Which Gothic end-games are won?

OK, thanks. These were the btm numbers. The wtm numbers you gave me on the chat last Monday also were a 100% match.

I must have goofed with the position, perhaps I simply mixed up btm and wtm positions. If al the numbers match, it is inconceivable that some of the positions themselves should be wrong.

So far I only built for statistics; interrogating the tablebase for a best move is only possible for wtm and in an extremely cumbersome way (by typing the hexdecimal index of the position!). I want to make the generator into a WinBoard engine, so that (after generation), you can play the endgame against it (or let it play against itself) like it was a normal Chess program.

Now that I am sure the results are correct, I will try to do some analysis, filtering out the positions that do not really belong in a givenend-game, because one of the pieces is trivially lost (by a fork or a skewer, or because two pieces are hanging at the same time). It might be to difficultfor me to do 5-men endings with the current generator, though, because the 16x16 board eats memory.

If the numbers match, so should everything else, I agree with that. Feel free to use the Gothic Vortex tablebases to check your info.

My favorite 5-piece endgame is:

http://www.gothicchess.com/endings/kakbn/game.htm

Archbishop vs. Bishop + Knight, a mate in 202 for the Archbishop!

Maybe I can put all of my generator summary files online somewhere.

OK, I guess I simply copied down the position wrong. When I rerun the generator, it indeed comes up exactly with the position you mentioned.

Funny thing is that if I run it for an 8x10, in stead of a 10x8 board, it finds another position. (Well, after all, there should be 4 such positions, so this is not too surprising.) This position has the Archbishop on a8 in stead of A1, but the Kings in the same place.

If the i- and j-files would not have been there, it would have been the diagonal mirror image of the position you gave.