New engine releases 2020

[Gabor Szots]

Yes the NN is included in the exe. I'm not sure how your compiler managed to get it under 2MB though . If people care, they can easily change line 13 of main.cpp to display the correct version

If I delete the nn from the source I obtain the same exe size which is 1437101 bytes, about 2 MB smaller than your exes. When I include the nn as well the size does not change. It seem it is left out somehow.

[Gabor Szots]

Yes the NN is included in the exe. I'm not sure how your compiler managed to get it under 2MB though . If people care, they can easily change line 13 of main.cpp to display the correct version

If I delete the nn from the source I obtain the same exe size which is 1437101 bytes, about 2 MB smaller than your exes. When I include the nn as well the size does not change. It seem it is left out somehow.

Sorry for the above, I missed that the source was not compiled again owing to the missing NN.

[Kieren Pearson]

I’ve noticed that non static compiles are much smaller, on my computer they are around 1MB. I’m not sure how the compiler does this but it might be doing some magic optimising the code that reads in the network into the float arrays. Considering that stuff can be determined at compile time it’s not surprising to me.

[JohnWoe]

I decided to release Mayhem 2.5.
Too many changes to remember. Mostly speedups. I can't remember.

This version is the first Mayhem to beat Glaurung 2.2 in Blitz match.
Which has been a difficult opponent for Mayhem.

Release: https://github.com/SamuraiDangyo/mayhem ... s/tag/v2.5
Source code: https://github.com/SamuraiDangyo/mayhem

Added KQKR heuristics. Was kinda annoying to see Mayhem always shuffle KQKR to ½-½.
Seems to work fine

[pgn][Event "Computer Chess Game"]
[Site "pc"]
[Date "2020.12.18"]
[Round "-"]
[White "Mayhem 2.5"]
[Black "Stockfish 12"]
[Result "1-0"]
[TimeControl "40/60"]
[FEN "8/8/5r2/5k2/3Q4/3K4/8/8 w - - 0 1"]
[SetUp "1"]

{--------------
. . . . . . . .
. . . . . . . .
. . . . . r . .
. . . . . k . .
. . . Q . . . .
. . . K . . . .
. . . . . . . .
. . . . . . . .
white to play
--------------}
1. Qe4+ {+8,36/12} Kg5 {-6,63/4 0,1} 2. Ke3 {+8,34/12 1,9} Rf7
{-4,74/16 1,0} 3. Qe5+ {+8,60/13 1,9} Kg6 {-4,58/16 0,6} 4. Ke4
{+8,71/13 1,8} Rf6 {-6,24/30 3} 5. Qe8+ {+8,85/13 1,7} Kh6 {-13,86/27 1,6}
6. Qg8 {+8,85/12 1,7} Rf1 {-13,79/23 0,9} 7. Qg4 {+9,18/13 1,6} Rb1
{-5,54/26 2,3} 8. Ke5 {+9,16/12 1,6} Rb5+ {-5,55/17 0,5} 9. Kf6
{+9,06/13 1,5} Rb6+ {-5,37/29 1,5} 10. Kf5 {+8,95/13 1,5} Rb5+
{-60,40/30 1,9} 11. Ke6 {+8,88/13 1,4} Rb6+ {-152,65/27 5} 12. Kd7
{+8,81/12 1,4} Rb5 {-152,65/26 0,4} 13. Qe2 {+8,51/11 1,4} Rg5
{-152,65/25 0,6} 14. Ke6 {+8,48/11 1,4} Rg6+ {-152,65/28 1,3} 15. Kf5
{+8,51/13 1,4} Rg5+ {-152,65/28 1,1} 16. Kf4 {+8,45/12 1,4} Rg7
{-152,65/29 1,2} 17. Qe5 {+8,32/11 1,4} Rf7+ {-128,56/21 0,5} 18. Kg4
{+8,20/13 1,4} Rg7+ {-152,65/30 2,4} 19. Kh4 {+8,31/13 1,4} Rf7
{-152,65/27 1,3} 20. Qg5+ {+7,99/11 1,4} Kh7 {-1,18/2 0,1} 21. Kh5
{+8,09/11 1,4} Rg7 {-1000,15/29 1,9} 22. Qe5 {+7,88/11 1,4} Rg1
{-1000,14/32 0,8} 23. Qe4+ {+9,60/13 1,4} Kg8 {-1000,15/32 0,8} 24. Kh6
{+9,69/12 1,4} Kf7 {-1000,15/32 0,7} 25. Qd5+ {+9,68/12 1,4} Ke8
{-1000,13/34 1,0} 26. Qa8+ {+9,72/12 1,4} Kd7 {-1000,12/37 0,9} 27. Qa7+
{+9,72/13 1,4} Kc6 {-1000,11/40 0,9} 28. Qxg1 {+9,73/13 1,4} Kb5
{-1000,09/46 0,9} 29. Qd4 {+13,29/11 1,4} Kc6 {-1000,08/51 1,0} 30. Kg6
{+13,74/11 1,4} Kc7 {-1000,07/55 1,1} 31. Qc5+ {+15,35/10 1,4} Kb7
{-1000,06/66 1,1} 32. Kf7 {+104,85/10 1,2} Kb8 {-1000,05/106 1,1} 33. Qc6
{+104,85/8 0,1} Ka7 {-60,27/1 0,1} 34. Ke7 {+104,85/7 0,1} Kb8
{-59,24/1 0,1} 35. Kd7 {+104,85/6 0,1} Ka7 {-60,27/4 0,1} 36. Qb5
{+104,85/5 0,1} Ka8 {-1000,02/17 0,1} 37. Kc7 {+104,85/4 0,1} Ka7
{-1000,01/46 0,1} 38. Qa4# {+104,85/3 0,1}
{Xboard adjudication: Checkmate} 1-0

[/pgn]

[jshriver]

I decided to release Mayhem 2.5.
Too many changes to remember. Mostly speedups. I can't remember.

Congrats! Look forward to playing and testing it.

[Chessqueen]

I decided to release Mayhem 2.5.
Too many changes to remember. Mostly speedups. I can't remember.

This version is the first Mayhem to beat Glaurung 2.2 in Blitz match.
Which has been a difficult opponent for Mayhem.

Release: https://github.com/SamuraiDangyo/mayhem ... s/tag/v2.5
Source code: https://github.com/SamuraiDangyo/mayhem

Added KQKR heuristics. Was kinda annoying to see Mayhem always shuffle KQKR to ½-½.
Seems to work fine

[pgn][Event "Computer Chess Game"]
[Site "pc"]
[Date "2020.12.18"]
[Round "-"]
[White "Mayhem 2.5"]
[Black "Stockfish 12"]
[Result "1-0"]
[TimeControl "40/60"]
[FEN "8/8/5r2/5k2/3Q4/3K4/8/8 w - - 0 1"]
[SetUp "1"]

{--------------
. . . . . . . .
. . . . . . . .
. . . . . r . .
. . . . . k . .
. . . Q . . . .
. . . K . . . .
. . . . . . . .
. . . . . . . .
white to play
--------------}
1. Qe4+ {+8,36/12} Kg5 {-6,63/4 0,1} 2. Ke3 {+8,34/12 1,9} Rf7
{-4,74/16 1,0} 3. Qe5+ {+8,60/13 1,9} Kg6 {-4,58/16 0,6} 4. Ke4
{+8,71/13 1,8} Rf6 {-6,24/30 3} 5. Qe8+ {+8,85/13 1,7} Kh6 {-13,86/27 1,6}
6. Qg8 {+8,85/12 1,7} Rf1 {-13,79/23 0,9} 7. Qg4 {+9,18/13 1,6} Rb1
{-5,54/26 2,3} 8. Ke5 {+9,16/12 1,6} Rb5+ {-5,55/17 0,5} 9. Kf6
{+9,06/13 1,5} Rb6+ {-5,37/29 1,5} 10. Kf5 {+8,95/13 1,5} Rb5+
{-60,40/30 1,9} 11. Ke6 {+8,88/13 1,4} Rb6+ {-152,65/27 5} 12. Kd7
{+8,81/12 1,4} Rb5 {-152,65/26 0,4} 13. Qe2 {+8,51/11 1,4} Rg5
{-152,65/25 0,6} 14. Ke6 {+8,48/11 1,4} Rg6+ {-152,65/28 1,3} 15. Kf5
{+8,51/13 1,4} Rg5+ {-152,65/28 1,1} 16. Kf4 {+8,45/12 1,4} Rg7
{-152,65/29 1,2} 17. Qe5 {+8,32/11 1,4} Rf7+ {-128,56/21 0,5} 18. Kg4
{+8,20/13 1,4} Rg7+ {-152,65/30 2,4} 19. Kh4 {+8,31/13 1,4} Rf7
{-152,65/27 1,3} 20. Qg5+ {+7,99/11 1,4} Kh7 {-1,18/2 0,1} 21. Kh5
{+8,09/11 1,4} Rg7 {-1000,15/29 1,9} 22. Qe5 {+7,88/11 1,4} Rg1
{-1000,14/32 0,8} 23. Qe4+ {+9,60/13 1,4} Kg8 {-1000,15/32 0,8} 24. Kh6
{+9,69/12 1,4} Kf7 {-1000,15/32 0,7} 25. Qd5+ {+9,68/12 1,4} Ke8
{-1000,13/34 1,0} 26. Qa8+ {+9,72/12 1,4} Kd7 {-1000,12/37 0,9} 27. Qa7+
{+9,72/13 1,4} Kc6 {-1000,11/40 0,9} 28. Qxg1 {+9,73/13 1,4} Kb5
{-1000,09/46 0,9} 29. Qd4 {+13,29/11 1,4} Kc6 {-1000,08/51 1,0} 30. Kg6
{+13,74/11 1,4} Kc7 {-1000,07/55 1,1} 31. Qc5+ {+15,35/10 1,4} Kb7
{-1000,06/66 1,1} 32. Kf7 {+104,85/10 1,2} Kb8 {-1000,05/106 1,1} 33. Qc6
{+104,85/8 0,1} Ka7 {-60,27/1 0,1} 34. Ke7 {+104,85/7 0,1} Kb8
{-59,24/1 0,1} 35. Kd7 {+104,85/6 0,1} Ka7 {-60,27/4 0,1} 36. Qb5
{+104,85/5 0,1} Ka8 {-1000,02/17 0,1} 37. Kc7 {+104,85/4 0,1} Ka7
{-1000,01/46 0,1} 38. Qa4# {+104,85/3 0,1}
{Xboard adjudication: Checkmate} 1-0

[/pgn]

Are you planning to use NN or SF NNUE very soon

[JohnWoe]

I decided to release Mayhem 2.5.
Too many changes to remember. Mostly speedups. I can't remember.

Congrats! Look forward to playing and testing it.

Thanks!
This 2.5 release should be a good one.

I decided to release Mayhem 2.5.
Too many changes to remember. Mostly speedups. I can't remember.

This version is the first Mayhem to beat Glaurung 2.2 in Blitz match.
Which has been a difficult opponent for Mayhem.

Release: https://github.com/SamuraiDangyo/mayhem ... s/tag/v2.5
Source code: https://github.com/SamuraiDangyo/mayhem

Added KQKR heuristics. Was kinda annoying to see Mayhem always shuffle KQKR to ½-½.
Seems to work fine

[pgn][Event "Computer Chess Game"]
[Site "pc"]
[Date "2020.12.18"]
[Round "-"]
[White "Mayhem 2.5"]
[Black "Stockfish 12"]
[Result "1-0"]
[TimeControl "40/60"]
[FEN "8/8/5r2/5k2/3Q4/3K4/8/8 w - - 0 1"]
[SetUp "1"]

{--------------
. . . . . . . .
. . . . . . . .
. . . . . r . .
. . . . . k . .
. . . Q . . . .
. . . K . . . .
. . . . . . . .
. . . . . . . .
white to play
--------------}
1. Qe4+ {+8,36/12} Kg5 {-6,63/4 0,1} 2. Ke3 {+8,34/12 1,9} Rf7
{-4,74/16 1,0} 3. Qe5+ {+8,60/13 1,9} Kg6 {-4,58/16 0,6} 4. Ke4
{+8,71/13 1,8} Rf6 {-6,24/30 3} 5. Qe8+ {+8,85/13 1,7} Kh6 {-13,86/27 1,6}
6. Qg8 {+8,85/12 1,7} Rf1 {-13,79/23 0,9} 7. Qg4 {+9,18/13 1,6} Rb1
{-5,54/26 2,3} 8. Ke5 {+9,16/12 1,6} Rb5+ {-5,55/17 0,5} 9. Kf6
{+9,06/13 1,5} Rb6+ {-5,37/29 1,5} 10. Kf5 {+8,95/13 1,5} Rb5+
{-60,40/30 1,9} 11. Ke6 {+8,88/13 1,4} Rb6+ {-152,65/27 5} 12. Kd7
{+8,81/12 1,4} Rb5 {-152,65/26 0,4} 13. Qe2 {+8,51/11 1,4} Rg5
{-152,65/25 0,6} 14. Ke6 {+8,48/11 1,4} Rg6+ {-152,65/28 1,3} 15. Kf5
{+8,51/13 1,4} Rg5+ {-152,65/28 1,1} 16. Kf4 {+8,45/12 1,4} Rg7
{-152,65/29 1,2} 17. Qe5 {+8,32/11 1,4} Rf7+ {-128,56/21 0,5} 18. Kg4
{+8,20/13 1,4} Rg7+ {-152,65/30 2,4} 19. Kh4 {+8,31/13 1,4} Rf7
{-152,65/27 1,3} 20. Qg5+ {+7,99/11 1,4} Kh7 {-1,18/2 0,1} 21. Kh5
{+8,09/11 1,4} Rg7 {-1000,15/29 1,9} 22. Qe5 {+7,88/11 1,4} Rg1
{-1000,14/32 0,8} 23. Qe4+ {+9,60/13 1,4} Kg8 {-1000,15/32 0,8} 24. Kh6
{+9,69/12 1,4} Kf7 {-1000,15/32 0,7} 25. Qd5+ {+9,68/12 1,4} Ke8
{-1000,13/34 1,0} 26. Qa8+ {+9,72/12 1,4} Kd7 {-1000,12/37 0,9} 27. Qa7+
{+9,72/13 1,4} Kc6 {-1000,11/40 0,9} 28. Qxg1 {+9,73/13 1,4} Kb5
{-1000,09/46 0,9} 29. Qd4 {+13,29/11 1,4} Kc6 {-1000,08/51 1,0} 30. Kg6
{+13,74/11 1,4} Kc7 {-1000,07/55 1,1} 31. Qc5+ {+15,35/10 1,4} Kb7
{-1000,06/66 1,1} 32. Kf7 {+104,85/10 1,2} Kb8 {-1000,05/106 1,1} 33. Qc6
{+104,85/8 0,1} Ka7 {-60,27/1 0,1} 34. Ke7 {+104,85/7 0,1} Kb8
{-59,24/1 0,1} 35. Kd7 {+104,85/6 0,1} Ka7 {-60,27/4 0,1} 36. Qb5
{+104,85/5 0,1} Ka8 {-1000,02/17 0,1} 37. Kc7 {+104,85/4 0,1} Ka7
{-1000,01/46 0,1} 38. Qa4# {+104,85/3 0,1}
{Xboard adjudication: Checkmate} 1-0

[/pgn]

Are you planning to use NN or SF NNUE very soon

Hello.
I already use SF NNUE. The latest EvalFile. Some modifications here and there in other parts of NNUE stuff. (Too C'ish to me and won't stand the test of time)

[Pi4Chess]

I decided to release Mayhem 2.5.
Too many changes to remember. Mostly speedups. I can't remember.

This version is the first Mayhem to beat Glaurung 2.2 in Blitz match.
Which has been a difficult opponent for Mayhem.

Release: https://github.com/SamuraiDangyo/mayhem ... s/tag/v2.5
Source code: https://github.com/SamuraiDangyo/mayhem

Added KQKR heuristics. Was kinda annoying to see Mayhem always shuffle KQKR to ½-½.
Seems to work fine

[pgn][Event "Computer Chess Game"]
[Site "pc"]
[Date "2020.12.18"]
[Round "-"]
[White "Mayhem 2.5"]
[Black "Stockfish 12"]
[Result "1-0"]
[TimeControl "40/60"]
[FEN "8/8/5r2/5k2/3Q4/3K4/8/8 w - - 0 1"]
[SetUp "1"]

{--------------
. . . . . . . .
. . . . . . . .
. . . . . r . .
. . . . . k . .
. . . Q . . . .
. . . K . . . .
. . . . . . . .
. . . . . . . .
white to play
--------------}
1. Qe4+ {+8,36/12} Kg5 {-6,63/4 0,1} 2. Ke3 {+8,34/12 1,9} Rf7
{-4,74/16 1,0} 3. Qe5+ {+8,60/13 1,9} Kg6 {-4,58/16 0,6} 4. Ke4
{+8,71/13 1,8} Rf6 {-6,24/30 3} 5. Qe8+ {+8,85/13 1,7} Kh6 {-13,86/27 1,6}
6. Qg8 {+8,85/12 1,7} Rf1 {-13,79/23 0,9} 7. Qg4 {+9,18/13 1,6} Rb1
{-5,54/26 2,3} 8. Ke5 {+9,16/12 1,6} Rb5+ {-5,55/17 0,5} 9. Kf6
{+9,06/13 1,5} Rb6+ {-5,37/29 1,5} 10. Kf5 {+8,95/13 1,5} Rb5+
{-60,40/30 1,9} 11. Ke6 {+8,88/13 1,4} Rb6+ {-152,65/27 5} 12. Kd7
{+8,81/12 1,4} Rb5 {-152,65/26 0,4} 13. Qe2 {+8,51/11 1,4} Rg5
{-152,65/25 0,6} 14. Ke6 {+8,48/11 1,4} Rg6+ {-152,65/28 1,3} 15. Kf5
{+8,51/13 1,4} Rg5+ {-152,65/28 1,1} 16. Kf4 {+8,45/12 1,4} Rg7
{-152,65/29 1,2} 17. Qe5 {+8,32/11 1,4} Rf7+ {-128,56/21 0,5} 18. Kg4
{+8,20/13 1,4} Rg7+ {-152,65/30 2,4} 19. Kh4 {+8,31/13 1,4} Rf7
{-152,65/27 1,3} 20. Qg5+ {+7,99/11 1,4} Kh7 {-1,18/2 0,1} 21. Kh5
{+8,09/11 1,4} Rg7 {-1000,15/29 1,9} 22. Qe5 {+7,88/11 1,4} Rg1
{-1000,14/32 0,8} 23. Qe4+ {+9,60/13 1,4} Kg8 {-1000,15/32 0,8} 24. Kh6
{+9,69/12 1,4} Kf7 {-1000,15/32 0,7} 25. Qd5+ {+9,68/12 1,4} Ke8
{-1000,13/34 1,0} 26. Qa8+ {+9,72/12 1,4} Kd7 {-1000,12/37 0,9} 27. Qa7+
{+9,72/13 1,4} Kc6 {-1000,11/40 0,9} 28. Qxg1 {+9,73/13 1,4} Kb5
{-1000,09/46 0,9} 29. Qd4 {+13,29/11 1,4} Kc6 {-1000,08/51 1,0} 30. Kg6
{+13,74/11 1,4} Kc7 {-1000,07/55 1,1} 31. Qc5+ {+15,35/10 1,4} Kb7
{-1000,06/66 1,1} 32. Kf7 {+104,85/10 1,2} Kb8 {-1000,05/106 1,1} 33. Qc6
{+104,85/8 0,1} Ka7 {-60,27/1 0,1} 34. Ke7 {+104,85/7 0,1} Kb8
{-59,24/1 0,1} 35. Kd7 {+104,85/6 0,1} Ka7 {-60,27/4 0,1} 36. Qb5
{+104,85/5 0,1} Ka8 {-1000,02/17 0,1} 37. Kc7 {+104,85/4 0,1} Ka7
{-1000,01/46 0,1} 38. Qa4# {+104,85/3 0,1}
{Xboard adjudication: Checkmate} 1-0

[/pgn]

hello,
Until this version i could not managed to compile for rpi4. so that's good news !
Now i have 1 question : is your engine only using 1 core ? I have 300 kns search speed for 1 core of rpi4

[Graham Banks]

I decided to release Mayhem 2.5.

Could you please include popcount and non-popcount exes as well?

[Gabor Szots]

Rasch r142 GS
https://github.com/Ingo60/rasch

I don't have much luck with this engine, it forfeits time in about 1 game out of 3. And that with incremental clock!

That's why I prefer testing released engines.