Please sign in to comment.
Use adaptive search depth at higher levels of play.
So far this is very simple. If the search doesn't take very long at all, it will be retried at the next depth. And again, if it still doesn't take long. This is slightly wasteful, in that the original search is thrown away, and the level is recalibrated every move, not estimated based on previous moves. But this waste is not huge, given that each deeper search is almost always an order of magnitude slower than the shallower one, so the waste is in the order of a few percent.
- Loading branch information...
Showing with 19 additions and 3 deletions.