Using ShredderBase

In the October 2006 Chess Life, GM Pal Benko's column 'Endgame Lab' discusses 'Computers and Endgames'. Benko gives a number of 5 and 6 piece endgames which he analyzed using tablebases. I need practice using this technique, so I checked his analysis using...

Shredder Computer Chess : Endgame Database
http://www.shredderchess.com/online-chess/online-databases/endgame-database.html

...Everything checked well except Benko's third position, shown in the diagram.

White to play and win
[FEN "8/7n/2K3kP/4P3/8/8/8/6N1 w - - 0 1"]

Benko gave 1.e6, which ShredderBase (SB) confirms as a 'Win in 29'. Now Benko continued 1...Kxh6, although SB says that 1...Kf6 loses slower. After 2.e7 Nf6 3.Kd6 Kg7 4.Ke6 Kg6 5.Ne2, Benko said, 'The computer pointed out 5.Nf3 is quicker'. SB says that both 5.Ne2 and 5.Nf3 'Win in 21.'

This is not an earth shattering disagreement, but I thought it curious that even people using computers can differ on optimum sequences. How long will it take to produce 7-piece tablebases?