dorsz wrote:im48 wrote:There may be cases where the compilations are weaker than the original, but here we also took into account the fact that Skynet is a very good compiler. His last compilation of Stockfish was the same as ChessMan's and that gives me confidence.
The differences are in speed only and they are really small differences. I'm using this tool for testing speed of my compilations.
https://github.com/zardav/FishBenchim48 wrote:But don't worry, you made two top engines in a very short time and congratulations on that!
I did two because it was optimal number for me. Swordfish as more active playing and tactical engine than Stockfish being only slightly weaker, and Blue Marlin where the most important thing is its sacrificial style. Here is one of the latest results from a test tournament before I released Swordfish 15.2, a proof: (www.mediafire.com/view/b5hpqn5n6kt1w25/15.2vs15.1.png/file#)
I had stronger candidate of new Blue Marlin but I did not release it. The reason is it failed with tactical tests and it failed with the positions I verify the engine to have enough sacrificial style. Blue Marlin 15.1 was clearly better.
(My engines are top engines because they are Stockfish derivatives, none is stronger)
I downloaded FishBench-6.0 and I want to see what it's about, it's completely unknown to me. Thanks!