I assume there is rational explanation for this ...

The following just might fall within the scant capabilities of Rybka embusques. Determine if any engines beside Fruit and its un-cromulent offspring find themselves at sea in this position.
http://www.talkchess.com/forum/viewtopic.php?p=427233 ...

I made a cursory glance at Loop 2007 (the 64-bit version). It uses exactly the same PST as Fruit 2.1 . The pawn eval uses the same scores. It shifts the mobility count as with Fruit (by 4,6,7,13) before multiplying, and then the arithmetic is the same (4 for N, 5 for B). I'm not that interested in ...

it does mean you are not going to resolve the score this iteration.

Why not? Other engines just restart the iteration, and resolve the fail low, and get a true score for the move that failed high-then low (though it's something that happens extremely rare anyway, I would have problems ...

I'm using Shredder Classic, here, when I put the engine to analyze a position, it goes at it endlessly, but I also can put it in "Human against the Machine mode", there, I can choose time control, I put it at 20ply (so, I make a move, and Stockfish reaches 20 plies and makes a move, then it's my ...

This is with a new version. The Fail High is resolved correctly, at a bit higher depth than Uly's Stockfish. If I'm correct the main variation with 1... Bb4 can also be slightly below 0.00, and in that case 1... a5 is preferred. But I did not go to extreme depths here:


r2k1r2/pp1n1q1p/1npP4/4p1P1 ...

I left it running a little longer...
.....
1...a5 2.a4 Kc8 3.Qg4 Kb8 4.Be2 Ka7 5.Rf1 Qg6
=/+ (-0.27 !) Depth: 26/9 00:06:00 284mN
1...a5 2.a4 Kc8 3.Nf5 Bb4 4.Qe3 Kb8 5.Be2 Ka7 6.Rg4 Nc5 7.Rf1 Rad8 8.Ne7 Qe6 9.Nf5 Kb8 10.Rh4 Qd7 11.c3 Ba3 12.Rh6 Rde8 13.Ra1 Nxe4 14.Qxe4 Qxf5 15.Qxf5 Rxf5 16.Rxa3 ...

Uly, send me some positions and whatever additional info you can give me via PM. I can try to isolate the issue (and I won't put your Corr games in jeopardy, promise).

I can say what I do in Rainbow Serpent that may help a little. I just let alpha rise a little, including oldalpha, when beta is ...

User is confused by a fail high (sent to GUI) followed by a fail low (not sent)

We will probably change SF behavior to the standard one followed by other engines, namely to send info to GUI only after any fail high / low is resolved.

But Marco, would that not be throwing out the baby with the ...

The behavior is _not_ correct. Perhaps you meant that what he observed actually happens? How can it be correct to get a fail high and then not play that move??? If you fail high and it is not a better move, that's a bug. If you fail high but play the original move because you didn't have time to ...

Okay, confirmed, what Ancalagon says is right, Stockfish goes back to the old move in the new iteration even at infinite analysis, so its behavior is correct.

What is weird is that in all the cases that I've seen, the move that is failing high and is discarded is better than the old move once you ...