I think it is theoretically impossible for a null window to occur in a PV node: a search with such a window can only fail low or high. A PV needs to have an exact score.zamar wrote:Okay, so they are very raremcostalba wrote:I have tested this condition in real games and never found a pv node where beta - alpha == 1However there are only a handful of PV-Nodes/game, so this condition is true 99.999% of time. 0.001% cases won't lead to crash, so it's okay
Stockfish and endgame K-KBN
Moderator: Ras
-
hgm
- Posts: 28461
- Joined: Fri Mar 10, 2006 10:06 am
- Location: Amsterdam
- Full name: H G Muller
Re: Stockfish and endgame K-KBN
-
zamar
- Posts: 613
- Joined: Sun Jan 18, 2009 7:03 am
Re: Stockfish and endgame K-KBN
Okay, I finally figured it out. search_pv is called after search-call when (value > alpha && value < beta). It's impossible to fullfil this condition when alpha = beta - 1.
Just a newbie mistake
BTW Marco, I think I found a new bug related to your qsearch-TT implementation, but I think I'd better send it by email. (This discussion doesn't belong to "Tournaments and Matches" category anyway.)
Just a newbie mistake
BTW Marco, I think I found a new bug related to your qsearch-TT implementation, but I think I'd better send it by email. (This discussion doesn't belong to "Tournaments and Matches" category anyway.)
Joona Kiiski