Skip to main content

Posts

SF PB 120624 wins New Engines Test, by Chess Engines Diary 2024.06.19

Another CEDR tournament in which we test new versions of chess engines for you. The winner was the SF PB 120624 engine with a large advantage. In second place is the SF-PRO 14/06/2024 (for now only available to our testers). The 3rd and 4th places were taken by the Cool Iris 12.10 and RapTora 3.0 engines. All CEDR  356.831 games download  (01.06.2024, 3'+3")  💾  650 (!) games from the tournament download 👍 @chessenginesdiary Country - Poland, City - Malbork  🕓 Time 3'+3"  💻HP Pavilion i5-1035G1 8GB RAM 🖬 GUI-Banksia Tech table: Engine KN/move NPS dep/mov time/mov mov/game time/game fails Altair 7.0.3 JA 2809 503532 21.7 5.6 63.2 352.4 Arasan 24.2.2 3427 519226 26.4 6.6 46.5 306.8 Berserk 13 4668 848540 32.6 5.5 59.8 329.2 Bread Engine 0.0.2 2184 350632 7.6 6.2 38.7 240.9 Caissa 1.18 5349 987497 27.3 5.4 58.2 315.2 Cool Iris 12.10 2541 396467 24.2 6.4 46.9 300.4 CorChess 20240606 2423 399605 29.1 6.1 50.3 304.7 Critter 1.6a 10433 1604357 17.6

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



Recent posts

End of the Torch Experiement?

 Andrew Grant: "Almost exactly one year ago we announced the release of Torch. We were a bit ambitious at the time, and announced it as the #2 engine, after beating Leela in the most recent Bullet event at CCC at the time. Today, that claim rings much more true, although still has some minor points of contention. Torch has beaten Leela in multiple Bullet and Blitz events in a row; Never managed to win one of the Rapid events; But has beaten Leela in a h2h classical event. Chesscom got into the space out of a need to develop a replacement for Komodo. They needed a strong chess engine which could be used for integrating their other products. Many features on Chesscom, from Game Review, to puzzle generation, to Bots (obviously!) are powered by high quality engines. The goal was always for that. In fact, despite me being the only full time employee working on Torch, a plurality of my time was spent on things not related to Torch's strength. Its a bit self-aggrandizing, but the Tor

Deep Ripper -1 wins TOP Engines Tournament, by Chess Engines Diary 2024.06.17

The tournament of the best chess engines was won by Deep Ripper -1 . In 2nd place Stockfish 20240605 Ivec . Places 3 and 4 were shared by Dragon 3.3 and Stockfish 16.1 engines. However, the results were "skewed" by numerous errors occurring in the engines. E.g. Lc0 BT4 - as many as 34 fails! All CEDR  356.831 games download  (01.06.2024, 3'+3")  💾  552 (!) games from the tournament download 👍 @chessenginesdiary Country - Poland, City - Malbork  🕓 Time 3'+3"  💻Dell G15 SE, i7-11800H RAM 16,0 GB 🖬 GUI-Banksia Tech table: Engine KN/move NPS dep/mov time/mov mov/game time/game fails Arasan 24.2.2 14999 2602150 31.0 5.8 61.7 355.6 Beast 15 MPV 9294 1832376 43.0 5.1 73.7 373.8 Berserk 13 19644 3956066 48.8 5.0 80.3 398.8 Big-SF 210324 8487 1687149 35.7 5.0 78.9 396.7 Caissa 1.18 20711 4088031 35.6 5.1 71.8 363.8 Chess-System-Tal-2.00 10054 2228374 27.4 4.5 71.9 324.5 CorChess 20240606 9822 1877715 39.6 5.2 69.4 363.0 Deep Ripper -1 10121 1

Chess engine: Bread Engine 0.0.4

  Bread Engine  is a chess engine written in c++. I started working on it in 2021, and only just finished. There is still a lot of room for improvement, but the engine is quite strong (for humans, at least). It uses NNUE (efficiently updatable neural network) to evaluate positions, as well as minimax search. Bread engine does not have a GUI built in, however it supports the uci protocol, you can therefore run it on any chess GUI. Author: Nonlinear2 Improved UCI commands, neural network is now always embedded in the executable.  Bread engine is now a single executable, that you can download and run directly. v.0.0.4: Version 0.0.4 has an improved neural network, as well as a few performance improvements. It is around 30 elo stronger than version 0.0.3. Score of bread_engine_0.0.4 vs bread_engine_0.0.3: 459 - 362 - 203 [0.547] 1024 ... bread_engine_0.0.4 playing White: 293 - 130 - 89 [0.659] 512 ... bread_engine_0.0.4 playing Black: 166 - 232 - 114 [0.436] 512 ... White vs Black: 525 - 2

Chess engine: Pedantic 1.1.0

  Pedantic - UCI engines Rating Chess Engines Diary CEDR=3129 v.1.1.0: Last HCE release which includes some small items waiting for release after 1.0.0 was made available. Playing strength improvements: AMD Ryzen 7 5800h ♦ 256mb Hash Table ♦ 3-6 Man Syzygy ♦ Concurrency 12 ♦ book-ply8-unifen-Q-0.25-0.40.pgn +41.7 +/- 7.9 Elo over Pedantic 1.0.0 in self-play at 20+0.2 time controls. +45.4 +/- 12.7 Elo over Pedantic 1.0.0 in self-play at 120+1 time controls. +40.5 +/- 20.8 Elo over Pedantic 1.0.0 in self-play at (12:00+8) time controls. 20+0.2 Self-Play Results Score of Pedantic 1.1.0 vs Pedantic 1.0.0: 1635 - 1098 - 1767 [0.560] 4500 ... Pedantic 1.1.0 playing White: 1073 - 337 - 840 [0.664] 2250 ... Pedantic 1.1.0 playing Black: 562 - 761 - 927 [0.456] 2250 ... White vs Black: 1834 - 899 - 1767 [0.604] 4500 Elo difference: 41.7 +/- 7.9, LOS: 100.0 %, DrawRatio: 39.3 % SPRT: llr 0 (0.0%), lbound -inf, ubound inf 120+1 Self-Play Results Score of Pedantic 1.1.0 vs Pedantic 1.0.0: 492 - 2

MEGA Dysk for donors: update 18.06.2024 ( 41 chess engines, 1-book, 3-GUI)

  Dear sponsors and friends. Especially for you, new engines, books and GUI added to the MEGA disk. Thank you for supporting our work. MEGA Dysk - update  18.06.2024  ( 41 chess engines) Windows: (34 files): A0 Lite TCEC 20 JA.7z Bagatur 5.1.7z Beast 15 MPV.7z Beast 15.7z Bread-Engine 0.0.2.7z Camel 1.6.0.7z Clarity 7.0.0.7z Clover 6.2.7z Deep Ripper -1.7z Deep Ripper exp.7z Dragonrose 0.26.7z Hazard JigSaw v5.8.7z Inanis 1.3.0.7z Killfish 150624.7z lc0-v0.31.0-rc3-windows-gpu-nvidia-cuda.7z lc0-v0.31.0-windows-gpu-nvidia-cuda-nodll.7z lc0-v0.31.0-windows-gpu-nvidia-cuda.7z Little Beast 15.7z Lynx 1.5.0.7z Marauders 3.6 Final BoSS.7z Mayhem 8.3.7z Peacekeeper 3.01.7z PlentyChess 2.0.0.7z RapTora 3.0.7z Ronja 0.9.1.7z RusticKnight 1.0.0.7z SF PB 120624.7z Simbelmyne 1.8.0.7z Simplex 0.9.8 JA.7z Smaug 2.2.1.7z Spaghet Pesto 1.0.7z Stockfish dev-20240605-5688b188.7z The Mad Prune 1.6.51 JA.7z Titan 1.1.0.7z MacOS (3 file): Lynx 1.5.0 for MacOS.7z Rocky 0.0.2 for MacOS.7z Simbelmyne 1.8.0

SF PB 120624 wins Test New Chess Engines, by Chess Engines Diary, 2024.06.17

Our next tournament in which we test new versions of chess engines. The clear winner was the SF PB 120624 engine. In places 2,3 and 4, with the same points, the following engines: Steel Fighter 2, Little Beast 15, Marauders 3.6 .     All CEDR  356.831 games download  (01.06.2024, 3'+3")  💾  650 (!) games from the tournament download 👍 @chessenginesdiary Country - Poland, City - Malbork  🕓 Time 3'+3"  💻Asus Zenbook, AMD Ryzen 9 6900HX, RAM 32,0 GB 🖬 GUI-Banksia Tech table: Engine KN/move NPS dep/mov time/mov mov/game time/game fails Beast 15 MPV 13660 2576898 50.4 5.3 62.6 331.9 Bread Engine 0.0.2 5249 872612 8.3 6.0 40.6 244.1 Camel 1.6.0 2238 411454 12.8 5.4 53.0 288.4 Clarity 7.0.0 11234 2164520 33.1 5.2 71.5 371.3 1 Clover 6.2 25927 5015056 38.0 5.2 67.2 347.4 CorChess 20240606 13645 2522482 43.7 5.4 66.2 358.2 Deep Ripper -1 12044 2399292 46.3 5.0 70.6 354.3 Dragonrose 0.26 2700 488518 10.1 5.5 44.8 247.4 Lc0 v0.31.0-rc3 129 25404 9.2 5

Chess engine: Bagatur 5.1

Bagatur - UCI chess engines,   Rating Chess Engines Diary CEDR= 3086 Author: Krasimir I. Topchiyski   v.5.1: NNUE in Java based on Bullet trainer and custom training process. Bagatur 4.1a - results: Booot 7.3 0/6 -6 6 Games Reckless 0.5.0 1.5/4 -1 4 Games Critter 1.6a 1/4 -2 4 Games Lizard 10.2 1/4 -2 4 Games Tenax 0.9.5 0.5/4 -3 4 Games Alexandria 6.0 0.5/4 -3 4 Games Superultra 2.0 0.5/4 -3 4 Games Leorik 3.0 0.5/4 -3 4 Games Clarity 5.0.0 0.5/4 -3 4 Games Caissa 1.17 0.5/4 -3 4 Games Brainlearn 27 0.5/4 -3 4 Games CorChess 20240209 0/4 -4 4 Games Willow 4.0.0.1 0/4 -4 4 Games Leptir 070224 0/4 -4 4 Games Eman 9.90 0/4 -4 4 Games Peacekeeper 2.30 0/4 -4 4 Games ShashChess 34.6 0/4 -4 4 Games Puffin 2.0 2/2 +2 2 Games Lambergar 0.4.1 2/2 +2 2 Games Odonata 0.8.0 2/2 +2 2 Games Mcu-max 1.0 2/2 +2 2 Games Bagatur 5.1 download  

Obsidian dev-12.29 wins Middle Class Engines Tournament (Tests by Jörn Gronemann, Heide, 2024.06.16 Time: 3'+3")

A great tournament with a lot of games played - 1100. The Obsidian dev-12.29 engine was by far the best with a clear advantage. In place 2, the Berserk 13 engine. Unexpectedly in place 3 - Ethereal 14.25 , a commercial engine. GPU NVIDIA  RTX 3060 ti. GUI-Banksia, For Every Engine only 1 CPU and 512 Hashtables, Book - 4 GM mvs book,  6TBs. 🕒🕒Time 3'+3"   1100 (!!!) tournaments games download   💾 All games CEDR  356.831 games download 01.06.2024 (3'+3")   @chessenginesdiary P Engine Score +/-/= 1: Obsidian dev-12.29 114.5 / 200 (+31 -2 =167) 2: Berserk 13 109.5 / 200 (+23 -4 =173) 3: Ethereal 14.25 107.5 / 200 (+21 -6 =173) 4: Alexandria-7.0.0 106.0 / 200 (+18 -6 =176) 5: PlentyChess 2.0.0 101.5 / 200 (+13 -10 =177) 6: RubiChess 20240112 101.5 / 200 (+17 -14 =169) 7: Caissa 1.18 99.5 /

Chess engine: Smaug 2.2.1 JA

Smaug 2.2.1 by Tord Romstad & Joona Kiiski. Smaug is slightly modified version of Glaurung 2.2 by Tord Romstad Re-compile Windows/Linux & Android clang profiled builds Jim Ablett  New J.A Compiles Smaug 2.2.1 JA download

SF-PRO 10.06.2024 and Beast 15 MPV wins New Engines Test, by Chess Engines Diary, 2024.06.16

Another tournament in which we "tested" 28 new versions of chess engines. Places 1 and 2 were taken by SF-PRO 10/06/2024 and Beast 15 MPV engines, which scored the same number of points. We would like to draw your attention to the very good result of the Titan 1.1.0 engine, it is only its second official version and it takes 7th place!   All CEDR  356.831 games download  (01.06.2024, 3'+3")  💾  756 (!) games from the tournament download 👍 @chessenginesdiary Country - Poland, City - Malbork  🕓 Time 3'+3"  💻HP Pavilion i5-1035G1 8GB RAM 🖬 GUI-Banksia Tech table: Engine KN/move NPS dep/mov time/mov mov/game time/game fails Bagatur 5.0 95 15388 11.6 6.2 34.9 214.8 5 Beast 15 MPV 2691 436882 30.2 6.2 44.3 272.6 Camel 1.6.0 842 145856 12.0 5.8 43.6 251.4 Clover 6.2 5880 991483 24.6 5.9 49.1 291.5 Cyrano 0.6b17 JA 5404 847727 15.0 6.4 47.4 302.2 Deep Ripper (exp) 2685 424745 29.5 6.3 44.2 279.5 DoctorB 1.2.1 JA 5295 1229182 7.4 4.3 33.9 146.1

New version chess engine: Lc0 v.0.31

LcZero - CEDR Rating=3698 We have an official version of the LcZero engine. We will start the first tests today! v0.31.0: In this version: The blas, cuda, eigen, metal and onnx backends now have support for multihead network architecture and can run BT3/BT4 nets. Updated the internal Elo model to better align with regular Elo for human players. There is a new XLA backend that uses OpenXLA compiler to produce code to execute the neural network. See https://github.com/LeelaChessZero/lc0/wiki/XLA-backend for details. Related are new leela2onnx options to output the HLO format that XLA understands. There is a vastly simplified lc0 interface available by renaming the executable to lc0simple. The backends can now suggest a minibatch size to the search, this is enabled by --minibatch-size=0 (the new default). If the cudnn backend detected an unsupported network architecture it will switch to the cuda backend. Two new selfplay options enable value and policy tournaments. A policy tournament