He stopped developping Brainfish as a Stockfish fork, because Stockfish includes large-page support and CFish includes it and the ability to use multiple books from Brainfish. And CFish is even better because it's faster.
Hence, instead of taking time to maintain a Stockfish derivative, the author prefer to be able to focus on the Cerebellum opening book, letting people use it with CFish.
chrisw wrote: ↑Fri Sep 25, 2020 10:47 am
Up to now I've not been using a book (multi-game testing was always using cutechess and a big random book, same for both sides, to give variance).
Since getting relatively trashed in the monthly Dutch blitz tournament, I hurriedly put a book together (semi-random move chooser, similar chooser method to polyglot based on frequency for the moment) and am testing engine plus book vs engine, and seem to be getting 63% win rate, or about 100 Elo or so. Does this figure? 100 Elo points for a primitive book?
Book takes about an hour to construct (Python) pulling positions from about 3 million CCRL PGNs, hashing them and accumulating the game results.
Book against no book depends on time control, on the opening strength of the engine and on the book chosen. Shorter TC ---> more Elo. Weaker engine in the openings ---> more Elo. Better book specifically built to play against no book engine ---> more Elo. So, if TC=blitz, classical AB Engine, Elo 2200, a well built book against no book can gain 200+ Elo points. If TC=long, engine is Lc0 (or SF NNUE) with a good net, any good book against no book ---> less than 30 Elo points.
Usually an engine and book plays versus another or same engine and book. These are book wars, and the differences even using solely strong engines, can be large, sometimes again larger than 50 Elo points.
