
Top
Joerg Oster
Posts: 937
Joined: Fri Mar 10, 2006 4:29 pm
Location: Germany

Re: Syzygy implementations of top engines

Post by Joerg Oster »

Laskos wrote:
Jouni wrote:Because most tests have something like +6,50 adjucation conversion of win is not necessary at all! BTW in all of my tests SF plays endings better than Houdini (syzygy or not).
Here was the Syzygy implementation with TBs at root. Aside Komodo, which seems to miss some wins, there is no Elo loss, but sometimes one would like to see the shortest Mate.
Well, I don't think this is caused by the syzygy-implementation,
but simply an effect of missing knowledge how to make progress in certain endgames, for instance KRPPKR with the pawns not far advanced.
The syzygy-implementation ensures the win, but cannot help in finding the mate.
And search alone is almost helpless in the more difficult endgames to find a mate in a reasonable amount of time.

And like already mentioned by Jouni, this goes unnoticed in general because of game adjudication.
Jörg Oster
Top
User avatar
Laskos
Posts: 10948
Joined: Wed Jul 26, 2006 10:21 pm
Full name: Kai Laskos

Re: Syzygy implementations of top engines

Post by Laskos »

Joerg Oster wrote:
Laskos wrote:
Jouni wrote:Because most tests have something like +6,50 adjucation conversion of win is not necessary at all! BTW in all of my tests SF plays endings better than Houdini (syzygy or not).
Here was the Syzygy implementation with TBs at root. Aside Komodo, which seems to miss some wins, there is no Elo loss, but sometimes one would like to see the shortest Mate.
Well, I don't think this is caused by the syzygy-implementation,
but simply an effect of missing knowledge how to make progress in certain endgames, for instance KRPPKR with the pawns not far advanced.
The syzygy-implementation ensures the win, but cannot help in finding the mate.
And search alone is almost helpless in the more difficult endgames to find a mate in a reasonable amount of time.

And like already mentioned by Jouni, this goes unnoticed in general because of game adjudication.
Isn't combining engine's eval and DTZ50 dependent on implementation? As general TB implementation, if I am not wrong, Texel's DTZ50 + DTM gives the shortest paths on root positions. IIRC Houdini's DTZ50 + DTM is not as good as Texel's, my impression was that in root TB positions, Nalimovs just neglect DTZ50, and might miss even conversion. But I haven't checked thoroughly.
Top

Return to “Computer Chess Club: General Topics”