ShashChess

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

Moderator: Ras

supernova
Posts: 87
Joined: Mon Apr 15, 2024 8:30 pm
Full name: Arthur Matheus

Re: ShashChess

Post by supernova »


The results continue unconvicing about Version 36. My testing at a 2-minute 3-second increment time control, Version 36 has achieved less favorable results against a Stockfish previous versions (knowing that 100 games cannot draw definitive conclusions).

ShashChess 36 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Stockfish 11 6t1024h : 100 : 38+ : 51= : 11- : 63.5%
ShashChess 36 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Stockfish 12 6t1024h : 100 : 13+ : 82= : 5- : 54.0%
ShashChess 36 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Stockfish 13 6t1024h : 99 : 3+ : 77= : 19- : 41.92%
ShashChess 36 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : Stockfish 14 6t1024h : 30 : 0+ : 23= : 7- : 38.33%

UCI Options:
1. **Debug Log File**: `<empty>`
2. **NumaPolicy**: auto
3. **Clear Hash**: (button)
4. **Ponder**: (checkbox unchecked)
5. **MultiPV**: 1
6. **Move Overhead**: 10
7. **Minimum Thinking Time**: 100
8. **Slow Mover**: 100
9. **nodestime**: 0
10. **UCI_Chess960**: (checkbox unchecked)
11. **UCI_ShowWDL**: (checkbox unchecked)
12. **CTG/BIN Book 1 File**: `<empty>`
13. **Book 1 Width**: 1
14. **Book 1 Depth**: 255
15. **(CTG) Book 1 Only Green**: (checkbox unchecked)
16. **CTG/BIN Book 2 File**: `<empty>`
17. **Book 2 Width**: 1
18. **Book 2 Depth**: 255
19. **(CTG) Book 2 Only Green**: (checkbox unchecked)
20. **SyzygyPath**: `F:\arena\TB`
21. **SyzygyProbeDepth**: 1
22. **Syzygy50MoveRule**: (checkbox unchecked)
23. **SyzygyProbeLimit**: 7
24. **EvalFile**: `nn-1111cfea1111.nnue`
25. **EvalFileSmall**: `nn-37f18f62d772.nnue`
26. **Full depth threads**: 0
27. **Persisted learning**: Standard
28. **Read only learning**: (checkbox unchecked)
29. **Experience Book**: (checkbox checked)
30. **Experience Book Max Moves**: 100
31. **MCTS**: (checkbox checked)
32. **MCTSThreads**: 1
33. **MCTS Multi Strategy**: 20
34. **MCTS Multi MinVisits**: 5
35. **LiveBook Proxy Url**: `<empty>`
36. **LiveBook Proxy Diversity**: (checkbox unchecked)
37. **LiveBook Lichess Games**: (checkbox unchecked)
38. **LiveBook Lichess Masters**: (checkbox unchecked)
39. **LiveBook Lichess Player**: `<empty>`
40. **LiveBook Lichess Player Color**: Both
41. **LiveBook ChessDB**: (checkbox checked)
42. **LiveBook Depth**: 255
43. **ChessDB Tablebase**: (checkbox checked)
44. **Lichess Tablebase**: (checkbox checked)
45. **ChessDB Contribute**: (checkbox checked)
46. **Variety**: Off
47. **Concurrent Experience**: (checkbox checked)
48. **High Tal**: (checkbox checked)
49. **Middle Tal**: (checkbox checked)
50. **Low Tal**: (checkbox unchecked)
51. **Capablanca**: (checkbox checked)
52. **Low Petrosian**: (checkbox unchecked)
53. **Middle Petrosian**: (checkbox checked)
54. **High Petrosian**: (checkbox checked)
supernova
Posts: 87
Joined: Mon Apr 15, 2024 8:30 pm
Full name: Arthur Matheus

Re: ShashChess

Post by supernova »

Hello,

I continue experiencing problems with Shaschess v36 and Brainlearn v29, as it has been losing many games due to time. This was not an issue with Shashchess v35.3. I have noticed a regression in Shaschess v36 compared to Shashchess v35.3 regarding chess results. I plan to stop using this version after the current Gauntlet, as it is resulting in too many time-related losses. I hope that a fix will be released soon.

Best regards.

Image

ShashChess 36 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h :
Brainlearn 29 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : 55 : 5+ : 45= : 5- : 50.0%
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h : 55 : 1+ : 43= : 11- : 40.91%
Stockfish 11 6t1024h : 100 : 38+ : 51= : 11- : 63.5%
Stockfish 12 6t1024h : 100 : 13+ : 82= : 5- : 54.0%
Stockfish 13 6t1024h : 99 : 3+ : 77= : 19- : 41.92%
Stockfish 14 6t1024h : 100 : 5+ : 82= : 13- : 46.0%
Stockfish 14.1 6t1024h : 55 : 2+ : 45= : 8- : 44.55%
Stockfish 15.1 6t1024h : 55 : 0+ : 44= : 11- : 40.0%
Vafra 14.12.2 6t1024h : 54 : 5+ : 41= : 8- : 47.22%
Vanilla 15 6t1024h : 100 : 4+ : 81= : 15- : 44.5%

Modern Times
Posts: 3710
Joined: Thu Jun 07, 2012 11:02 pm

Re: ShashChess

Post by Modern Times »

The fix for the time losses is to increase the Move Overhead from 10 to something like 100.
supernova
Posts: 87
Joined: Mon Apr 15, 2024 8:30 pm
Full name: Arthur Matheus

Re: ShashChess

Post by supernova »

Modern Times wrote: Fri Oct 04, 2024 5:51 pm The fix for the time losses is to increase the Move Overhead from 10 to something like 100.
Thank you for the tip. I will try from now. However, It looks like the time losses can be related to a new issue too. The engine stopped working. I had not seen this problem in V35.3.
Modern Times
Posts: 3710
Joined: Thu Jun 07, 2012 11:02 pm

Re: ShashChess

Post by Modern Times »

Yes, could be something else going on as well.
supernova
Posts: 87
Joined: Mon Apr 15, 2024 8:30 pm
Full name: Arthur Matheus

Re: ShashChess

Post by supernova »

Modern Times wrote: Fri Oct 04, 2024 10:56 pm Yes, could be something else going on as well.
I have adhered to your guidelines, but I am still experiencing time-related losses. I attempted to record the logs, but the engine resets them each time it is restarted after a game. I have stopped because the reponsible of this derivative is unresponsive in Github. Best regards.

Image

User avatar
Graham Banks
Posts: 44161
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: ShashChess

Post by Graham Banks »

supernova wrote: Tue Oct 08, 2024 2:35 am
Modern Times wrote: Fri Oct 04, 2024 10:56 pm Yes, could be something else going on as well.
I have adhered to your guidelines, but I am still experiencing time-related losses. I attempted to record the logs, but the engine resets them each time it is restarted after a game. I have stopped because the reponsible of this derivative is unresponsive in Github. Best regards.

Image

I've seen move overhead=1000 suggested. Try that. Never had any time loss issues with Shashchess here.
gbanksnz at gmail.com
Modern Times
Posts: 3710
Joined: Thu Jun 07, 2012 11:02 pm

Re: ShashChess

Post by Modern Times »

Yes 100 was just a starting point. Especially if that is Arena. Maybe try 300 next.
supernova
Posts: 87
Joined: Mon Apr 15, 2024 8:30 pm
Full name: Arthur Matheus

Re: ShashChess

Post by supernova »

Graham Banks wrote: Tue Oct 08, 2024 2:37 am I've seen move overhead=1000 suggested. Try that. Never had any time loss issues with Shashchess here.
Hello, I hope these changes will address the issues we've been facing
  • I've increased the overhead to 1000.
  • I've added the executable to the Antivirus permitted list of files.
So far, I've only experienced one loss of time. I appreciate your assistance both Modern Times and Graham Banks. Thanks.

Regards.
supernova
Posts: 87
Joined: Mon Apr 15, 2024 8:30 pm
Full name: Arthur Matheus

Re: ShashChess

Post by supernova »

Hello everyone,

I have noticed a couple of time losses. The results indicate a decline compared to the previous version:
ShashChess 35.3 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h
ShashChess 36 MCTS.1-PL.On-SS.AllOn-LiveBook.On.D255V0 6t1024h

Results: 63.0/100 with a score of +27-1=72 for v35.3.

I will pause the tests for now.

Thank you all!