Mate or not, Stockfish?

In this Queen vs Rook position, the cloud-based analysis says it's a draw (depth 61, evaluation "0")

However, after e.g. Qc8, stockfish (PNACL) quickly finds me a mate in 23 moves.

What's going on? Just an error in the cloud database, or something else? Is there a way for black to draw this against best play (I guess not?)?

(Just out of interest... I myself would not have the slightest chance to win this endgame with the queen against a computer or a good player...)

@ lovlas
Yes, that's what I read, that's why I'm wondering where the "0" computer evaluation is coming from. But maybe it's just a database glitch or something...

Could be. I was wondering if maybe the position was reached in some sort of threefold repetition fashion and therefor saved as a draw. But I have no idea, just speculating.

There is no Stockfish when Stockfish is accessible to 7 men TB.
When Tablebases are turned on ( within number of TB pieces), engines are automatically switched off.
If there is bug, it is not Stockfish, it is TB probe bug or GUI communication bug.


good point. If you click on the tablebase, it indeed shows the winning moves for white, e.g. Kf5 - mate in 40. So I think the stored cloud analysis either didn't use the tablebase, and maybe the computer at this depth only found lines that take more than 50 moves. Or there was a technical communication/programming error or whatever.

In any case, happy to know that white wins, theoretically at least. (;

100% white mates
kf5 or kf4 best move ( lichess ,shredder endgame database win 21 )
my engine says mate in 22 ( no endgame database )
Engine says 0 because engine missed Qc8+

You can't post in the forums yet. Play some games!