Test Conditions
- OS: Windows 11 Home - HW: 13th Gen Intel(R) Core(TM) i7-13700HX, Lenovo 5 Legion
- Gui: Cute Chess Gui - Tournameent Type: Gaunlet - Time Setup: 40m/1m
- Games by Match: 100 as much as possible.
- Ponder: ON - Hash Table: 512Mb - Threads: 1 - Syzygy endgame tablebases: 3,4,5 complete + most important 6,7.
- Opening Book: Different Opening Test Sets consolidated in Polyglot Book
and progressively removing bad opening lines.
- PGN games are available upon request.
TACTICAL 170325 Tactical.0-TacticalDepth.0-Experience.On :
Brainlearn 30 PL.On : 100 : 3+ : 91= : 6- : 48.5%
Stockfish 15 : 100 : 6+ : 89= : 5- : 50.5%
Stockfish 15.1 : 100 : 7+ : 91= : 2- : 52.5%
Stockfish 16 : 100 : 3+ : 93= : 4- : 49.5%
Stockfish 16.1 : 100 : 2+ : 90= : 8- : 47.0%
Stockfish 17 : 100 : 2+ : 96= : 2- : 50.0%
Stockfish Dev.25022808 : 100 : 3+ : 85= : 12- : 45.5%
Remark: TACTICAL 170325 appears to be weaker than the last Stockfish Development Version,
suggesting a strength difference comparable to that between Stockfish 16.1 and 17.
Gaunlets for Fun
Gaunlets for Fun: TACTICAL 170325 vs Stockfishes
Gaunlets for Fun: Cool Iris v17 vs Stockfishes
Gaunlets for Fun: Cool Iris v17 vs Stockfishes Test Conditions - OS: Windows 11 Home - HW: 13th Gen Intel(R) Core(TM) i7-13700HX, Lenovo 5 Legion - Gui: Cute Chess Gui - Tournameent Type: Gaunlet - Time Setup: 40m/2m - Games by Match: 100 as much as possible. - Ponder: ON - Hash Table: 512Mb - Threads: 1 - Syzygy endgame tablebases: 3,4,5 complete + most important 6,7. - Opening Book: Different Opening Test Sets. Now: lacrosse-nunn-silver-reviewed.pgn, Bias [ -0.80, 0.80 ] - PGN games are available upon request. Cool Iris v17 Stockfish 15 : 100 : 13+ : 87= : 0- : 56.5% Stockfish 15.1 : 100 : 6+ : 94= : 0- : 53.0% Stockfish 16 : 100 : 7+ : 93= : 0- : 53.5% Stockfish 16.1 : 100 : 8+ : 91= : 1- : 53.5% Stockfish 17 : 100 : 4+ : 95= : 1- : 51.5% Stockfish 17.1 : 100 : 4+ : 95= : 1- : 51.5% Stockfish Dev.25042718 : 100 : 1+ : 99= : 0- : 50.5%
Gaunlets for Fun: SIERRA 250425 vs Stockfishes
Gaunlets for Fun: SIERRA 250425 vs Stockfishes Test Conditions - OS: Windows 11 Home - HW: 13th Gen Intel(R) Core(TM) i7-13700HX, Lenovo 5 Legion - Gui: Cute Chess Gui - Tournameent Type: Gaunlet - Time Setup: 40m/2m - Games by Match: 100 as much as possible. - Ponder: ON - Hash Table: 512Mb - Threads: 1 - Syzygy endgame tablebases: 3,4,5 complete + most important 6,7. - Opening Book: Different Opening Test Sets. Now: whc-wcc-reviewed.pgn, Bias [ -0.80, 0.80 ] SIERRA 250425 : Stockfish 15 : 100 : 9+ : 88= : 3- : 53.0% Stockfish 15.1 : 100 : 4+ : 95= : 1- : 51.5% Stockfish 16 : 100 : 4+ : 94= : 2- : 51.0% Stockfish 16.1 : 100 : 3+ : 96= : 1- : 51.0% Stockfish 17 : 100 : 2+ : 92= : 6- : 48.0% Stockfish 17.1 : 100 : 1+ : 93= : 6- : 47.5% Stockfish Dev.25042718 : 100 : 0+ : 95= : 5- : 47.5% (Eng) SIERRA 250425 faced its most formidable challenge against Stockfish 17 and 17.1. The matches were based on a meticulously crafted 23-ply opening set, rigorously analyzed by 35 engines to avoid weak lines. This repertoire seamlessly combined human championship openings (1984-2024) with computer championship lines (2011-2024) (Sp) SIERRA 250425 enfrentó su desafío más difícil contra Stockfish 17 y 17.1. Las partidas se basaron en un repertorio de aperturas de 23 movimientos, meticulosamente analizado por 35 motores para remover las líneas deficientes. Este conjunto fusionó de manera experta aperturas de campeonatos humanos (1984-2024) con líneas de campeonatos computacionales (2011-2024).
Re: Gaunlets for Fun
Dear Supernova, TACTICAL is a bit older than a year old code. SIERRA is based on Stockfish 17.1, with a tendency toward move history based on the pawn structure.
François-André Danica Philidor, once said that pawns are like the soul of chess, so SIERRA tentatively pursues this principle. In many pawn structures, pieces have "naturally good" and "naturally bad" squares; the engine takes this information and adds it to the move value when ordering moves and pruning search heuristics, although it usually doesn't work.
And Cool Iris is a slightly different architecture with a large net different from the current green net.
Additionally, Cool Iris and SIERRA (the firts and the last attempt), have served me well in my compilation tests on clang64, using a third-party and PGO, which generally gives them a 5 percent speedup against SF dev.
Best regards;
François-André Danica Philidor, once said that pawns are like the soul of chess, so SIERRA tentatively pursues this principle. In many pawn structures, pieces have "naturally good" and "naturally bad" squares; the engine takes this information and adds it to the move value when ordering moves and pruning search heuristics, although it usually doesn't work.
And Cool Iris is a slightly different architecture with a large net different from the current green net.
Additionally, Cool Iris and SIERRA (the firts and the last attempt), have served me well in my compilation tests on clang64, using a third-party and PGO, which generally gives them a 5 percent speedup against SF dev.
Best regards;
Mysterious Stockfish Blunder
From one of my current games.
Gaunlets for Fun: SF_PB_280425 vs Stockfishes & Lc0
Gaunlets for Fun: SF_PB_280425 vs Stockfishes & Lc0 Test Conditions - OS: Windows 11 Home - HW: 13th Gen Intel(R) Core(TM) i7-13700HX, Lenovo 5 Legion - Gui: Cute Chess Gui - Tournameent Type: Gaunlet - Time Setup: 40m/2m - Games by Match: 100 as much as possible. - Ponder: ON - Hash Table: 1024Mb - Threads: 2 - Syzygy endgame tablebases: 3,4,5 complete + most important 6,7. - Opening Book: Different Opening Test Sets. Now: 2025-05.pgn, Bias [ -0.80, 0.80 ] SF_PB_280425 : Brainlearn 30 : 100 : 2+ : 97= : 1- : 50.5% Lc0.cuda 0.31.2 754042.15.192 : 100 : 13+ : 87= : 0- : 56.5% Lc0.cuda 0.31.2 791556.15.192 : 100 : 9+ : 91= : 0- : 54.5% Lc0.cuda 0.31.2 811971.19.512 : 100 : 10+ : 90= : 0- : 55.0% Lc0.cuda 0.31.2 824884-15.768 : 100 : 8+ : 92= : 0- : 54.0% Stockfish 15 : 100 : 4+ : 96= : 0- : 52.0% Stockfish 15.1 : 100 : 6+ : 94= : 0- : 53.0% Stockfish 16 : 100 : 4+ : 96= : 0- : 52.0% Stockfish 16.1 : 100 : 0+ : 100= : 0- : 50.0% Stockfish 17 : 100 : 0+ : 98= : 2- : 49.0% Stockfish 17.1 : 100 : 1+ : 98= : 1- : 50.0% Stockfish Dev.25042718 : 100 : 0+ : 100= : 0- : 50.0%
Gaunlets for Fun: SF_PB_280425 vs Stockfishes & Lc0
Gaunlets for Fun: SF_PB_280425 vs Stockfishes & Lc0 Test Conditions - OS: Windows 11 Home - HW: 13th Gen Intel(R) Core(TM) i7-13700HX, Lenovo 5 Legion - Gui: Gui: Arena Chess Gui 3.5.1 - Tournameent Type: Gaunlet - Time Setup: 40m/2m - Tournameent Type: Gaunlet - Time Setup: 40m/2m - Games by Match: 100 as much as possible. - Ponder: ON - Hash Table: 1024Mb - Threads: 2 - Syzygy endgame tablebases: 3,4,5 complete + most important 6,7. - Opening Book: Different Opening Test Sets. Now: 2025-05.pgn, Bias [ -0.80, 0.80 ] SF_PB_280425 : Brainlearn 30 : 100 : 2+ : 97= : 1- : 50.5% Lc0.cuda 0.31.2 754042.15.192 : 100 : 13+ : 87= : 0- : 56.5% Lc0.cuda 0.31.2 791556.15.192 : 100 : 9+ : 91= : 0- : 54.5% Lc0.cuda 0.31.2 811971.19.512 : 100 : 10+ : 90= : 0- : 55.0% Lc0.cuda 0.31.2 824884-15.768 : 100 : 8+ : 92= : 0- : 54.0% Stockfish 15 : 100 : 4+ : 96= : 0- : 52.0% Stockfish 15.1 : 100 : 6+ : 94= : 0- : 53.0% Stockfish 16 : 100 : 4+ : 96= : 0- : 52.0% Stockfish 16.1 : 100 : 0+ : 100= : 0- : 50.0% Stockfish 17 : 100 : 0+ : 98= : 2- : 49.0% Stockfish 17.1 : 100 : 1+ : 98= : 1- : 50.0% Stockfish Dev.25042718 : 100 : 0+ : 100= : 0- : 50.0%
Notice: I updated the GUI used for the Gaunlet.
Gaunlets for Fun: CoolIris v18 vs Stockfishes
Gaunlets for Fun: CoolIris v18 vs Stockfishes Test Conditions - OS: Windows 11 Home - HW: 13th Gen Intel(R) Core(TM) i7-13700HX, Lenovo 5 Legion, NVIDIA GeForce RTX 4060 GPU - Gui: Arena Chess Gui 3.5.1 - Tournameent Type: Gaunlet - Time Setup: 40m/2m - Games by Match: 100 as much as possible. - Ponder: ON - Hash Table: 1024Mb - Threads: 2 - Syzygy endgame tablebases: 3,4,5 complete + most important 6,7. - Opening Book: Different Opening Test Sets. Now: 2025-05.pgn, Bias [ -0.80, 0.80 ] CoolIris v18 : Brainlearn 30 : 100 : 0+ : 99= : 1- : 49.5% CoolIris v11.80 : 100 : 0+ : 98= : 2- : 49.0% CoolIris v13.90 : 100 : 1+ : 100= : 0- : 50.5% CoolIris v13.97 : 100 : 4+ : 95= : 1- : 51.5% CoolIris v14 : 100 : 3+ : 97= : 0- : 51.5% CoolIris v14-MPV : 100 : 0+ : 100= : 0- : 50.0% CoolIris v15 : 100 : 2+ : 97= : 1- : 50.5% CoolIris v16 : 100 : 1+ : 97= : 2- : 49.5% CoolIris v17 : 100 : 3+ : 97= : 0- : 51.5% Stockfish 15 : 100 : 11+ : 89= : 0- : 55.5% Stockfish 15.1 : 100 : 2+ : 98= : 0- : 51.0% Stockfish 16 : 100 : 4+ : 95= : 1- : 51.5% Stockfish 16.1 : 100 : 1+ : 99= : 0- : 50.5% Stockfish 17 : 100 : 4+ : 96= : 0- : 52.0% Stockfish 17.1 : 100 : 1+ : 99= : 0- : 50.5% Stockfish Dev.25042718 : 100 : 1+ : 98= : 1- : 50.0%