Author: Will Singleton
Date: 19:38:17 04/27/99
Go up one level in this thread
On April 27, 1999 at 16:21:17, KarinsDad wrote: >On April 27, 1999 at 15:47:09, Will Singleton wrote: > > >> >>Thanks for the explanation, sounds like you're making good progress. It always >>seemed to me that the overhead of incrementally updating the attack-tables at >>each node would negate the advantage derived at move-gen. But I guess you >>haven't got far enough to test that trade-off yet. >> >> >>Will > >Yes, incrementally updating the attack tables is a BEAR. > >I was working on that last night and decided to not even bother yet. > >The steps required are: > >1) Clear all piece attacks for the source square of the moving piece >2) Clear all piece attacks for any long range piece that attacks the source >square >3) If a piece is taken, clear all piece attacks for any piece taken at the >destination square >4) If a piece is not taken, clear all piece attacks for any long range piece >that attacks the destination square and then set all piece attacks for any long >range piece that attacks the destination square >5) Set all piece attacks for any long range piece that attacks the source square >6) Set all piece attacks for the destination square of the moving piece > >However, this will result in considerably less (maybe 70%) work than creating >the attack tables from scratch since most pawn, knight, and king (and all long >range pieces that did not attack the source or destination square) pieces will >not be affected. > >I also could not come up with a way to quickly turn off the bits (for example in >step 4 above) that are not needed without turning them all off and then >re-performing the calculation for each piece. I'm sure there is a way to do it >(which would speed this up), but I haven't attempted to figure it out yet. > >Your idea that the incremental calculation basically offsets the gain in move >gen may be correct. However, since I am also going to use a derivative of the >piece attack tables in the evaluation, I am sort of killing two birds with one >stone. So, I'm hoping that I will gain more than I lose. Also, I'm hoping that I >can use the Piece Attack tables for other things (such as quickly calculating >open files, etc.), so I'm hoping that it will give me even more. > >KarinsDad :) I look forward to seeing the results! Will
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.