The stabilizer judge is a very time time.
I started to take a lot of weight, I have a lot of optimization of this part of the program, but it still consumes approximately half of the entire valuation. Later, I found that most of the chess games, basically no stabilization in 40 steps, so many calculations were actually wasted. And starting from the 40-step final search begins.
Now I think, starting from step 30 to join a stable judge may be a good attention. That is to say to the different stages of the game, the valuation function must change, huh, huh.
