Stefan Pohl Computer ChessHome of famous UHO openings and EAS RatinglistLatest Website-News (2024/07/27): Testrun of Stockfish 240719 finished: +1 Celo to Stockfish 240711: 3845 Celo is the new all-time highscore in my UHO-Top15 Ratinglist.
Dont forget to take a look at my EAS-Ratinglist (The world's first engine-ratinglist not measuring strength of engines, but engines's style of play).
Next testrun: Clover 7.0 and Torch 3 Next NN-testrun: Lc0 0.32dev BT4-6077500 (best BT4-net in my previous tests, and the new binary from Ergodice, which should be a little stronger (OpenBench-test here)). Download the Cuda binary here
Great news & sad news: A.Grant just provided me the download of the final Celo-gaining version of Torch, Torch 3. Many thanks. Bud sad news, that Torch 3 will be the final version (for now...). Here some words by A.Grant in an email to me: "As you may know, chesscom decided to stop funding the development of Torch strictly for strength as of about two months ago. The project ran for a little over a year. While we were able to very quickly become #3, and after some time make a strong argument for a clear #2, becoming #1 is just not in the cards.
Of course our initial ambition was to be #1 -- how could you aim for anything else? But in reality, I was the only full time employee on the project, and we were always going to be limited in what could be done. It would take an incredible effort -- in terms of raw manpower and in compute -- to match the dozens of developers and millions of dollars of hardware behind Stockfish.
In the end though, Chesscom's goal was actually to create a new engine to replace Komodo, to use in their internal tools and products. Torch is already powering a large portion of the chesscom infrastructure, and will soon replace Komodo entirely there.
It has been the joy of my lifetime (thus far!) to work on this project. The authors of Koivisto and Berserk and I had always talked about coming together, saying that we could do quite well if we just unified. And that turned out to be true."
Stay tuned UHO-Top15 Engines Ratinglist (+ regular testing of Stockfish Dev-versions)
Playing conditions:
The 15 strongest engines and the latest StockfishDev version are playing 1000 games vs. each opponent: 15000 games per engine are played. A RoundRobin Tournament with 120000 games overall !
Hardware: AMD Ryzen 3900 12-core (24 threads) notebook with 32GB RAM. Speed: (singlethread, TurboBoost-mode switched off, chess starting position) Stockfish 16.1: 500 kn/s (when 20 games are running simultaneously) Hash: 256MB per engine GUI: Cutechess-cli (GUI ends game, when a 5-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, 5 Syzygy for cutechess-cli Openings: My UHO_2022_6mvs_+120_+129 openings are used (first 500 lines) (UHO 2022 openings are part of my Anti Draw Openings download-package). Ponder, Large Memory Pages & learning: Off Thinking time: 3min+1sec per game/engine (average game-duration: 7 min 50sec), around 3500-4000 games in 24 hours are played. One testrun (15000 games) takes around 100 hours (=4 days and some hours) 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 https://github.com/official-stockfish/Stockfish/releases (except the official SF-release versions, which are taken form the official Stockfish website). Read the explanation, why I chose my testing-conditions, like I did here
Latest update: 2024/07/25: Stockfish 240719 (+1 Celo to Stockfish 240711)
(Ordo-calculation fixed to Stockfish 16.1 = 3833 Celo. Mention, there are no human or "realistic" Elo numbers in enginechess, since the engines reached superhuman strength, so I decided to use "Celo" (=Computer Elo) instead)
See the tournament cross-table as a picture (done by Fritz18 GUI) here See the individual statistics of engine-results here See the Engines Aggressiveness Score Ratinglist here See the Gamepair-statistics of the UHO-Top15 Ratinglist here See the full UHO Ratinglist (with EAS-Ratinglist): here Download the current gamebase, the full-list gamebase and the archive here See the old SPCC-Ratinglist (full list) and EAS-List from 2020 until 2023/08/28: here
(best Stockfish Celo: Stockfish 240719: 3845 Celo, latest full release (Stockfish 16.1) had 3833 Celo) (highest EAS-Score since Stockfish 16.1 release: Stockfish 240519: 203882 EAS-points) (Read the explanation of the Lc0-entry in the Ratinglist here). Program Celo + - Games Score Av.Op. Draws 1 Stockfish 240719 avx2 : 3845 4 4 15000 71.2% 3680 47.4%
White Wins : 57968 (48.3 %) Below the UHO-Top15 Ratinglist, recalculated with Gamepairs (using my Gamepair Rescoring Tool V1.5), Ordo-calculation fixed to Stockfish 16.1 = 3833 Celo, 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)." See the head-to-head gamepair-statistics of the UHO-Top15 Ratinglist here # PLAYER : Celo Error Pairs W D L (%) CFS(%)
Aborted testruns, because the tested engine was too weak to enter the UHO-Top15 Ratinglist: Minic 3.41, Booot 7.3, Black Marlin 9.0, Equisetum 1.0, Stormphrax 5 Below you find a diagram of the progress of Stockfish in my tests since February 2023. And below that diagram, the older diagrams.
You can save the diagrams on your PC with mouseclick (right button) and then choose "save image"...
Sie sind Besucher Nr. |