Stefan Pohl Computer Chess

private website for chessengine-tests


Latest Website-News (2021/05/11): AB-testrun of Igel 3.0.5 finished: +31 Elo to Igel 3.0.0

 

Update of my AntiDraw Openings collection! New WOMP openings included. 

WOMP = "W"hite "O"ne "M"ore "P"awn Openings. Means, that an opening-line was created, in which the black pawn on a7 is deleted:

1. Na3 a5 2. Nb5 a4 3. Nc3 Nf6 4. Nxa4 Ng8 5. Nc3 Nf6 6. Nb1 Ng8 
Behind these 6 moves, human opening moves (4, 5 and 6 moves deep) out of the Megabase 2021 were added and the endpositions were checked by KomodoDragon 1.0 avx with 10 seconds/move on a AMD Ryzen 12 core machine. There are 7 folders with different WOMP V1 openings-sets, with increasing advantage for white (from +0.90 up to +1.50). Read more in the ReadMe-file in the WOMP-folder in the download-package. Download the AntiDraw openings in the "Anti Draw Openings"- section or right here

 

Some 1000 games quicktests of nnue-nets (SF 210131 default vs. SF 210131 testnet). Latest update: 2021/05/06. Some of the best nets were retested with 3x longer thinkingtime. Look at the results here

Direct download of the strongest net so far (3x longer time) is here

 

Stay tuned.


Stockfish testing

 

Playing conditions:

 

Hardware: Since 20/07/21 AMD Ryzen 3900 12-core (24 threads) notebook with 32GB RAM. Now, 20 games are played simultaneously (!), so from now, each testrun will have 6000 or 7000 games (instead of 5000 before) and will take only 2 days, not 6-7 days as before! From now, all engine-binaries are popcount/avx2, of course, because bmi2-compiles are extremly slow on AMD. To keep the rating-list engine-names consistent, the "bmi2"- or "pext"-extension in the engine-name is still in use for older engines - otherwise ORDO will not calculate all played games by this engine as one engine...

Speed: (singlethread, TurboBoost-mode switched off, chess starting position) Stockfish: 1.3 mn/s, Komodo: 1.1 mn/s

Hash: 256MB per engine

GUI: Cutechess-cli (GUI ends game, when a 5-piece endgame is on the board)

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

Openings: HERT_500 testset (by Thomas Zipproth) (download the file at the "Download & Links"-section or here)

Ponder, Large Memory Pages & learning: Off

Thinking time: 180''+1000ms (= 3'+1'') per game/engine (average game-duration: around  7.5 minutes). One 7000 games-testrun takes about 2 days.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 abrok.eu (except the official SF-release versions, which are taken form the official Stockfish website).

Download BrainFish (and the Cerebellum-Libraries)here

 

To avoid distortions in the Ordo Elo-calculation, from now, only 3x Stockfish (latest official release + the latest 2 dev-versions) and 1x Brainfish are stored in the gamebase (all older engine-versions games will be deleted, every time, when a new version was tested). Stockfish and BrainFish older Elo-results can still be seen in the Elo-diagrams below. BrainFish plays always with the latest Cerebellum-Libraries of course, because otherwise BrainFish = Stockfish.

 

Latest update: 2021/05/11: Igel 3.0.5  (+31 Elo to Igel 3.0.0)

 

(Ordo-calculation fixed to Stockfish 13 = 3723 Elo)

 

See the individual statistics of engine-results here

Download the current gamebase here

 

     Program                    Elo    +    -   Games   Score   Av.Op.  Draws

   1 Stockfish 210406 avx2    : 3732    8    8  7000    76.9 %   3501   43.9 %
   2 Stockfish 210428 avx2    : 3729    7    7  7000    76.1 %   3508   45.9 %
   3 Stockfish 210415 avx2    : 3727    8    8  7000    76.5 %   3501   45.1 %
   4 CFish 12 3xCerebellum    : 3727    9    9  7000    86.1 %   3391   27.3 %
   5 CFish 210324 avx2        : 3725    7    7  7000    76.2 %   3501   45.1 %
   6 Stockfish 13 210218      : 3723    4    4 22000    71.9 %   3533   53.0 %
   7 SF Fat Fritz 2 avx2      : 3719    7    7  8000    73.7 %   3514   49.3 %
   8 CFish 12 avx2            : 3705    9    9  7000    84.6 %   3391   29.1 %
   9 Stockfish 12 200902      : 3685    4    4 26000    78.7 %   3431   38.5 %
  10 SF Fat Fritz 2 github    : 3678    7    7  7000    73.3 %   3484   48.5 %
  11 SF miniNNUE 210307       : 3658    7    7  7000    69.5 %   3501   51.0 %
  12 KomodoDragon 2.0 avx2    : 3651    6    6  8000    73.0 %   3468   46.8 %
  13 KomodoDragon 1.0 avx2    : 3649    3    3 27000    68.4 %   3490   49.2 %
  14 KomodoDragon 2.0 MCTS    : 3566    6    6  7000    62.4 %   3475   53.4 %
  15 Stockfish 11 200118      : 3563    6    6 12000    65.5 %   3435   42.8 %
  16 KomodoDragon 1.0 MCTS    : 3481    6    6  7000    57.6 %   3427   57.1 %
  17 Komodo 14.1 x64          : 3456    6    6  8000    56.3 %   3413   55.6 %
  18 Komodo 14 bmi2           : 3446    4    4 16000    52.4 %   3435   52.9 %
  19 RubiChess 2.1 avx2       : 3444    5    5 12000    40.2 %   3524   53.3 %
  20 Nemorino 6.00 avx2       : 3441    3    3 35000    44.9 %   3488   51.4 %
  21 Houdini 6 pext           : 3439    2    2 45000    46.1 %   3476   47.7 %
  22 Fire 8.1 popc            : 3434    4    4 14000    43.2 %   3490   50.2 %
  23 Fire 8 popc              : 3432    5    5 13000    38.2 %   3535   47.1 %
  24 Pedone 3.1 avx2          : 3426    5    5 12000    40.0 %   3507   50.7 %
  25 Slow Chess 2.5 avx2      : 3422    3    3 25000    39.5 %   3512   46.8 %
  26 Igel 3.0.5 popavx2       : 3416    7    7  7000    38.7 %   3508   57.5 %
  27 Nemorino 6.05 avx2       : 3407    7    7  7000    42.8 %   3468   51.0 %
  28 Ethereal 12.75 avx2      : 3399    3    3 31000    42.4 %   3468   49.2 %
  29 Ethereal 12.62 avx2      : 3393    6    6  8000    49.1 %   3405   54.6 %
  30 Igel 3.0.0 popavx2       : 3385    6    6  8000    36.9 %   3492   53.0 %
  31 Slow Chess 2.4 popc      : 3375    5    5 12000    43.7 %   3429   52.3 %
  32 Ethereal 12.50 popc      : 3361    6    6  7000    45.8 %   3400   55.2 %
  33 RubiChess 2.0 avx2       : 3356    5    5 12000    30.2 %   3527   45.5 %
  34 Slow Chess 2.3 popc      : 3347    5    5 13000    44.2 %   3394   53.8 %
  35 Komodo 14 MCTS           : 3343    7    7  5000    44.4 %   3388   53.4 %
  36 Ethereal 12.25 pext      : 3342    6    6 10000    33.8 %   3486   47.6 %
  37 Slow Chess 2.2 popc      : 3332    6    6  9000    30.9 %   3502   43.2 %
  38 Pedone 3 avx2            : 3331    6    6  8000    33.4 %   3467   46.7 %
  39 Igel 2.9.0 popavx2       : 3329    5    5 11000    34.5 %   3453   50.1 %
  40 RubiChess 1.9dev nnue    : 3322    6    6  8000    31.8 %   3472   45.6 %
  41 Ethereal 12.00 pext      : 3320    7    7  7000    42.1 %   3382   50.6 %
  42 Xiphos 0.6 bmi2          : 3308    4    4 23000    36.9 %   3421   48.6 %
  43 Fire 7.1 popc            : 3308    4    4 20000    33.9 %   3445   48.3 %
  44 Booot 6.5 popc           : 3294    6    6  8000    30.7 %   3460   40.1 %
  45 Minic 2.51 nasc_nutr     : 3285    6    6  7000    31.2 %   3438   45.0 %
  46 rofChade 2.3 bmi2        : 3266    6    6  9000    38.2 %   3357   49.9 %
  47 Minic 3.06 noct_nadir    : 3172    7    7  7000    25.7 %   3364   40.6 %

The version-numbers (180622 for example) of the engines are the date of the latest patch, which was included in the Stockfish sourcecode, not the release-date of the engine-file. Especially the asmFish-engines are often released much later!!

Below you find a diagram of the progress of Stockfish in my tests since August 2020

And below that diagram, the older diagrams.

 

You can save the diagrams (as a JPG-picture (in originial size)) on your PC with mouseclick (right button) and then choose "save image"...

The Elo-ratings of older Stockfish dev-versions in the Ordo-calculation can be a little different to the Elo-"dots" in the diagram, because the results/games of new Stockfish dev-versions - when getting part of the Ordo-calculation - can change the Elo-ratings of the opponent engines and that can change the Elo-ratings of older Stockfish dev-versions (in the Ordo-calculation / ratinglist, but not in the diagram, where all Elo-"dots" are the rating of one Stockfish dev-version at the moment, when the testrun of that Stockfish dev-version was finished).

 

 

 

 

 

 


Sie sind Besucher Nr.