The CCLR rating results show that while several engines are close, none of them have proven superior to Stockfish.
It’s unfortunate the Swordfish engine was not included in the tournament. Swordfish is a Stockfish derivative that the developer says can be useful for analyzing sharp positions where Stockfish can overlook some tactics. As I reported back in May, in order to test Swordfish, I participated in a quad correspondence tournament that included one ICCF Master.
Where possible I played gambit openings to test its ability to find hidden tactics that Stockfish may have missed. Of course, the assumption was that my opponents were using Stockfish. Swordfish did not find any hidden resources and, in fact, in many cases its moves matched those of Stockfish.
A couple of years ago I also tested the highly rated Berserk engine in correspondence play because of its tactical ability and the possibility that it might find some hidden resources in gambit openings. It didn’t and by move 20 my opponent was starting to register a slight advantage of 1/4 to 1/3 of a Pawn and my feeling was that Berserk was starting to lose control, so it was time to switch to Stockfish. The game was drawn in 27 moves.
Even though the claim is made that some engines are better than Stockfish in some areas I have not found that to bet he case. That said, let me hasten to add that I am only a casual user and have not done extensive testing. My procedure has been to let the engine run 15 minutes to maybe 3 hours on 10 core. If the evaluation of more than one move was close then they were all examined.
The conclusion is that for correspondence play where engine use is permitted, Stockfish is still the absolute strongest.
No comments:
Post a Comment