Stockfish AI with Atomkraft piece values plays very well. It is worth trying:
( Windows 64 bit executable is in the 'psrc' directory )
I finally tested these engines against the original multiple variant version of Stockfish by playing an engine tournanemt between the different flavors of Stockfish, BigBang and Atomkraft, from the starting position, at 40 moves / 3 minutes time control, with each pariing played with both colors. The games between BigBang and Atomkraft were removed from the schedule.
StockfishAI finished on a +2 score, with all games decisive. StockfishTFA finished on an equal score with 2 draws. Stockfish finished on a -3 score with 3 draws.
It can be seen that there is a marked difference between Stockfish AI which plays with Atomkraft piece values and the other two flavors of Stockfish, which play with original Stockfish piece values ( in case of StockfishTFA this is modified with pawn and queen bonuses ). Stockfish plays very aggressive atomic chess, similarly to Atomkaft and this sharp playing style mostly pays off. StockfishTFA and Stockfish are more conservative, with StockfishTFA having a much better evaluation than Stockfish at this time control for games played form the starting position.
It has to be noted that also Atomkraft's games were all decisive with a whopping +4 score. It is clear that Atomkraft at this time control is still the strongest engine for games played from the starting position.
All the games with move times and evaluations:
That's very cool that Stockfish with ATOMKRAFT piece values outperformed Stockfish. I'll carefully think about the best way to change material weights since that affects middlegame/endgame play.
You have to be careful because Atomkraft is Atomkraft and Stockfish is Stockfish. Which make Stockfish play much better at quick time controls, may backfire at longer time controls ( as long TC games against BigBang show ). Also StockfishAI does not like being forced into positions it would not play by its own. So if you force it to play 1. e3 or 1. Nh3, it does not understand the position and cannot convert its advantage.
This topic has been archived and can no longer be replied to.