Endgame mistake : could win but draws about koivisto

[Luecx]

Did the engine have Tablebases?

[Luecx]

Also how sure are you Nc5 is not winning? The score is +8. This seems to be very decisive. Or is this some cursed win?

[Luecx]

So i checked the position. its infact not playing the optimal moves without TB but I would not say that this is easy or even trivial

[AlexBrunetti]

Also how sure are you Nc5 is not winning? The score is +8. This seems to be very decisive. Or is this some cursed win?

Nc5 can't win because White plays NxP drawing immediately.

Alex

[Luecx]

I see.

Knowing that NNKvk is drawn is not known by Koivisto. Guess this is some edge case.

[tissatussa]

TBs were OFF in my CuteChess settings .. I used the online TBs at https://syzygy-tables.info to inspect this position .. indeed NNKvk is "some edge case" .. however NNKvkp is winning when one opponent (!) pawn is on the c-file or f-file .. eg. see https://en.wikipedia.org/wiki/Two_knights_endgame .. i report this because many other engines DO move h4 in this position, although that might be luck : they may not be aware of the theory ..

[altarchess]

This should be fine as even 4 man TB's cover this case.