The current move ordering is evaluating the board state for EVERY board in a given depth.
This is obviously extremely bad performance-vise and refactoring this
should result in a decent performance boost, though it will also reduce
alpha-beta cutoffs.
The behavious of the current move ordering actually explains the following things
(A) The reason the timing of the best move gen is so much higher then it should be
(B) Why there are so much more hash collisions then possible
The current move ordering is evaluating the board state for EVERY board in a given depth. This is obviously extremely bad performance-vise and refactoring this should result in a decent performance boost, though it will also reduce alpha-beta cutoffs.
The behavious of the current move ordering actually explains the following things
(A) The reason the timing of the best move gen is so much higher then it should be
(B) Why there are so much more hash collisions then possible