New engine releases 2020

[Werner]

I have released Francesca 0.30. Available from www.silentshark.co.uk
Not a huge jump in strength, maybe 30 ELO or so.
Changes
Small changes. Minor optimisations. More aggressive null move pruning. IID replaced by IIR (thanks to Ed Schroder for idea). Tweak to futility to make more aggressive or not, depending on if position is improving. Separate search functions when in check. Minor eval tweaks - tuning needed
Happy Christmas to everyone. Let's hope 2021 is a better year, and the world gets on top of the Coronavirus.

Hi - from 1000 games I have had 30 games with lost in time. Most times in won positions (from white or black). Example
[Event "Comb2"] [Site "AGANDO_HP_PC"] [Date "2020.12.26"] [Round "17"] [White "Francesca MAD 0.30"] [Black "Counter 3.5 x64 1CPU"] [Result "0-1"] [ECO "B90"] [WhiteElo "2200"] [BlackElo "2200"] [PlyCount "116"] [EventDate "2020.??.??"] [TimeControl "40/480:40/480:40/480"] 1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. h3 g6 7. g4 Bg7 8. Bg2 O-O 9. Be3 Nc6 10. Qd2 Bd7 11. O-O-O Rc8 12. f4 {0.41/32 22} Na5 {-0.64/17 12} 13. e5 {0.27/32 10} Nc4 {-0.60/19 22} 14. exf6 {0.27/34 10} Nxd2 {-0.28/20 16} 15. fxg7 {-1.31/35 18} Kxg7 {-0.34/20 13} 16. Rxd2 {-1.34/35 11} Rxc3 {-0.29/ 18 16} 17. bxc3 {-0.81/31 18} Qa5 {-0.23/20 17} 18. Kb1 {-0.45/32 10} Qb6+ { -0.15/19 12} 19. Ka1 {-0.37/34 10} Qc7 {-0.17/20 12} 20. Ne2 {-0.55/36 54} Bc6 {-0.11/19 14} 21. Bd5 {-0.54/35 19} e6 {-0.14/20 14} 22. Bxc6 {-0.32/33 11} Qxc6 {-0.17/20 13} 23. Rhd1 {-0.17/35 10} Qe4 {-0.18/21 21} 24. Rd3 {-0.17/35 20} e5 {-0.18/19 28} 25. Re1 {-0.23/30 15} Qc6 {-0.32/18 15} 26. Bd2 {-0.01/34 10} Re8 {-0.14/19 18} 27. fxe5 {-0.01/36 9} dxe5 {-0.59/17 15} 28. Ng3 { 0.00/35 9} Re6 {-0.60/21 15} 29. Ne4 {0.08/34 10} h6 {-0.68/20 22} 30. Rf3 { 0.16/33 12} b6 {-0.60/19 12} 31. h4 {0.25/32 15} Re7 {-0.94/19 20} 32. Bc1 { 0.65/33 15} Qa4 {-1.00/20 12} 33. h5 {0.84/35 14} g5 {-1.07/19 13} 34. Kb1 { 1.24/35 11} Kf8 {-1.47/19 26} 35. Rf6 {0.92/37 14} Re6 {-1.19/21 14} 36. Rxe6 { 1.08/41 14} fxe6 {-1.15/23 11} 37. Nf2 {1.35/37 9} Qb5+ {-0.91/23 11} 38. Bb2 { 1.43/40 8} Qa4 {-0.91/25 13} 39. Rxe5 {1.37/40 9} Qf4 {-1.16/23 23} 40. Nd3 { 1.48/39 13} Qxg4 {-1.17/21 10} 41. Bc1 {1.57/38 17} Ke7 {-1.07/21 12} 42. Be3 { 1.67/37 9} Qxh5 {-1.07/20 9} 43. Bxb6 {1.97/30 8} Qf3 {-0.92/20 9} 44. Kb2 { 1.41/29 16} g4 {-0.62/19 12} 45. Re3 {1.35/31 15} Qd5 {-0.82/20 10} 46. a4 { 1.24/32 10} Qg5 {-0.69/20 11} 47. Re5 {1.47/34 9} Qg8 {-0.80/22 29} 48. Bd4 { 1.52/33 8} g3 {-0.83/20 17} 49. Nc5 {1.66/36 9} h5 {-0.78/21 32} 50. Rxe6+ { 1.92/33 10} Qxe6 {-1.26/22 30} 51. Nxe6 {4.14/42 7} Kxe6 {-2.10/24 28} 52. Ba7 {3.36/42 11} h4 {-2.79/22 24} 53. Bb8 {3.92/42 8} Kd5 {-2.95/25 21} 54. Kb3 { 4.09/42 7} a5 {-2.26/25 8} 55. c4+ {13.53/43 7} Ke4 {-4.44/25 11} 56. c5 { 19.97/45 8} Kd5 {-5.04/26 7} 57. c6 {31.97/46 6} Kxc6 {-14.45/24 15} 58. c4 { 63.26/48 45} Kb7 {time forfeit -16.45/26 13 Weiss überschreitet die Zeit} 0-1

[silentshark]

I have released Francesca 0.30. Available from www.silentshark.co.uk
Not a huge jump in strength, maybe 30 ELO or so.
Changes
Small changes. Minor optimisations. More aggressive null move pruning. IID replaced by IIR (thanks to Ed Schroder for idea). Tweak to futility to make more aggressive or not, depending on if position is improving. Separate search functions when in check. Minor eval tweaks - tuning needed
Happy Christmas to everyone. Let's hope 2021 is a better year, and the world gets on top of the Coronavirus.

Hi - from 1000 games I have had 30 games with lost in time. Most times in won positions (from white or black). Example
[Event "Comb2"] [Site "AGANDO_HP_PC"] [Date "2020.12.26"] [Round "17"] [White "Francesca MAD 0.30"] [Black "Counter 3.5 x64 1CPU"] [Result "0-1"] [ECO "B90"] [WhiteElo "2200"] [BlackElo "2200"] [PlyCount "116"] [EventDate "2020.??.??"] [TimeControl "40/480:40/480:40/480"] 1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. h3 g6 7. g4 Bg7 8. Bg2 O-O 9. Be3 Nc6 10. Qd2 Bd7 11. O-O-O Rc8 12. f4 {0.41/32 22} Na5 {-0.64/17 12} 13. e5 {0.27/32 10} Nc4 {-0.60/19 22} 14. exf6 {0.27/34 10} Nxd2 {-0.28/20 16} 15. fxg7 {-1.31/35 18} Kxg7 {-0.34/20 13} 16. Rxd2 {-1.34/35 11} Rxc3 {-0.29/ 18 16} 17. bxc3 {-0.81/31 18} Qa5 {-0.23/20 17} 18. Kb1 {-0.45/32 10} Qb6+ { -0.15/19 12} 19. Ka1 {-0.37/34 10} Qc7 {-0.17/20 12} 20. Ne2 {-0.55/36 54} Bc6 {-0.11/19 14} 21. Bd5 {-0.54/35 19} e6 {-0.14/20 14} 22. Bxc6 {-0.32/33 11} Qxc6 {-0.17/20 13} 23. Rhd1 {-0.17/35 10} Qe4 {-0.18/21 21} 24. Rd3 {-0.17/35 20} e5 {-0.18/19 28} 25. Re1 {-0.23/30 15} Qc6 {-0.32/18 15} 26. Bd2 {-0.01/34 10} Re8 {-0.14/19 18} 27. fxe5 {-0.01/36 9} dxe5 {-0.59/17 15} 28. Ng3 { 0.00/35 9} Re6 {-0.60/21 15} 29. Ne4 {0.08/34 10} h6 {-0.68/20 22} 30. Rf3 { 0.16/33 12} b6 {-0.60/19 12} 31. h4 {0.25/32 15} Re7 {-0.94/19 20} 32. Bc1 { 0.65/33 15} Qa4 {-1.00/20 12} 33. h5 {0.84/35 14} g5 {-1.07/19 13} 34. Kb1 { 1.24/35 11} Kf8 {-1.47/19 26} 35. Rf6 {0.92/37 14} Re6 {-1.19/21 14} 36. Rxe6 { 1.08/41 14} fxe6 {-1.15/23 11} 37. Nf2 {1.35/37 9} Qb5+ {-0.91/23 11} 38. Bb2 { 1.43/40 8} Qa4 {-0.91/25 13} 39. Rxe5 {1.37/40 9} Qf4 {-1.16/23 23} 40. Nd3 { 1.48/39 13} Qxg4 {-1.17/21 10} 41. Bc1 {1.57/38 17} Ke7 {-1.07/21 12} 42. Be3 { 1.67/37 9} Qxh5 {-1.07/20 9} 43. Bxb6 {1.97/30 8} Qf3 {-0.92/20 9} 44. Kb2 { 1.41/29 16} g4 {-0.62/19 12} 45. Re3 {1.35/31 15} Qd5 {-0.82/20 10} 46. a4 { 1.24/32 10} Qg5 {-0.69/20 11} 47. Re5 {1.47/34 9} Qg8 {-0.80/22 29} 48. Bd4 { 1.52/33 8} g3 {-0.83/20 17} 49. Nc5 {1.66/36 9} h5 {-0.78/21 32} 50. Rxe6+ { 1.92/33 10} Qxe6 {-1.26/22 30} 51. Nxe6 {4.14/42 7} Kxe6 {-2.10/24 28} 52. Ba7 {3.36/42 11} h4 {-2.79/22 24} 53. Bb8 {3.92/42 8} Kd5 {-2.95/25 21} 54. Kb3 { 4.09/42 7} a5 {-2.26/25 8} 55. c4+ {13.53/43 7} Ke4 {-4.44/25 11} 56. c5 { 19.97/45 8} Kd5 {-5.04/26 7} 57. c6 {31.97/46 6} Kxc6 {-14.45/24 15} 58. c4 { 63.26/48 45} Kb7 {time forfeit -16.45/26 13 Weiss überschreitet die Zeit} 0-1

Thanks for flagging.. I will see if I can replicate with the same time controls.

Most of my testing calls a win if one side is ahead / behind by 5 pawns for more than a few moves. I think there may be a bug here which your testing has uncovered. I will look into it!

[silentshark]

Thanks for flagging.. I will see if I can replicate with the same time controls.

Most of my testing calls a win if one side is ahead / behind by 5 pawns for more than a few moves. I think there may be a bug here which your testing has uncovered. I will look into it!

Your prompt got me looking into things, and I've identified a few bugs which (I hope are resolved now). One relates to the losing on time when winning bug. Another one relates to mate score in the hash table. And for good measure, I had a play with the gcc compiler and its funky fsanitize options, which pointed out a couple more issues. More testing now, and hopefully will release 0.30a in a couple of days.

Cheers,
Tom

[tmokonen]

Dumb 1.6:
https://github.com/abulmo/Dumb/releases/tag/1.6

The zip file for the Windows compiles contains two files, supposedly 32 and 64 bits, but the compiles are identical 64 bit compiles.

[silentshark]

Thanks for flagging.. I will see if I can replicate with the same time controls.

Most of my testing calls a win if one side is ahead / behind by 5 pawns for more than a few moves. I think there may be a bug here which your testing has uncovered. I will look into it!

Your prompt got me looking into things, and I've identified a few bugs which (I hope are resolved now). One relates to the losing on time when winning bug. Another one relates to mate score in the hash table. And for good measure, I had a play with the gcc compiler and its funky fsanitize options, which pointed out a couple more issues. More testing now, and hopefully will release 0.30a in a couple of days.

Cheers,
Tom

I have released Francesca 0.30a, which is a bugfix version, as per above. If you happen to have downloaded 0.30, please replace it with 0.30a. Apologies for any inconvenience. Bugs, eh?

You can grab as normal from www.silentshark.co.uk

Cheers,
Tom

[abulmo2]

Dumb 1.6:
https://github.com/abulmo/Dumb/releases/tag/1.6

The zip file for the Windows compiles contains two files, supposedly 32 and 64 bits, but the compiles are identical 64 bit compiles.

Sorry, I did not know there was still some 32-bits Windows in the wild . I have uploaded a new zip file that should contain a 32-bits version of Dumb-1.6. Apart support for FRC, Dumb-1.6 brings no improvement over version 1.5, so I see no much interests in testing it as an orthodox chess engine.
As I enjoy playing with its code, I have been able to squeeze some speed improvements out of it since the release of version 1.6. I am looking if I can improve some search algorithms without adding too much code to get some more ELO and will release a version 1.7 within a few weeks.

Have fun with it.

[Roland Chastain]

Have fun with it.

Hello! For your information, I cannot use the Linux binary of Dumb 1.6.

Code: Select all

[roland@localhost 16]$ ./dumb-1.6
./dumb-1.6: /lib64/libc.so.6: version `GLIBC_2.32' not found (required by ./dumb-1.6)
[roland@localhost 16]$ 

It seems that I have to compile the source. I search how I can install a D compiler...

Code: Select all

make: ldc2: Command not found

[mhouppin]

Have fun with it.

Hello! For your information, I cannot use the Linux binary of Dumb 1.6.

Code: Select all

[roland@localhost 16]$ ./dumb-1.6
./dumb-1.6: /lib64/libc.so.6: version `GLIBC_2.32' not found (required by ./dumb-1.6)
[roland@localhost 16]$ 

It seems that I have to compile the source. I search how I can install a D compiler...

Code: Select all

make: ldc2: Command not found

Hi Roland,
If you have snapd installed on your Linux distro, you can install the D compiler with

Code: Select all

sudo snap install ldc2 --classic

[Roland Chastain]

Hi Roland,
If you have snapd installed on your Linux distro, you can install the D compiler with

Code: Select all

sudo snap install ldc2 --classic

Hi Morgan. Thank you for your answer. But, unfortunately, snapd doesn't seem to be available on my distro (Mageia). I am looking for another solution.

[Roland Chastain]

I am looking for another solution.

I downloaded a compiler here and could compile Dumb 1.6.