Stefan Pohl Computer Chess

Home of famous UHO openings and EAS Ratinglist


SPCC Super 3 Tournament

 

 

Latest update: 2024/07/23 (next updates will follow every 7-10 days...)

Tournament overall runtime (average: 47.2 games per day are played): 71 days

Download all played games since start (2024/05/13) here

 

An endless RoundRobin-tournament with 3 engines, which are at the same level of strength (around 3400 Elo) but are completely different in their inner structure and their way of thinking.

Why? The strongest engine since more a decade (Stockfish) is open source, so many, many other engines are (at least) "inspired" by Stockfish... And additionally, a lot of engines (including Stockfish!) are using Lc0-training-data for building their neural nets: The high-end computerchess has become very incestuous... To say this clear here: This is not good or bad, it is just the reality of high-end computerchess in these days.

So, IMHO, it is very interesting to run a tournament with engines, which are completely different, not only in their playing-style, but also in their inner structure and way of thinking, but on a close level of playing-strength.

The Super 3 tournament is not about the results (as you can see below, all 3 engines are at the same level of stregth), but about generating interesting enginegames.

             | Search    | Evaluation   | nodes per second (early middlegame)
-------------|-----------|--------------|-------------------------------------
Lc0 CPU      | MCTS      | float-neural |      1.100
-------------|-----------|--------------|-------------------------------------
Revenge 1.0  | AlphaBeta | int-neural   | 11.000.000 (10.000x faster than Lc0)
-------------|-----------|--------------|-------------------------------------
Komodo 14.1  | AlphaBeta | Handcrafted  | 19.000.000 (17.300x faster than Lc0)
-------------|-----------|--------------|-------------------------------------

As you can see, these 3 engines have nothing in common, considering their way of thinking. Komodo 14.1 uses a classical handcrafted evaluation, Lc0 CPU uses a (float) neural net, and Revenge 1 uses a (integer) nnue-net, like most modern engines in these days. Because floating-point calculations are brutally slow on CPUs, Lc0 CPU is way slower than the 2 opponents... (this is the reason, why Lc0 normally runs on the GPU, not the CPU). If you want to learn more about the neural-net of Lc0 and about nnue-nets, I recommend this e-book by Dominik Klein, which can be downloaded for free as a PDF-file.

And, additionally, Lc0 uses a complete different search (MCTS). And Revenge 1 is one of the most aggressive playing engines of all time (see the EAS-Ratinglist below).

 

HardwareAMD Ryzen 7840HS 8-core (16 threads) notebook with 32GB RAM. Turboboost off.

Speed: See above. Each engine uses 14 threads, when thinking (Lc0 cpu dnll has the UCI option "Threads" like any normal CPU-engine, so it uses the CPU like all other engines) - the GPU stays (of course) unused.

Hash: 8 GB per engine (20.000.000 NNCachesize for Lc0 - enough for storing all evaluated positions of a complete game)

GUI: CutechessGUI (GUI ends game, when a 6-piece endgame is on the board, all other games are played until mate or draw by chess-rules (3fold, 50-moves, stalemate))

Tablebases: None for engines, 6 Syzygy for CutechessGUI

Openings: My UHO_2024_8mvs_+085_+094.pgn openings are used (randomly mixed, each opening repeated with reversed colors, of course (=Gamepairs))

Ponder, Large Memory Pages & learning: Off

Thinking time: 10min+5sec per game/engine (average game-duration: 30 minutes), so only 50 games are played in 24 hours = high quality enginechess

 

Here you can see the shortest wins with sacrifices, played between the latest two site-updates, filtered by my Interesting Wins Search Tool. Download this cool tool in the "Downloads & Links" section or right here

Many thanks to ChessBase for the pgn-replayer tool, which is very easy to use (only 3 lines of code!) and very powerful - use the fan (propeller?)-icon right near the arrows below the chessboard, to start and stop the online-analyzing with the Fritz-engine! Perhaps you have to clear your browser-cache to see the latest games - otherwise the pgn-replayer does not update the games correctly...if you can not see the chessboard, check, that your browser has Javascript activated or if an AdBlocker is the problem.

 

 

 

 

Below the results (first normal Celos (by ORDO), followed by gamepair-rescored Celos, followed by EAS-Ratinglist).

 

     Program               Celo    +    - Games    Score   Av.Op. Draws

   1 Lc0 791921 CPU      : 3450    7    7  2220    51.1%   3442   48.1%
   2 Komodo 14.1 HCE     : 3445    7    7  2220    50.1%   3445   49.5%
   3 Revenge 1.0 avx2    : 3439    7    7  2220    48.8%   3448   48.9%


Games        : 3330 (finished)

White Wins   : 1605 (48.2 %)
Black Wins   : 98    (2.9 %)
Draws        : 1627 (48.9 %)


Gamepairs:

   # PLAYER              :    Celo  Error   Pairs    W    D    L   (%)  CFS(%)
   1 Lc0 791921 CPU      :    3450   ----    1110  295  564  251  52.0      92
   2 Komodo 14.1 HCE     :    3438     16    1110  272  555  283  49.5      71
   3 Revenge 1.0 avx2    :    3434     17    1110  259  559  292  48.5     ---


------------------------------------------------------------------- 
--- Number of all Gamepairs          : 1665 
--- Number of drawn Gamepairs overall: 839 (= 50.39%) 
--- Number of 1:1 drawn Gamepairs    : 415 (= 24.92%) 
--- Number of 2-draws drawn Gamepairs: 424 (= 25.47%) 
------------------------------------------------------------------- 

 

Head to head statistics:

 

1) Lc0 791921 CPU   3450 :   1110 (+295,=564,-251),  52.0 %

   vs.                    :  games (   +,   =,   -),   (%) :   Diff,   SD, CFS (%)
   Komodo 14.1 HCE        :    555 ( 155, 280, 120),  53.2 :    +12,    8,   91.8
   Revenge 1.0 avx2       :    555 ( 140, 284, 131),  50.8 :    +16,    9,   97.0

 

2) Komodo 14.1 HCE  3438 :   1110 (+272,=555,-283),  49.5 %

   vs.                    :  games (   +,   =,   -),   (%) :   Diff,   SD, CFS (%)
   Lc0 791921 CPU         :    555 ( 120, 280, 155),  46.8 :    -12,    8,    8.2
   Revenge 1.0 avx2       :    555 ( 152, 275, 128),  52.2 :     +5,    8,   71.0

 

3) Revenge 1.0 avx2 3434 :   1110 (+259,=559,-292),  48.5 %

   vs.                    :  games (   +,   =,   -),   (%) :   Diff,   SD, CFS (%)
   Lc0 791921 CPU         :    555 ( 131, 284, 140),  49.2 :    -16,    9,    3.0
   Komodo 14.1 HCE        :    555 ( 128, 275, 152),  47.8 :     -5,    8,   29.0


 

Here the EAS-Ratinglist, calculated by my EAS-Tool:

                                 bad  avg.win 
Rank  EAS-Score  sacs   shorts  draws  moves  Engine/player 
-------------------------------------------------------------------
   1    159566  27.22%  17.96%  13.72%   73   Revenge 1.0 avx2  
   2     74524  11.21%  14.95%  26.09%   72   Komodo 14.1 HCE  
   3     54114  09.15%  12.65%  23.60%   73   Lc0 791921 CPU  
-------------------------------------------------------------------
*** Average length of all won games:     73 moves

 

 

A: Most high-value sacrifices (3+ pawnunits): [1]:05.93% Revenge 1.0 avx2   

                                              [2]:02.14% Komodo 14.1 HCE   

                                              [3]:00.33% Lc0 791921 CPU 


B: Most sacrifices overall                  : [1]:27.22% Revenge 1.0 avx2   

                                              [2]:11.21% Komodo 14.1 HCE   

                                              [3]:09.15% Lc0 791921 CPU 


C: Very short wins (40 moves or less)       : [1]:01.48% Revenge 1.0 avx2   

                                              [2]:00.53% Komodo 14.1 HCE   

                                              [3]:00.33% Lc0 791921 CPU 


D: Most short wins overall                  : [1]:17.96% Revenge 1.0 avx2   

                                              [2]:14.95% Komodo 14.1 HCE   

                                              [3]:12.65% Lc0 791921 CPU 


E: Average length of all won games          : [1]:072 Komodo 14.1 HCE   

                                              [2]:073 Lc0 791921 CPU   

                                              [3]:073 Revenge 1.0 avx2 


F: Smallest number of bad draws             : [1]:13.72% Revenge 1.0 avx2   

                                              [2]:23.60% Lc0 791921 CPU   

                                              [3]:26.09% Komodo 14.1 HCE