Difference between revisions of "Copy-Make"
GerdIsenberg (talk | contribs) (Created page with "'''Home * Chess * Position * Copy-Make''' While traversing a Search Tree, the '''Copy-Make''' approach keeps and updates local copies...") |
GerdIsenberg (talk | contribs) |
||
| (3 intermediate revisions by the same user not shown) | |||
| Line 1: | Line 1: | ||
'''[[Main Page|Home]] * [[Chess]] * [[Chess Position|Position]] * Copy-Make''' | '''[[Main Page|Home]] * [[Chess]] * [[Chess Position|Position]] * Copy-Make''' | ||
| − | While traversing a [[Search Tree]], the '''Copy-Make''' approach keeps and updates local copies of certain aspects of a [[Chess Position|chess position]] inside an array indexed by [[Ply|ply]], which could also be interpreted as explicit, random accessible search [[Stack|stack]]. It usually refers the irreversible aspects of the position, like [[En passant|ep state]], [[Castling | + | While traversing a [[Search Tree]], the '''Copy-Make''' approach keeps and updates local copies of certain aspects of a [[Chess Position|chess position]] inside an array indexed by [[Ply|ply]], which could also be interpreted as explicit, random accessible search [[Stack|stack]]. It usually refers the irreversible aspects of the position, like [[En passant|ep state]], [[Castling Rights|castling rights]] and the [[Halfmove Clock|halfmove clock]], which can not [[Incremental Updates|incrementally updated]] during [[Unmake Move|unmake move]]. Some programs even keep reversible stuff inside an array, to avoid incremental update during unmake. Copy-Make is required, if aspects need to be accessed randomly in the current branch from the [[Root|root]] (or even starting game position) to the current one. |
=Copy-Make= | =Copy-Make= | ||
| Line 42: | Line 42: | ||
: [https://groups.google.com/group/rec.games.chess.computer/msg/730c03a83bf92807 Re: cheaper search ?] by [[Shaun Press]], [[Computer Chess Forums|rgcc]], April 28, 1997 » [[Vanilla Chess]], [[KnightCap]] | : [https://groups.google.com/group/rec.games.chess.computer/msg/730c03a83bf92807 Re: cheaper search ?] by [[Shaun Press]], [[Computer Chess Forums|rgcc]], April 28, 1997 » [[Vanilla Chess]], [[KnightCap]] | ||
* [https://www.stmintz.com/ccc/index.php?id=40653 Unmake move v copy the board] by Hugh Cumper, [[CCC]], January 24, 1999 | * [https://www.stmintz.com/ccc/index.php?id=40653 Unmake move v copy the board] by Hugh Cumper, [[CCC]], January 24, 1999 | ||
| + | * [https://www.stmintz.com/ccc/index.php?id=60557 Move Make/Unmake Questions] by [[Daniel Homan]], [[CCC]], July 15, 1999 » [[Unmake Move]] | ||
==2000 ...== | ==2000 ...== | ||
* [https://www.stmintz.com/ccc/index.php?id=312031 The need to unmake move] by [[Mathieu Pagé]], [[CCC]], August 19, 2003 » [[Unmake Move]] | * [https://www.stmintz.com/ccc/index.php?id=312031 The need to unmake move] by [[Mathieu Pagé]], [[CCC]], August 19, 2003 » [[Unmake Move]] | ||
Latest revision as of 11:32, 21 June 2020
Home * Chess * Position * Copy-Make
While traversing a Search Tree, the Copy-Make approach keeps and updates local copies of certain aspects of a chess position inside an array indexed by ply, which could also be interpreted as explicit, random accessible search stack. It usually refers the irreversible aspects of the position, like ep state, castling rights and the halfmove clock, which can not incrementally updated during unmake move. Some programs even keep reversible stuff inside an array, to avoid incremental update during unmake. Copy-Make is required, if aspects need to be accessed randomly in the current branch from the root (or even starting game position) to the current one.
Contents
Copy-Make
// make
memcpy (&position[ply+1].irrvrsAspects,
&position[ply ].irrvrsAspects,
sizeof(irrvrsAspects));
ply++;
update (position[ply], move)
...
// unmake
ply--;
// position[ply] is still valid
Stack
The alternative, to maintain those irreversible aspects inside a global structure, would require a stack (LIFO), with push and global update during make, and pop from stack to global structure during unmake, and therefor higher memory bandwidth for copying back and forth.
// make push (position.irreversibleAspects); ply++; update (position, move) ... // unmake ply--; pop (position.irreversibleAspects); // position is restored from stack
See also
Forum Posts
1995 ...
- cheaper search ? by James F. Long, rgcc, April 27, 1997 » Tristram
- Re: cheaper search ? by Shaun Press, rgcc, April 28, 1997 » Vanilla Chess, KnightCap
- Unmake move v copy the board by Hugh Cumper, CCC, January 24, 1999
- Move Make/Unmake Questions by Daniel Homan, CCC, July 15, 1999 » Unmake Move
2000 ...
- The need to unmake move by Mathieu Pagé, CCC, August 19, 2003 » Unmake Move
- undo move vs. Position Cloning by BoldReceiver, CCC, September 16, 2009
- Re: undo move vs. Position Cloning by Marco Costalba, CCC, September 16, 2009 » Stockfish
- Re: undo move vs. Position Cloning by Don Dailey, CCC, September 16, 2009 » Doch
- copy/make vs make/unmake test results by Robert Hyatt, CCC, September 19, 2009 » Crafty
2010 ...
- Copy Board vs Unmake Move by ChrisJ, OpenChess Forum, September 29, 2010
- performance of copy-make by Rein Halbersma, CCC, August 02, 2011
- How costly is taking moves back ? by Piotr Lopusiewicz, CCC, April 30, 2013
- Saving info before making a move by CDaley11, OpenChess Forum, December 30, 2013
- copy/make vs make/unmake by Robert Hyatt, CCC, January 07, 2014
- Memory usage in make/unmake vs logic complexity by Matthew Lai, CCC, August 30, 2014
2015 ...
- Unifying make/undo and copy-make by Rein Halbersma, CCC, December 21, 2015
- Copy-make vs Make/Unmake ? by Mahmoud Uthman, CCC, November 12, 2016
