OpenChess

Hi all,
I'm trying to read as many relevant threads on here as possible and one topic I haven't seen anything on is searching without the make/unmake move combo. What I'm thinking about is that since most search algorithms are implemented in a recursive manner could it not be more efficient (faster) to "makemove" on temporary position and then let exiting the recursion "pop" the position back to where it was before the "makemove"? I know that each call to the recursion would require the copying of a minimal set of data, but wouldn't this be compensated for by the lack of an "unmake" (including the game-state/game-history)?

Any information regarding discussion of this topic, together with thoughts on this, very welcome.

Thanks,
Graham....

GrahamA wrote:Hi all,
I'm trying to read as many relevant threads on here as possible and one topic I haven't seen anything on is searching without the make/unmake move combo. What I'm thinking about is that since most search algorithms are implemented in a recursive manner could it not be more efficient (faster) to "makemove" on temporary position and then let exiting the recursion "pop" the position back to where it was before the "makemove"? I know that each call to the recursion would require the copying of a minimal set of data, but wouldn't this be compensated for by the lack of an "unmake" (including the game-state/game-history)?

Any information regarding discussion of this topic, together with thoughts on this, very welcome.

Thanks,
Graham....

Search for copy/make. CCC has many such discussions. Gets rid of Unmake, but adds a LOT of memory accesses to copy the complete board position. So much in fact that most choose to do Make/Unmake to avoid all the excessive memory traffic that is expensive. Early Crafty's were Copy/Make. But not for 18 or so years now. I even tried doing it again a couple of years ago, but still found it too expensive and slower than make/unmake.

Thanks for your knowledgeable reply.

Graham....