Stefan Pohl Computer Chess

Latest Website-News (2024/05/14): Testrun of Stockfish 240509 finished: +1 Celo to Stockfish 240428. 

NN-testrun of Lc0 0.31dev 819344 (new new from T80 training) finished: See the result on NN vs SF 15.1 testing

 

Dont forget to take a look at my EAS-Ratinglist (The world's first engine-ratinglist not measuring strength of engines, but engines's style of play).

 

Next testrun: Titan 1.0 - just to see, if Titan 1.0 is strong enough for my UHO-Top15 Ratinglist or not.

 

BUG ALERT There was a bug found in pgn-extract, which affects my EAS-Tool and my other tools: Do not use any pgn-files, containing games, that are starting from a FEN-code. Then pgn-extract can detect false sacs and so will my tools, which are using pgn-extract! If the games in the pgn-file, given to the EAS-Tool, start from move 1 and from the normal chess starting-position, everything works fine - that is the reason, I never recognized this bug, because my games all start from move 1.

 

Stay tuned

UHO-Top15 Engines Ratinglist (+ regular testing of Stockfish Dev-versions)

 

Playing conditions:

 

The 15 strongest engines and the latest StockfishDev version are playing 1000 games vs. each opponent: 15000 games per engine are played. A RoundRobin Tournament with 120000 games overall !

 

Hardware: AMD Ryzen 3900 12-core (24 threads) notebook with 32GB RAM. 

Speed: (singlethread, TurboBoost-mode switched off, chess starting position) Stockfish 16.1: 500 kn/s (when 20 games are running simultaneously)

Hash: 256MB per engine

GUI: Cutechess-cli (GUI ends game, when a 5-piece endgame is on the board, all other games are played until mate or draw by chess-rules (3fold, 50-moves, stalemate, insufficent material))

Tablebases: None for engines, 5 Syzygy for cutechess-cli

Openings: My UHO_2022_6mvs_+120_+129 openings are used (first 500 lines) (part of my UHO 2022 download). 

Ponder, Large Memory Pages & learning: Off

Thinking time: 3min+1sec per game/engine (average game-duration: 7 min 50sec), around 3500-4000 games in 24 hours are played. One testrun (15000 games) takes around 100 hours (=4 days and some hours)

The version-numbers of the Stockfish engines are the date of the latest patch, which was included in the Stockfish sourcecode, not the release-date of the engine-file, written backwards (year,month,day))(example: 200807 = August, 7, 2020). The used SF compile is the AVX2-compile, which is the fastest on my AMD Ryzen CPU. SF binaries are taken from https://github.com/official-stockfish/Stockfish/releases (except the official SF-release versions, which are taken form the official Stockfish website).

Read the explanation, why I chose my testing-conditions, like I did here

 

Latest update: 2024/05/14: Stockfish 240509 (+1 Celo to Stockfish 240428)

 

(Ordo-calculation fixed to Stockfish 16.1 = 3833 Celo. Mention, there are no human or "realistic" Elo numbers in enginechess, since the engines reached superhuman strength, so I decided to use "Celo" (=Computer Elo) instead)

 

See the tournament cross-table as a picture (done by Fritz18 GUI) here

See the individual statistics of engine-results here

See the Engines Aggressiveness Score Ratinglist here

See the Gamepair-statistics of the UHO-Top15 Ratinglist here

See the full UHO Ratinglist (with EAS-Ratinglist): here

Download the current gamebase, the full-list gamebase and the archive here

See the old SPCC-Ratinglist (full list) and EAS-List from 2020 until 2023/08/28: here

 

(best Stockfish Celo: Stockfish 240424: 3844 Celo, latest full release (Stockfish 16.1) had 3833 Celo)

(Read the explanation of the Lc0-entry in the Ratinglist here).

     Program                    Celo    +    - Games    Score   Av.Op. Draws

   1 Stockfish 240509 avx2    : 3842    4    4 15000    71.6%   3674   47.5%
   2 Stockfish 16.1 240224    : 3833    4    4 15000    70.5%   3674   47.8%
   3 Torch 2 popavx2          : 3801    4    4 15000    66.5%   3676   48.0%
  [  Lc0 0.31dev T3-2815      : 3781   15   15  1000 (NN vs SF15.1 testing)]
   4 Berserk 13 avx2          : 3749    4    4 15000    59.3%   3680   48.9%
   5 KomodoDragon 3.3 avx2    : 3738    4    4 15000    57.8%   3681   49.7%
   6 Ethereal 14.38 avx2      : 3702    4    4 15000    52.7%   3683   49.0%
   7 Obsidian 12.0 avx2       : 3697    4    4 15000    52.0%   3683   50.7%
   8 Caissa 1.18 avx2         : 3680    4    4 15000    49.5%   3685   49.2%
   9 RubiChess 240112 avx2    : 3656    4    4 15000    46.1%   3686   49.0%
  10 PlentyChess 1.0 avx2     : 3634    4    4 15000    42.9%   3688   50.0%
  11 Alexandria 6.1.0 avx2    : 3613    4    4 15000    40.0%   3689   49.4%
  12 Seer 2.8.0 avx2          : 3610    4    4 15000    39.5%   3689   49.0%
  13 Uralochka 3.41a avx2     : 3601    4    4 15000    38.2%   3690   48.3%
  14 CSTal 2.0 avx2           : 3600    4    4 15000    38.2%   3690   51.0%
  15 Rebel 16.3 avx2          : 3599    4    4 15000    38.0%   3690   49.8%
  16 Clover 6.1 avx2          : 3592    4    4 15000    37.0%   3690   50.2%

Games        : 120000 (finished)

White Wins   : 58337 (48.6 %)
Black Wins   : 2600 (2.2 %)
Draws        : 59063 (49.2 %)

Below the UHO-Top15 Ratinglist, recalculated with Gamepairs (using my Gamepair Rescoring Tool V1.5), realizing Vondele's (Stockfish Maintainer) idea: "Thinking uniquely in game pairs makes sense with the biased openings used these days. While pentanomial makes  sense it is a bit complicated so we could simplify and score game pairs only (not games) as W-L-D (a traditional score of 2-0, or 1.5-0.5 is just a W)."

See the head-to-head gamepair-statistics of the UHO-Top15 Ratinglist here

   # PLAYER                   :  RATING  ERROR  PLAYED     W     D     L   (%)  CFS(%)
   1 Stockfish 240509 avx2    :    3874     13    7500  6271  1099   130  90.9     100
   2 Stockfish 16.1 240224    :    3833   ----    7500  6041  1238   221  88.8     100
   3 Torch 2 popavx2          :    3708     12    7500  5209  1718   573  80.9     100
   4 Berserk 13 avx2          :    3546     12    7500  3908  2269  1323  67.2     100
   5 KomodoDragon 3.3 avx2    :    3523     12    7500  3684  2388  1428  65.0     100
   6 Ethereal 14.38 avx2      :    3426     12    7500  2766  2691  2043  54.8     100
   7 Obsidian 12.0 avx2       :    3412     12    7500  2607  2777  2116  53.3     100
   8 Caissa 1.18 avx2         :    3375     12    7500  2325  2711  2464  49.1     100
   9 RubiChess 240112 avx2    :    3327     12    7500  1878  2763  2859  43.5     100
  10 PlentyChess 1.0 avx2     :    3269     12    7500  1367  2792  3341  36.8     100
  11 Alexandria 6.1.0 avx2    :    3226     12    7500  1084  2625  3791  32.0     100
  12 Seer 2.8.0 avx2          :    3213     12    7500  1007  2565  3928  30.5     100
  13 CSTal 2.0 avx2           :    3184     12    7500   836  2456  4208  27.5      55
  14 Rebel 16.3 avx2          :    3184     13    7500   813  2492  4195  27.5      65
  15 Uralochka 3.41a avx2     :    3182     12    7500   797  2494  4209  27.3     100
  16 Clover 6.1 avx2          :    3159     13    7500   733  2270  4497  24.9     ---

------------------------------------------------------------------- 
--- Number of all Gamepairs          : 60000 
--- Number of drawn Gamepairs overall: 18674 (= 31.12%) 
--- Number of 1:1 drawn Gamepairs    : 8839  (= 14.73%) 
--- Number of 2-draws drawn Gamepairs: 9835  (= 16.39%) 
------------------------------------------------------------------- 

Aborted testruns, because the tested engine was too weak to enter the UHO-Top15 Ratinglist:

Minic 3.40, Stormphrax 4.1.0, Booot 7.3, Black Marlin 9.0, Equisetum 1.0

Below you find a diagram of the progress of Stockfish in my tests since February 2023. 

And below that diagram, the older diagrams.

 

You can save the diagrams on your PC with mouseclick (right button) and then choose "save image"...