Larry, are those algorithms for starting/finishing iterations not been applied in X time per move time controls? Does the engine knows that a move have to be made in X time? or acts like an infinite analysis that is suddenly truncated after X time?lkaufman wrote: But I don't agree about testing with X time per move. This is a highly inefficient way to test. Most (all?) engines make some effort to finish iterations and/or not to start ones they cannot finish. This is highly beneficial. Even for analysis, this is an important feature. Testing with fixed time justs wastes a lot of time and fails to give credit for good algorithms for starting/finishing iterations etc.
Komodo 6, Houdiini 3, latest Stockfish dev. all equal.
Moderator: Ras
-
Aser Huerga
- Posts: 812
- Joined: Tue Jun 16, 2009 10:09 am
- Location: Spain
Re: Komodo 6, Houdiini 3, latest Stockfish dev. all equal.
-
lkaufman
- Posts: 6284
- Joined: Sun Jan 10, 2010 6:15 am
- Location: Maryland USA
- Full name: Larry Kaufman
Re: Komodo 6, Houdiini 3, latest Stockfish dev. all equal.
All engines that I know of treat X seconds per move as just infinite analysis stopping at X seconds. One could devise algorithms that optimize for X seconds per move testing, but since no one tests this way there is no incentive to work on doing this.