SPCC: Testrun of Torch 2 finished

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

User avatar
pohl4711
Posts: 2698
Joined: Sat Sep 03, 2011 7:25 am
Location: Berlin, Germany
Full name: Stefan Pohl

SPCC: Testrun of Torch 2 finished

Post by pohl4711 »

My UHO-Top15 Ratinglist is the world's first engine-ratinglist, using UHO-openings, and the world's first ratinglist offering additionally Gamepair-statistics.

Ratinglist-testrun of Torch 2 finished.

https://www.sp-cc.de

Also take a look at the EAS-Ratinglist, the world's first engine-ratinglist not measuring strength of engines but engines's style of play:
https://www.sp-cc.de/eas-ratinglist.htm


(Perhaps you have to clear your browsercache (press STRG+SHIFT+DEL) or reload the website))
User avatar
pohl4711
Posts: 2698
Joined: Sat Sep 03, 2011 7:25 am
Location: Berlin, Germany
Full name: Stefan Pohl

Re: SPCC: Testrun of Torch 2 finished

Post by pohl4711 »

Testrun of Torch 2 finished: +24 Elo to Torch 1 and nearly the same result than Torch 1 in the EAS Ratinglist (only 2750 EAS-points difference in the RoundRobin EAS-calculation = nearly nothing, so the new and bigger nnue-net of Torch 2 did not change the playing-style of Torch)

Comparison of 5 months of progress of Torch and Stockfish:

Middle of October 2023:
Torch 1: 3782 Elo, Stockfish 231015: 3825 Elo

Middle of March 2024:
Torch 2: 3806 Elo, Stockfish 240314: 3838 Elo

Stockfish gained +13 Elo, Torch gained +24 Elo in these 5 months. So, Torch was catching up 11 Elo in the last 5 months. To catch up the missing 32 Elo to Stockfish, Torch should need around 15 months from now (3 * 5 months)...if the speed of progress of Torch and Stockfish does not change...

Mention, the Elo-progress and the EAS-scoring of Torch 2 in the full ratinglist is a bit higher than the numbers provided here, because Torch 1 played a lot more games than Torch 2 in the full ratinglist, which drops Elo and EAS of Torch 1 a little bit...
User avatar
pohl4711
Posts: 2698
Joined: Sat Sep 03, 2011 7:25 am
Location: Berlin, Germany
Full name: Stefan Pohl

Re: SPCC: Testrun of Torch 2 finished

Post by pohl4711 »

Here a nice win of Torch 2 vs. Stockfish 16.1 from my testrun. More cool games of my Torch 2-testrun can be found on my website:
https://www.sp-cc.de/view-games-with-sacs.htm
(I use the pgn-viewer by ChessBase: You have to switch off AdBlockers and activate Javascript in your browser, otherwise the pgn-viewer will not work)

In my testrun, the first 6 moves are from my UHO 2022 openings (white always has an advantage). Engines started thinking from move 7 of white. Game ended, because it was 5-piece Syzygy-position on the board.

[pgn]
[Event "UHO Ratinglist"]
[Site "?"]
[Date "2024.03.22"]
[Round "296"]
[White "Torch 2 popavx2"]
[Black "Stockfish 16.1 240224"]
[Result "1-0"]
[ECO "C13"]
[Opening "French"]
[Variation "Albin-Alekhine-Chatard attack, Spielmann Variation"]
[TimeControl "180+1"]
[Annotator "IWS-Tool: 5+ PawnUnits Sacrifice found in this game"]
[Termination "adjudication"]
[PlyCount "93"]
[GameDuration "00:06:09"]
[GameEndTime "2024-03-22T20:01:59.379 Mitteleuropäische Zeit"]
[GameStartTime "2024-03-22T19:55:49.877 Mitteleuropäische Zeit"]

1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. Bg5 Be7 5. e5 Nfd7 6. h4 O-O 7. Qh5 c5 8.
Bd3 g6 9. Qh6 cxd4 10. Nf3 Nxe5 11. Nxe5 dxc3 12. Nxg6 fxg6 13. Bxg6 hxg6
14. Qxg6+ Kh8 15. Rh3 cxb2 16. Rb1 Qb6 17. Qh5+ Kg8 18. Be3 Qb4+ 19. Bd2
Qe4+ 20. Re3 Qxg2 21. Rg3+ Qxg3 22. fxg3 Nd7 23. Qg6+ Kh8 24. Rxb2 b6 25.
Bh6 Rg8 26. Qh5 Nf6 27. Qf7 Ng4 28. Bg5 Bxg5 29. hxg5 e5 30. g6 Rg7 31.
Qxd5 Bb7 32. Qe6 Be4 33. Qxg4 Bxg6 34. c4 Rf8 35. Rh2+ Rh7 36. Rh4 Kg7 37.
Qe6 Rxh4 38. gxh4 Rf5 39. Qe7+ Bf7 40. Qxa7 Rf4 41. h5 Kh6 42. Qxb6+ Kxh5
43. c5 Bxa2 44. c6 Rf6 45. Kd2 Bd5 46. Qd8 Rxc6 47. Qxd5 1-0
[/pgn]
Modern Times
Posts: 3699
Joined: Thu Jun 07, 2012 11:02 pm

Re: SPCC: Testrun of Torch 2 finished

Post by Modern Times »

+24 Elo, good result!
User avatar
pohl4711
Posts: 2698
Joined: Sat Sep 03, 2011 7:25 am
Location: Berlin, Germany
Full name: Stefan Pohl

Re: SPCC: Testrun of Torch 2 finished

Post by pohl4711 »

Yes, but from my tests, we can learn much more: The playing-style of Torch 2 is identical to Torch 1... Here the first 5 of my EAS-ratinglist, the old list, with Torch 1, and the new list, containing Torch 2. As you can see, not only the EAS-Score of Torch 1 and Torch 2 is nearly identical, but the single stats are nearly identical, too. So, the playing style of Torch 1 and 2 is definitly the same (but very different to the playing style of Stockfish (Stockfish plays way more sacrifices and plays much less drawish, but Torch plays a little bit more straightforward for the win))

Code: Select all

                                 bad  avg.win 
Rank  EAS-Score  sacs   shorts  draws  moves  Engine/player 
-------------------------------------------------------------------
   1    200545  24.55%  24.25%  08.74%   70   Stockfish 16.1 240224  
   2    184199  23.79%  23.06%  09.08%   71   Stockfish 240314 avx2  
   3    154012  16.84%  25.58%  13.82%   69   Torch 1 popavx2  
   4    115165  17.44%  19.46%  15.38%   73   KomodoDragon 3.3 avx2  
   5    100235  16.21%  17.42%  16.16%   75   RubiChess 240112 avx2  

                                 bad  avg.win 
Rank  EAS-Score  sacs   shorts  draws  moves  Engine/player 
-------------------------------------------------------------------
   1    199635  24.45%  24.50%  08.87%   70   Stockfish 16.1 240224  
   2    184690  23.98%  23.34%  09.04%   71   Stockfish 240314 avx2  
   3    156760  17.21%  25.90%  12.45%   69   Torch 2 popavx2  
   4    118415  17.43%  19.53%  15.75%   72   KomodoDragon 3.3 avx2  
   5    100559  16.16%  17.61%  16.54%   75   RubiChess 240112 avx2