Stefan Pohl Computer Chess

Home of famous UHO openings and EAS Ratinglist

Lc0 or other GPU-Neural Nets versus Stockfish 15.1 testing

Playing conditions:

Hardware: Ryzen 7 6800H 2.6GHz Notebook, RTX 3060 GPU, Windows 11 64bit, 32GB RAM

Cuda version installed: Cuda 11.7

Speed: Stockfish 15.1 plays with 14 Threads (=7 cores). Lc0 minibatchsize parameter is set to the best value for each netsize, depending on Lc0's benchmark with backendbench --clippy.

Hash: 2048 GB Hash for Stockfish 15.1 / 8192 RamLimitMb for Lc0

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: UHO_2022_6mvs_+120_+129.pgn. Download my UHO 2022 openings here

Ponder, Large Memory Pages & learning: Off

Thinking time: 2min+2sec for Lc0 and 1min+1sec for Stockfish 15.1: I measured nps on my system and compared these values with the TCEC: My CPU is way too fast, compared with Lc0 running on my RTX 3060 GPU, so it makes sense to set the thinking-time of Stockfish to only 50% of the thinking-time of Lc0. For compensating the fast CPU and the fact, that in TCEC Lc0 benefits from fast hardware and long thinking-time (both is better for Lc0, not for Stockfish)

One testrun takes around nearly 5 days. Average game-duration: 6min 45sec

Each Lc0 / Neural Net plays 1000 games vs. Stockfish 15.1 with my UHO 2022 openings

Learn more about Lc0 (getting started in a GUI, links to net-downloads, FAQs, development-informations and the Leela-Blog) here

Latest update: 2023/05/26: Lc0 0.30dev BT2-4510 (successor net of TCEC 24 net)

Download latest Lc0 dev binaries here

Download all played games (games of the old test-setups, too): here

Program Elo + - Games Score Av.Op. Draws

1 Stockfish 15.1 avx2 : 0 5 5 9000 58.9% -64 49.1% 2 Lc0 0.30dev T1-4000 (15x768) : -39 15 15 1000 44.5% 0 49.8% 3 Lc0 0.30dev 811107 (19x512) : -41 15 15 1000 44.1% 0 46.1% 4 Lc0 0.30dev TCEC 24 : -42 13 13 1000 44.1% 0 51.0% 5 Lc0 0.30rc1 T1-4000 (15x768) : -44 15 15 1000 43.7% 0 49.8% 6 Lc0 0.30dev BT2-4510 (15x768) : -45 14 14 1000 43.5% 0 47.5% 7 Lc0 0.30dev T1-30875 (15x768) : -45 15 15 1000 43.5% 0 47.5% 8 Lc0 0.30dev TCEC 20 : -90 15 15 1000 37.5% 0 50.5% 9 Lc0 0.30dev TCEC 22 : -95 15 15 1000 36.8% 0 49.4% 10 Lc0 0.30dev TCEC 18 : -133 16 16 1000 31.9% 0 50.5%

Games : 9000 (finished)

White Wins : 4538 (50.4 %) Black Wins : 41 (0.5 %) Draws : 4421 (49.1 %)

Below the 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 15.1 avx2 : 0 ---- 4500 2019 2066 415 67.8 100 2 Lc0 0.30dev T1-4000 (15x768) : -79 23 500 62 265 173 38.9 59 3 Lc0 0.30dev 811107 (19x512) : -83 22 500 53 278 169 38.4 61 4 Lc0 0.30dev TCEC 24 : -87 22 500 56 266 178 37.8 57 5 Lc0 0.30rc1 T1-4000 (15x768) : -90 22 500 62 250 188 37.4 56 6 Lc0 0.30dev T1-30875 (15x768) : -93 22 500 60 251 189 37.1 54 7 Lc0 0.30dev BT2-4510 (15x768) : -94 22 500 60 249 191 36.9 100 8 Lc0 0.30dev TCEC 20 : -190 24 500 25 203 272 25.3 82 9 Lc0 0.30dev TCEC 22 : -206 24 500 25 186 289 23.6 100 10 Lc0 0.30dev TCEC 18 : -315 32 500 12 118 370 14.2 ---

You can download my Gamepairs Rescorer Tool right here

Mention, that this is not a ratinglist, but only a performance test of Lc0 with different NNs versus Stockfish. For a real ratinglist including Lc0 running on a RTX-GPU (with a valid Leela-Ratio of 1.0), please visit Andreas Strangmueller's excellent website. Just click here

Datenschutzerklärung