Chess engine and EGTB: solve this FEN

Hello,

The FEN: 8/Q7/5r1p/8/6k1/7P/7K/8 b

It is black to move.

I would like to know if your engine(s), coupled with the 6-men Syzygy EGTB is able to find the mate in #26?

I am experimenting weird outputs from many engines on different computer system and I need to know if the problem is only here, or if others have the same problem at solving it.

The story behind all this:  In a computer chess game, my engine was winning starting from move 47.  Its eval jumped to +128 (I was using a Stockfish 8 derived engine) and my opponent was using a latest Stockfish 8 build.  After move 58. gxh3+ (which is the FEN posted above), my opponent's eval gives him a losing game and a mate in #26 for me.  On my side, my eval was still +128 (I was with white) but never was any mating solution found.  The game went on for 70 more moves and eventually ended in a draw !!!

If you try the FEN above on any online end-game tablebase viewer, it will give you that the game ends in effectively in 26 moves.  You can check one here at http://www.shredderchess.com/online-chess/online-databases/endgame-database.html

So I was dumbfounded by this and started testing this on all my systems. I checked my Syzygy EGTB files for possible corruption but their checksum were still ok.  Yet, on any system being 4,6, 12 or 24 CPUs, I can't solve this mate and the EGTB can't either.

So I am asking for your help!

Thx!