Skip to main content

Since you’re here...

... we have a small favour to ask. More people, like you, are reading and supporting our blog: "Chess Engines Diary". And unlike many other sites and blogs, we made the choice to keep our articles open for all, regardless of where they live or what they can afford to pay.

We hope you will consider supporting us today. We need your support to continue to exist, because good entries are more and more work time. Every reader contribution, however big or small, is so valuable. Support "Chess Engines Diary" even a small amount– and it only takes a minute. Thank you.

============================== My email: jotes@go2.pl



Stockfish 18012314_Nicolet - new version!

Stockfish, chess engine UCI

Rating JCER = 3389
🔬 Author: Stéphane Nicolet

Timestamp: 1516714005 
Contempt 20 
Set the default contempt value of Stockfish to 20 centipawns. 
The contempt feature of Stockfish tries to prevent the engine from simplifying the position too quickly when it feels that it is very slightly behind, instead keeping the tension a little bit longer. 

Various tests in November 2017 have proved that our current imple- mentation works well against SF7 (which is about 130 Elo weaker than current master) and than the Elo gain is an increasing function of contempt, going (against SF7) from +0 Elo when contempt is set at zero centipawns, to +30 Elo when contempt is 40 centipawns. 

This november discussion left open the decision of which "default" value for contempt we should use for Stockfish, taking into account the various uses ofStockfish (opening preparation for humans, computer online tournaments,analysis tool for web pages, human/computer play, etc). 

This pull request proposes to set the default contempt value of SF to twenty centipawns, which turns out to be the highest value which is not a regression against current master, as this seemed to be a good compromise between risk and safety. A couple of SPRT[-3..1] tests were done to bisect this value: 
Contempt 10: http://tests.stockfishchess.org/tests/view/5a5d42d20ebc5902977e2901 (PASSED) 
Contempt 15: http://tests.stockfishchess.org/tests/view/5a5d41740ebc5902977e28fa (PASSED) 
Contempt 20: http://tests.stockfishchess.org/tests/view/5a5d42060ebc5902977e28fc (PASSED) 
Contempt 25: http://tests.stockfishchess.org/tests/view/5a5d433f0ebc5902977e2904 (FAILED) 

Surprisingly, a test at "very long time control" hinted that using contempt 20 is not only be non-regressive against contempt 0, but may actually exhibit some small Elo gain, giving a likehood of superio-rity of 88.7% after 8500 games: 

VLTC: 
ELO: 2.28 +-3.7 (95%) LOS: 88.7% 
Total: 8521 W: 1096 L: 1040 D: 6385 

Finally, there was some concerns that a contempt value of 20 would be worse than a value of 7, but a test with 20000 games at STC was neutral: 

STC: 
ELO: 0.45 +-3.1 (95%) LOS: 61.2% 
Total: 20000 W: 4222 L: 4196 D: 11582 

See the comments in pull request 1361 for the long, nice discussion (180 entries :-)) leading to the decision to propose contempt 20 as the default value: 
https://github.com/official-stockfish/Stockfish/pull/1361 

Whether Stockfish should strictly adhere to the Komodo and Houdini semantics and add the UCI commands to force the contempt to be White in the so-called "analysis mode" is still under discussion, and may be or may not be the object of a future commit. 

Bench: 5783344 

⭐⭐⭐   Stockfish 18012314_Nicolet - download


Comments

Popular posts from this blog

New Strong Engines Test, by Chess Engines Diary, 2024.04.12

  💾  552 (!) games from the tournament download   👍 @chessenginesdiary  Country - Poland, City -  Malbork 🕓 Time 3'+3" 💻HP Pavilion i5-1035G1 8GB RAM 🖬 GUI-Banksia All  CEDR  317.321 games download  (01.04.2024 - 3'+3")  HypnoS 030424, Yuli GM Pro 16, Stockfish 16.1 and ShashChess 35 . These four engines scored the same number of points and placed at the top of the table in this strong tournament. Tech table: Engine KN/move NPS dep/mov time/mov mov/game time/game fails Alexandria 6.1.0 3218 576429 30.2 5.6 56.7 316.2 Berserk 13 4779 841025 36.7 5.7 56.0 318.1 Brainlearn 28 2275 366977 31.4 6.2 43.5 269.4 Caissa 1.18 5492 923486 30.2 5.9 50.4 299.7 Clover 6.1.19 4803 854377 31.4 5.6 62.2 349.4 Cool Iris 12.10 2201 356730 27.6 6.2 51.2 315.5 CorChess 20240331 2253 372774 27.4 6.0 52.2 315.8 Crystal 8 3582 571994 20.3 6.3 47.7 298.8 Fire 9.3 5225 876565 22.0 6.0 55.5 331.1 Fisherov chess monk 1.2 3487 619728 36.0 5.6 60.4 340.0 Hyp

New version chess engine: Raid 3.6 TacticaL RetreaT (Stockfish derivative)

Raid - is a Stockfish derivative . Rating CEDR=3743 Raid 3.2 - results: Polyfish 240114 7/14 +0 14 Games Hazard 4.21 DuaL 6.5/12 +1 12 Games Leptir 100124 6/12 +0 12 Games Arasan 24.1 7.5/10 +5 10 Games Uralochka 3.41 dev1 6/10 +2 10 Games Caissa 1.16 5.5/10 +1 10 Games SF-PRO 07.01.2024 5/10 +0 10 Games Stockfish 20240117 5/10 +0 10 Games Minic 3.41 7/8 +6 8 Games Pawn 3.0 6.5/8 +5 8 Games Texel 1.11 5.5/8 +3 8 Games RubiChess 20240112 5/8 +2 8 Games CorChess 20240103 4/8 +0 8 Games Counter 5.5 4.5/6 +3 6 Games Obsidian 10.0 3.5/6 +1 6 Games Fisherov chess monk 1.2 3/6 +0 6 Games Avalanche 2.1.0 3.5/4 +3 4 Games Starzix 4.0 2.5/4 +1 4 Games Cool Iris 11.80 2/4 +0 4 Games Blue Marlin 15.7 2/4 +0 4 Games AI 28.0 2/4 +0 4 Games

New chess engine: HypnoS 240324 (derived from Stockfish)

  HypnoS is a free and strong UCI chess engine derived from Stockfish  that analyzes chess positions and computes the optimal moves. HypnoS does not include a graphical user interface (GUI) that is required  to display a chessboard and to make it easy to input moves. These GUIs are  developed independently from HypnoS and are available online. HypnoS development is currently supported on the Openbench framework. OpenBench (created by Andrew Grant) is an open-source Sequential Probability Ratio Testing (SPRT) framework designed for self-play testing of engines. OpenBench makes use of distributed computing, allowing anyone to contribute CPU time to further the development of some of the world's most powerful engines. HypnoS 240324 download

Chess engine: Incognito 8.2 (experimental version of CorChess)

  Eduard Nemeth: "by Solista: This engine is identical in code to 8.01 but without Random Op. MultiPV mode. Some users only want to use normal MultiPV mode. Download Win 64-Bit (avx2, bmi2, sse41, source code).". Rating CEDR=3741 Incognito 5 PRO engine results SF CorChess 010124 25.5/50 +1 50 Games Predator AI 23/46 +0 46 Games Polyfish 240105 22.5/44 +1 44 Games Sawfish 2TC 21/42 +0 42 Games SF PB 070124 14.5/29 +0 29 Games Dragon 3.3 15.5/26 +5 26 Games Stockfish 20240108 12/24 +0 24 Games Deep Blue 20230628 12/24 +0 24 Games AbbyStein 2.8 12/24 +0 24 Games Berserk 12.1 11/20 +2 20 Games Cool Iris 11.80 10.5/20 +1 20 Games Ethereal 14.25 10.5/16 +5 16 Games Stockfish 20231231 8/16 +0 16 Games Raid 3.1 8/16 +0 16 Games SF-PRO 25.12.2023 8/16 +0 16 Games Killfish 071223 8/16 +0 16 Games Crystal 7 7/14 +0 14 Games TACTICAL 161223 6/12 +0 12 Games Fisherov chess monk 1.2 6/12 +0 12 Games Seer 2.8.0 6.5/10 +3 10 Games SF-PRO 07.01.2024 5/10 +0 10 Games Incognito 8.2 download

New version chess engine: Lc0 v.0.31.0-rc2

LcZero - CEDR Rating=3712 v0.31.0-rc2: In this version: Changed cuda compilation options to use -arch=native or -arch=all-major if no specific version is requested, with fallback for older cuda that don't support those options. Updated android builds to use openblas 0.3.27. A few small fixes. Individual statistics:  Lc0 BT4 Clover 6.1 5.5/8 +3 8 Games RofChade 3.1 4.5/8 +1 8 Games Dragon 3.3 4.5/8 +1 8 Games SF-PRO 19.11.2023 4/8 +0 8 Games YuliGM PrO 15 3.5/8 -1 8 Games ShashChess 34.5 3.5/8 -1 8 Games Sawfish 2TC 3.5/6 +1 6 Games Polyfish 231120 3.5/6 +1 6 Games Hazard 4.1 3/6 +0 6 Games Raid 3.1 3/6 +0 6 Games Eman 9.80 3/6 +0 6 Games Incognito 4 3/6 +0 6 Games Vanilla 14c 3/6 +0 6 Games Patzer AI X-256 2.5/6 -1 6 Games SugaR AI SE 2.5/5 +0 5 Games Altair 6.0.0 3.5/4 +3 4 Games Caissa 1.15 3/4 +2 4 Games Seer 2.8.0 2.5/4 +1 4 Games Fisherov chess monk 1.2 2.5/4 +1 4 Games Crystal 7 CMB 2.5/4 +1 4 Games Chess-System-Tal-2 2.5/4 +1 4 Games Lc0 v.0.31.0-rc2 (gpu-nvidia-cudnn) d