Sapeli 1.0 - New chess engine

[Roland Chastain]

Hello! I updated my "all versions" package (for Windows), after I found a way to build without dependency to cygwin1.dll. The package include 32 bits binaries, so maybe it can be useful to someone, in addition to the binaries offered by Günther.

I started from v1.17, since I couldn't compile the previous versions.

Regards.

Roland

[JohnWoe]

Thanks for the builds!
Added them here: https://github.com/SamuraiDangyo/Sapeli ... naries/bin

I think 1.0 -> 1.17 were a bit messy code.

The latest version 1.69 is a lot easier to read. Made mgen much simpler. Much easier to read its code. Not published it yet. It's gonna be a little bit faster.

Edit:
Released: https://github.com/SamuraiDangyo/Sapeli ... /tag/v1.69
Source code: https://github.com/SamuraiDangyo/Sapeli

+ Slightly faster
+ Easier to read source code

[Ras]

The UCI spec says that "isready" must also be answered during search - but only "stop" is evaluated. This might make GUIs kill Sapeli because they may assume it is hanging.

[Roland Chastain]

Thanks for the builds!
Added them here: https://github.com/SamuraiDangyo/Sapeli ... naries/bin

You're welcome. Maybe you could make some cleaning in your bin directory (that is, delete the previous versions of my package)? And by the way, it isn't really important, but my first name is Roland, and not Ronald.

I will update the package tonight, to include v1.69. The link will be the same.[edit] Done.[/edit]

[Guenther]

...
The latest version 1.69 is a lot easier to read. Made mgen much simpler. Much easier to read its code. Not published it yet. It's gonna be a little bit faster.

Edit:
Released: https://github.com/SamuraiDangyo/Sapeli ... /tag/v1.69
Source code: https://github.com/SamuraiDangyo/Sapeli

+ Slightly faster
+ Easier to read source code

Updated my compilations too.
http://rwbc-chess.de/download.htm

[JohnWoe]

The UCI spec says that "isready" must also be answered during search - but only "stop" is evaluated. This might make GUIs kill Sapeli because they may assume it is hanging.

Thanks! I didn't know that. Fixed.

Thanks for the builds!
Added them here: https://github.com/SamuraiDangyo/Sapeli ... naries/bin

You're welcome. Maybe you could make some cleaning in your bin directory (that is, delete the previous versions of my package)? And by the way, it isn't really important, but my first name is Roland, and not Ronald.

I will update the package tonight, to include v1.69. The link will be the same.[edit] Done.[/edit]

My bad. Fixed.

Been testing against Stockfish that there's no illegal activity in 40/1s Chess960 games. Sometimes Sapeli even scores a win against the latest Stockfish.
[pgn][Event "?"]
[Site "?"]
[Date "2020.01.17"]
[Round "88"]
[White "Stockfish 160120 64"]
[Black "Sapeli 1.70"]
[Result "0-1"]
[TimeControl "40/1"]
[Variant "fischerandom"]
[PlyCount "150"]
[FEN "rkbqrbnn/pppppppp/8/8/8/8/PPPPPPPP/RKBQRBNN w EAea - 0 1"]
[SetUp "1"]

{--------------
r k b q r b n n
p p p p p p p p
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
P P P P P P P P
R K B Q R B N N
white to play
--------------}
1. e4 {+0,72/8 1,1}
{s}
1... d5 {-0,04/6 3}
{s}
2. Qh5 {+1,51/8 1,1}
{s}
2... Nf6 {+0,81/6 3}
{s}
3. Qd1 {-1,27/9 1,0}
{s}
3... dxe4 {+0,90/6 3}
{s}
4. Ng3 {-0,75/10 1,1}
{s}
4... Qd4 {+1,05/6 2,9}
{s}
5. d3 {-0,53/9 1,2}
{s}
5... exd3 {+0,99/5 2,8}
{s}
6. Nf3 {-0,19/8 1,1}
{s}
6... Qxf2 {+1,20/5 2,8}
{s}
7. Qxd3 {-0,36/9 1,2}
{s}
7... Qb6 {+0,90/5 2,6}
{s}
8. Be3 {-0,30/8 1,1}
{s}
8... Qd6 {+0,90/5 2,6}
{s}
9. Ne4 {-0,44/7 1,1}
{s}
9... Nxe4 {+1,07/5 2,5}
{s}
10. Qxe4 {+0,11/9 1,1}
{s}
10... g6 {+0,97/5 2,4}
{s}
11. Bf4 {+3,40/9 1,2}
{s}
11... Qc5 {+1,02/5 2,2}
{s}
12. Re3 {+3,57/8 1,1}
{s}
12... Bf5 {+1,41/4 2,2}
{s}
13. Rc3 {-2,07/8 1,2}
{s}
13... Qxc3 {+3,38/6 2,2}
{s}
14. Qa4 {-4,65/11 1,2}
{s}
14... Qc6 {+5,94/6 2,1}
{s}
15. Bb5 {-2,45/10 1,2}
{s}
15... Qxc2+ {+3,42/6 2,2}
{s}
16. Qxc2 {-1,80/1 0,1}
{s}
16... Bxc2+ {+3,42/6 2,2}
{s}
17. Kxc2 {-3,94/11 2,3}
{s}
17... Rd8 {+3,56/6 2,2}
{s}
18. Nd2 {-4,19/11 2,4}
{s}
18... a6 {+3,79/5 2,2}
{s}
19. Bc4 {-4,86/11 2,6}
{s}
19... b5 {+3,81/5 2,2}
{s}
20. Be2 {-4,26/10 2,7}
{s}
20... Rd5 {+3,88/5 2,1}
{s}
21. Nb3 {-2,99/10 2,1}
{s}
21... O-O-O {+3,97/5 2,1}
{s}
22. a4 {-2,61/11 3}
{s}
22... g5 {+4,11/5 2,2}
{s}
23. Bg3 {-2,40/11 2,5}
{s}
23... Ng6 {+4,11/5 2,1}
{s}
24. axb5 {-1,27/11 1,9}
{s}
24... axb5 {+3,89/6 2,2}
{s}
25. Bf3 {-2,44/11 3}
{s}
25... e6 {+2,89/6 2,2}
{s}
26. Ra8+ {-2,58/13 4}
{s}
26... Kd7 {+2,93/7 2,3}
{s}
27. Bxd5 {-3,00/11 1,8}
{s}
27... exd5 {+2,69/7 2,2}
{s}
28. Ra6 {-3,38/12 4}
{s}
28... f5 {+3,05/6 2,3}
{s}
29. Nd4 {-3,70/12 4}
{s}
29... f4 {+3,12/6 2,2}
{s}
30. Bf2 {-4,23/10 2,1}
{s}
30... Bc5 {+3,17/6 2,3}
{s}
31. Bg1 {-4,54/11 4}
{s}
31... Re8 {+3,39/6 2,3}
{s}
32. Ra5 {-5,54/11 4}
{s}
32... c6 {+3,47/5 2,2}
{s}
33. b4 {-5,26/13 3}
{s}
33... Bxd4 {+3,98/6 2,3}
{s}
34. Bxd4 {-5,42/12 1,9}
{s}
34... Re2+ {+3,12/5 2,3}
{s}
35. Kd3 {-5,65/12 4}
{s}
35... Rxg2 {+3,42/6 2,3}
{s}
36. Ra7+ {-4,30/11 1,7}
{s}
36... Ke6 {+3,37/7 2,3}
{s}
37. Rc7 {-6,37/12 5}
{s}
37... Rxh2 {+4,12/6 2,2}
{s}
38. Rxc6+ {-7,31/11 4}
{s}
38... Kf5 {+4,07/6 2,3}
{s}
39. Rf6+ {-6,93/11 4}
{s}
39... Kg4 40. Rd6 {-7,39/12 5}
{s}
40... f3 {+4,30/7 3}
{s}
41. Be3 {-7,87/10 1,1}
{s}
41... Ne7 {+4,32/6 3}
{s}
42. Rd7 {-6,84/9 1,1}
{s}
42... Nc6 {+4,46/7 3}
{s}
43. Kc3 {-7,45/9 1,0}
{s}
43... f2 {+5,19/7 3}
{s}
44. Bxf2 {-6,92/10 1,2}
{s}
44... Rxf2 {+5,28/7 3}
{s}
45. Rxh7 {-7,54/8 1,1}
{s}
45... d4+ {+6,53/7 3}
{s}
46. Kd3 {-7,20/9 1,1}
{s}
46... Rf4 {+6,55/7 3}
{s}
47. Rb7 {-6,69/9 1,2}
{s}
47... Nxb4+ {+6,69/7 2,9}
{s}
48. Ke2 {-7,66/9 1,2}
{s}
48... d3+ {+6,75/7 2,7}
{s}
49. Ke3 {-7,42/9 1,1}
{s}
49... Rf3+ {+6,77/7 2,6}
{s}
50. Kd2 {-6,69/11 1,1}
{s}
50... Rf2+ {+6,95/7 2,6}
{s}
51. Kc3 {-7,35/11 1,3}
{s}
51... Rf1 {+5,74/6 2,5}
{s}
52. Kxb4 {-4,61/9 1,7}
{s}
52... d2 {+6,18/7 2,4}
{s}
53. Rd7 {-54,54/12 2,2}
{s}
53... d1=R {+6,29/7 2,2}
{s}
54. Re7 {-7,54/10 2,4}
{s}
54... Rb1+ {+7,46/6 2,2}
{s}
55. Kc3 {-7,66/10 2,7}
{s}
55... b4+ {+7,70/6 2,4}
{s}
56. Kc4 {-7,61/10 2,6}
{s}
56... Rf4+ {+7,75/6 2,4}
{s}
57. Kd3 {-7,71/12 2,9}
{s}
57... b3 {+7,80/6 2,3}
{s}
58. Kc3 {-7,94/10 2,9}
{s}
58... Rf3+ {+7,83/6 2,3}
{s}
59. Kb4 {-7,75/9 1,4}
{s}
59... b2 {+7,88/6 2,3}
{s}
60. Re2 {-47,96/12 3}
{s}
60... Kf5 {+8,03/6 2,3}
{s}
61. Rd2 {-16,94/13 3}
{s}
61... g4 {+8,31/6 2,4}
{s}
62. Rd5+ {-19,61/12 3}
{s}
62... Ke6 {+8,32/7 2,3}
{s}
63. Rd2 {-18,87/13 2,8}
{s}
63... g3 {+8,36/6 2,3}
{s}
64. Re2+ {-10,10/15 3}
{s}
64... Kd5 {+8,37/7 2,4}
{s}
65. Rd2+ {-18,45/20 3}
{s}
65... Ke4 {+8,39/7 2,4}
{s}
66. Re2+ {-18,62/16 3}
{s}
66... Re3 {+8,47/7 2,4}
{s}
67. Rc2 {-65,55/13 3}
{s}
67... Kf3 {+11,90/7 2,4}
{s}
68. Kb5 {-72,40/13 3}
{s}
68... g2 {+11,91/6 2,3}
{s}
69. Kb4 {-83,32/11 3}
{s}
69... g1=Q {+16,75/6 2,4}
{s}
70. Kb5
{-M22/10 0.031s}
70... Rb3+ {+19,78/5 2,3}
{s}
71. Kc4
{-M8/16 0.015s}
71... Qg8+ {+24,31/5 2,4}
{s}
72. Kd4
{-M8/20 0.014s}
72... Rd1+ {+24,78/5 2,3}
{s}
73. Rd2
{-M6/24 0.013s}
73... Rxd2+ {+30,10/4 2,5}
{s}
74. Ke5
{-M4/245 0.014s}
74... Rb6 {+104,85/4 1,5}
{s}
75. Kf5
{-M2/1 0s}
75... Qd5#
{+104.85/3 0.001s, Black mates} 0-1
[/pgn]

Of course Stockfish kills Sapeli in the long run

Code: Select all

Score of Sapeli 1.70 vs Stockfish 160120 64: 4 - 523 - 3  [0.010] 530
Elo difference: -791.75 +/- 179.03
Finished match

[Roland Chastain]

My bad. Fixed.

My main preoccupation wasn't relative to the spelling of my name, but to the many packages including the same files. As I provide to you an updated ZIP archive with all the versions, it doesn't seem useful to keep the previous versions of the package. But, of course, it's up to you.

[JohnWoe]

I decided to publish Sapeli 1.70

Release: https://github.com/SamuraiDangyo/Sapeli ... /tag/v1.70
Master: https://github.com/SamuraiDangyo/Sapeli

Changes:
Removed dev branch.
Some bug fixes + cleanups + speedups.
data.h -> Can be modified by users.
consts.h -> Can't be. Sapeli will crash if you do.
Improved code so it's easier to read.

1+0.01 960. Against SF 11. No time losses / illegal moves / crashes. Score is pretty bad tho
Summary: https://github.com/SamuraiDangyo/bins/b ... _11_64.txt

Code: Select all

Score of Sapeli 1.70 vs Stockfish 11 64: 14 - 980 - 6  [0.017] 1000
Elo difference: -704.84 +/- 85.33
Finished match

I wanted to add ponder to this version. But the UCI spec is so awful I decided not to implement it here. The only pondering Sapeli (source code) is found here: https://github.com/SamuraiDangyo/Sapeli ... ponder.zip

[Guenther]

I have omitted 1.70 by accident and already uploaded 1.71 (though not released by yourself).
BTW I have created a single download now, which contains all my compilations.

Due to server migration it might take some time until it is reachable from the new server.
(still directing to the old IP here, despite request was sent 6 hours ago and TTL was announced as 4 hours...)

Edit:
While writing this it finally worked ;-)

[JohnWoe]

Thanks for the builds!
Added them here: https://github.com/SamuraiDangyo/Sapeli ... ersions.7z

Yes, I never released 1.71. But it is pretty much the same as 1.70. I added a few const modifiers and documented the code. So it is pretty the same as 1.70. I'll move to 1.72 in the future.

I'm pretty happy with 1.71 atm. On my laptop I got bench 6.6M NPS. I have never seen so high numbers. Seems to have no problems converting KNBK and KQKR into wins. This increased speed seems to help as Sapeli can complete higher depths. Especially in endgames.

I also uploaded some .epd here if anybody finds them useful. I sometimes run these through Sapeli:

WAC: https://github.com/SamuraiDangyo/bins/b ... er/wac.epd
Silver 201 test suite: https://github.com/SamuraiDangyo/bins/b ... er_wac.epd
Bratko-Kopec test suite: https://github.com/SamuraiDangyo/bins/b ... -suite.epd
Colditz-test suite: https://github.com/SamuraiDangyo/bins/b ... -suite.epd
Hard-Talkchess-2020-set: https://github.com/SamuraiDangyo/bins/b ... 20-set.epd