Stefan Pohl Computer Chess

private website for chessengine-tests

UHO Top 10 Ratinglist with Stockfish Dev progression

The 10 strongest engines of my SPCC Ratinglist and the latest StockfishDev version are playing 1000 games vs. each opponent: 10000 games per engine are played. A RoundRobin Tournament with 55000 games overall !
Results evaluated by ORDO, reference engine is the latest official Stockfish (Stockfish 15.1 right now), its Elo is set to 0. Why? Because using my UHO openings spread Elo-results wider than balanced openings, so it makes no sense to set the reference-Elo to any "realistic" value of 3800 or so... And mention, that all Computer Elo (Celo) in the various ratinglists are fictional in these days of superstrong engines, because the Celos can not be compared with human Elos anymore and since a long time.

Latest testrun: 2023/03/19: Stockfish 230314 (0 Elo to Stockfish 230308)
Reference point (opponent) is the latest official SF-release (Stockfish 15.1 right now).

Hardware: AMD Ryzen 3900 12-core (24 threads) notebook with 32GB RAM. 20 games are played simultaneously

Speed: Singlethread, TurboBoost-mode switched off, chess starting position: Stockfish 15: 750000 n/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: 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: 8 min 7sec (little bit longer than the games of the SPCC-Ratinglist take (7min 45sec), because using UHO-openings lowers the number of early 3fold-draws a lot))

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).

ORDO calculation fixed to reference-engine (Elo = 0)

See the individual statistics of engine results here

See the EAS-Ratinglists of the UHO Top10 Ratinglist games here

You can download all played games from my Google-Drive. Download here

Program Elo + - Games Score Av.Op. Draws

1 Stockfish 230314 avx2 : 14 5 5 10000 73.1% -170 46.2% 2 Stockfish 15.1 221204 : 0 5 5 10000 71.4% -168 46.7% 3 KomodoDragon 3.2 avx2 : -52 5 5 10000 64.6% -163 47.8% 4 Berserk 11 avx2 : -129 5 5 10000 53.7% -155 49.6% 5 Ethereal 14.00 nnue : -163 5 5 10000 48.8% -152 49.9% 6 Koivisto 9 avx2 : -190 4 4 10000 44.8% -149 49.1% 7 RubiChess 221120 avx2 : -214 5 5 10000 41.3% -147 47.4% 8 Revenge 3.0 avx2 : -214 4 4 10000 41.3% -147 49.4% 9 Rebel 16.1 : -236 4 4 10000 38.1% -145 49.2% 10 Fire 8.NN avx2 : -238 5 5 10000 37.9% -144 47.7% 11 Igel 3.4.0 popavx2 : -259 5 5 10000 35.0% -142 48.2%

Games : 55000 (finished)

White Wins : 26643 (48.4 %) Black Wins : 1791 (3.3 %) Draws : 26566 (48.3 %)

Here the progress in regression-testing since Stockfish 15.1 (2022/12/04), with Elo of SF 15.1 set to 0 in a diagram:

Below the UHO Top10 gamebase recalculated with my Gamepairs Rescorer Batch-Tool. 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)."

# PLAYER : RATING ERROR PLAYED W D L (%) CFS(%) 1 Stockfish 230314 avx2 : 57 17 5000 4322 590 88 92.3 100 2 Stockfish 15.1 221204 : 0 ---- 5000 4124 710 166 89.6 100 3 KomodoDragon 3.2 avx2 : -215 16 5000 3342 989 669 76.7 100 4 Berserk 11 avx2 : -461 17 5000 2016 1565 1419 56.0 100 5 Ethereal 14.00 nnue : -533 17 5000 1581 1666 1753 48.3 100 6 Koivisto 9 avx2 : -610 17 5000 1150 1671 2179 39.7 100 7 RubiChess 221120 avx2 : -651 18 5000 891 1721 2388 35.0 100 8 Revenge 3.0 avx2 : -666 18 5000 808 1717 2475 33.3 100 9 Rebel 16.1 : -714 17 5000 595 1614 2791 28.0 66 10 Fire 8.NN avx2 : -717 17 5000 593 1593 2814 27.8 100 11 Igel 3.4.0 popavx2 : -760 18 5000 426 1468 3106 23.2 ---

You can download my Gamepairs Rescorer Tool right here

Below the head to head statistics of the Gamepairs-rescoring, with all WDL-statistics:

Head to head statistics:

1) Stockfish 230314 avx2 57 : 5000 (+4322,=590,-88), 92.3 %

vs. : games ( +, =, -), (%) : Diff, SD, CFS (%) Stockfish 15.1 221204 : 500 ( 142, 290, 68), 57.4 : +57, 9, 100.0 KomodoDragon 3.2 avx2 : 500 ( 340, 144, 16), 82.4 : +271, 9, 100.0 Berserk 11 avx2 : 500 ( 450, 48, 2), 94.8 : +518, 9, 100.0 Ethereal 14.00 nnue : 500 ( 469, 31, 0), 96.9 : +590, 10, 100.0 Koivisto 9 avx2 : 500 ( 477, 22, 1), 97.6 : +667, 10, 100.0 RubiChess 221120 avx2 : 500 ( 488, 12, 0), 98.8 : +708, 10, 100.0 Revenge 3.0 avx2 : 500 ( 482, 17, 1), 98.1 : +723, 10, 100.0 Rebel 16.1 : 500 ( 490, 10, 0), 99.0 : +771, 9, 100.0 Fire 8.NN avx2 : 500 ( 492, 8, 0), 99.2 : +773, 9, 100.0 Igel 3.4.0 popavx2 : 500 ( 492, 8, 0), 99.2 : +817, 10, 100.0

2) Stockfish 15.1 221204 0 : 5000 (+4124,=710,-166), 89.6 %

vs. : games ( +, =, -), (%) : Diff, SD, CFS (%) Stockfish 230314 avx2 : 500 ( 68, 290, 142), 42.6 : -57, 9, 0.0 KomodoDragon 3.2 avx2 : 500 ( 284, 197, 19), 76.5 : +215, 8, 100.0 Berserk 11 avx2 : 500 ( 431, 67, 2), 92.9 : +461, 9, 100.0 Ethereal 14.00 nnue : 500 ( 456, 42, 2), 95.4 : +533, 9, 100.0 Koivisto 9 avx2 : 500 ( 472, 27, 1), 97.1 : +610, 9, 100.0 RubiChess 221120 avx2 : 500 ( 476, 24, 0), 97.6 : +651, 9, 100.0 Revenge 3.0 avx2 : 500 ( 484, 16, 0), 98.4 : +666, 9, 100.0 Rebel 16.1 : 500 ( 485, 15, 0), 98.5 : +714, 9, 100.0 Fire 8.NN avx2 : 500 ( 481, 19, 0), 98.1 : +717, 9, 100.0 Igel 3.4.0 popavx2 : 500 ( 487, 13, 0), 98.7 : +760, 9, 100.0

3) KomodoDragon 3.2 avx2 -215 : 5000 (+3342,=989,-669), 76.7 %

vs. : games ( +, =, -), (%) : Diff, SD, CFS (%) Stockfish 230314 avx2 : 500 ( 16, 144, 340), 17.6 : -271, 9, 0.0 Stockfish 15.1 221204 : 500 ( 19, 197, 284), 23.5 : -215, 8, 0.0 Berserk 11 avx2 : 500 ( 311, 175, 14), 79.7 : +246, 7, 100.0 Ethereal 14.00 nnue : 500 ( 361, 125, 14), 84.7 : +319, 7, 100.0 Koivisto 9 avx2 : 500 ( 411, 86, 3), 90.8 : +395, 8, 100.0 RubiChess 221120 avx2 : 500 ( 427, 70, 3), 92.4 : +437, 8, 100.0 Revenge 3.0 avx2 : 500 ( 432, 65, 3), 92.9 : +452, 8, 100.0 Rebel 16.1 : 500 ( 448, 48, 4), 94.4 : +500, 7, 100.0 Fire 8.NN avx2 : 500 ( 448, 49, 3), 94.5 : +502, 7, 100.0 Igel 3.4.0 popavx2 : 500 ( 469, 30, 1), 96.8 : +546, 8, 100.0

4) Berserk 11 avx2 -461 : 5000 (+2016,=1565,-1419), 56.0 %

vs. : games ( +, =, -), (%) : Diff, SD, CFS (%) Stockfish 230314 avx2 : 500 ( 2, 48, 450), 5.2 : -518, 9, 0.0 Stockfish 15.1 221204 : 500 ( 2, 67, 431), 7.1 : -461, 9, 0.0 KomodoDragon 3.2 avx2 : 500 ( 14, 175, 311), 20.3 : -246, 7, 0.0 Ethereal 14.00 nnue : 500 ( 181, 242, 77), 60.4 : +73, 6, 100.0 Koivisto 9 avx2 : 500 ( 266, 200, 34), 73.2 : +149, 6, 100.0 RubiChess 221120 avx2 : 500 ( 253, 209, 38), 71.5 : +191, 6, 100.0 Revenge 3.0 avx2 : 500 ( 284, 186, 30), 75.4 : +206, 6, 100.0 Rebel 16.1 : 500 ( 331, 157, 12), 81.9 : +254, 6, 100.0 Fire 8.NN avx2 : 500 ( 331, 153, 16), 81.5 : +256, 6, 100.0 Igel 3.4.0 popavx2 : 500 ( 352, 128, 20), 83.2 : +300, 6, 100.0

5) Ethereal 14.00 nnue -533 : 5000 (+1581,=1666,-1753), 48.3 %

vs. : games ( +, =, -), (%) : Diff, SD, CFS (%) Stockfish 230314 avx2 : 500 ( 0, 31, 469), 3.1 : -590, 10, 0.0 Stockfish 15.1 221204 : 500 ( 2, 42, 456), 4.6 : -533, 9, 0.0 KomodoDragon 3.2 avx2 : 500 ( 14, 125, 361), 15.3 : -319, 7, 0.0 Berserk 11 avx2 : 500 ( 77, 242, 181), 39.6 : -73, 6, 0.0 Koivisto 9 avx2 : 500 ( 189, 242, 69), 62.0 : +76, 5, 100.0 RubiChess 221120 avx2 : 500 ( 203, 229, 68), 63.5 : +118, 6, 100.0 Revenge 3.0 avx2 : 500 ( 231, 217, 52), 67.9 : +133, 6, 100.0 Rebel 16.1 : 500 ( 272, 193, 35), 73.7 : +181, 6, 100.0 Fire 8.NN avx2 : 500 ( 278, 182, 40), 73.8 : +183, 6, 100.0 Igel 3.4.0 popavx2 : 500 ( 315, 163, 22), 79.3 : +227, 6, 100.0

6) Koivisto 9 avx2 -610 : 5000 (+1150,=1671,-2179), 39.7 %

vs. : games ( +, =, -), (%) : Diff, SD, CFS (%) Stockfish 230314 avx2 : 500 ( 1, 22, 477), 2.4 : -667, 10, 0.0 Stockfish 15.1 221204 : 500 ( 1, 27, 472), 2.9 : -610, 9, 0.0 KomodoDragon 3.2 avx2 : 500 ( 3, 86, 411), 9.2 : -395, 8, 0.0 Berserk 11 avx2 : 500 ( 34, 200, 266), 26.8 : -149, 6, 0.0 Ethereal 14.00 nnue : 500 ( 69, 242, 189), 38.0 : -76, 5, 0.0 RubiChess 221120 avx2 : 500 ( 160, 237, 103), 55.7 : +41, 6, 100.0 Revenge 3.0 avx2 : 500 ( 166, 246, 88), 57.8 : +56, 6, 100.0 Rebel 16.1 : 500 ( 223, 213, 64), 65.9 : +104, 6, 100.0 Fire 8.NN avx2 : 500 ( 225, 208, 67), 65.8 : +107, 6, 100.0 Igel 3.4.0 popavx2 : 500 ( 268, 190, 42), 72.6 : +151, 6, 100.0

7) RubiChess 221120 avx2 -651 : 5000 (+891,=1721,-2388), 35.0 %

vs. : games ( +, =, -), (%) : Diff, SD, CFS (%) Stockfish 230314 avx2 : 500 ( 0, 12, 488), 1.2 : -708, 10, 0.0 Stockfish 15.1 221204 : 500 ( 0, 24, 476), 2.4 : -651, 9, 0.0 KomodoDragon 3.2 avx2 : 500 ( 3, 70, 427), 7.6 : -437, 8, 0.0 Berserk 11 avx2 : 500 ( 38, 209, 253), 28.5 : -191, 6, 0.0 Ethereal 14.00 nnue : 500 ( 68, 229, 203), 36.5 : -118, 6, 0.0 Koivisto 9 avx2 : 500 ( 103, 237, 160), 44.3 : -41, 6, 0.0 Revenge 3.0 avx2 : 500 ( 137, 247, 116), 52.1 : +15, 6, 99.6 Rebel 16.1 : 500 ( 167, 236, 97), 57.0 : +63, 6, 100.0 Fire 8.NN avx2 : 500 ( 170, 246, 84), 58.6 : +65, 6, 100.0 Igel 3.4.0 popavx2 : 500 ( 205, 211, 84), 62.1 : +109, 6, 100.0

8) Revenge 3.0 avx2 -666 : 5000 (+808,=1717,-2475), 33.3 %

vs. : games ( +, =, -), (%) : Diff, SD, CFS (%) Stockfish 230314 avx2 : 500 ( 1, 17, 482), 1.9 : -723, 10, 0.0 Stockfish 15.1 221204 : 500 ( 0, 16, 484), 1.6 : -666, 9, 0.0 KomodoDragon 3.2 avx2 : 500 ( 3, 65, 432), 7.1 : -452, 8, 0.0 Berserk 11 avx2 : 500 ( 30, 186, 284), 24.6 : -206, 6, 0.0 Ethereal 14.00 nnue : 500 ( 52, 217, 231), 32.1 : -133, 6, 0.0 Koivisto 9 avx2 : 500 ( 88, 246, 166), 42.2 : -56, 6, 0.0 RubiChess 221120 avx2 : 500 ( 116, 247, 137), 47.9 : -15, 6, 0.4 Rebel 16.1 : 500 ( 154, 254, 92), 56.2 : +48, 6, 100.0 Fire 8.NN avx2 : 500 ( 175, 233, 92), 58.3 : +50, 5, 100.0 Igel 3.4.0 popavx2 : 500 ( 189, 236, 75), 61.4 : +94, 6, 100.0

9) Rebel 16.1 -714 : 5000 (+595,=1614,-2791), 28.0 %

vs. : games ( +, =, -), (%) : Diff, SD, CFS (%) Stockfish 230314 avx2 : 500 ( 0, 10, 490), 1.0 : -771, 9, 0.0 Stockfish 15.1 221204 : 500 ( 0, 15, 485), 1.5 : -714, 9, 0.0 KomodoDragon 3.2 avx2 : 500 ( 4, 48, 448), 5.6 : -500, 7, 0.0 Berserk 11 avx2 : 500 ( 12, 157, 331), 18.1 : -254, 6, 0.0 Ethereal 14.00 nnue : 500 ( 35, 193, 272), 26.3 : -181, 6, 0.0 Koivisto 9 avx2 : 500 ( 64, 213, 223), 34.1 : -104, 6, 0.0 RubiChess 221120 avx2 : 500 ( 97, 236, 167), 43.0 : -63, 6, 0.0 Revenge 3.0 avx2 : 500 ( 92, 254, 154), 43.8 : -48, 6, 0.0 Fire 8.NN avx2 : 500 ( 129, 247, 124), 50.5 : +2, 6, 66.3 Igel 3.4.0 popavx2 : 500 ( 162, 241, 97), 56.5 : +46, 6, 100.0

10) Fire 8.NN avx2 -717 : 5000 (+593,=1593,-2814), 27.8 %

vs. : games ( +, =, -), (%) : Diff, SD, CFS (%) Stockfish 230314 avx2 : 500 ( 0, 8, 492), 0.8 : -773, 9, 0.0 Stockfish 15.1 221204 : 500 ( 0, 19, 481), 1.9 : -717, 9, 0.0 KomodoDragon 3.2 avx2 : 500 ( 3, 49, 448), 5.5 : -502, 7, 0.0 Berserk 11 avx2 : 500 ( 16, 153, 331), 18.5 : -256, 6, 0.0 Ethereal 14.00 nnue : 500 ( 40, 182, 278), 26.2 : -183, 6, 0.0 Koivisto 9 avx2 : 500 ( 67, 208, 225), 34.2 : -107, 6, 0.0 RubiChess 221120 avx2 : 500 ( 84, 246, 170), 41.4 : -65, 6, 0.0 Revenge 3.0 avx2 : 500 ( 92, 233, 175), 41.7 : -50, 5, 0.0 Rebel 16.1 : 500 ( 124, 247, 129), 49.5 : -2, 6, 33.7 Igel 3.4.0 popavx2 : 500 ( 167, 248, 85), 58.2 : +44, 6, 100.0

11) Igel 3.4.0 popavx2 -760 : 5000 (+426,=1468,-3106), 23.2 %

vs. : games ( +, =, -), (%) : Diff, SD, CFS (%) Stockfish 230314 avx2 : 500 ( 0, 8, 492), 0.8 : -817, 10, 0.0 Stockfish 15.1 221204 : 500 ( 0, 13, 487), 1.3 : -760, 9, 0.0 KomodoDragon 3.2 avx2 : 500 ( 1, 30, 469), 3.2 : -546, 8, 0.0 Berserk 11 avx2 : 500 ( 20, 128, 352), 16.8 : -300, 6, 0.0 Ethereal 14.00 nnue : 500 ( 22, 163, 315), 20.7 : -227, 6, 0.0 Koivisto 9 avx2 : 500 ( 42, 190, 268), 27.4 : -151, 6, 0.0 RubiChess 221120 avx2 : 500 ( 84, 211, 205), 37.9 : -109, 6, 0.0 Revenge 3.0 avx2 : 500 ( 75, 236, 189), 38.6 : -94, 6, 0.0 Rebel 16.1 : 500 ( 97, 241, 162), 43.5 : -46, 6, 0.0 Fire 8.NN avx2 : 500 ( 85, 248, 167), 41.8 : -44, 6, 0.0

Datenschutzerklärung