Author: Miguel A. Ballicora
Date: 09:47:19 05/13/02
Go up one level in this thread
On May 13, 2002 at 11:20:56, Robert Hyatt wrote: >On May 13, 2002 at 00:22:28, Russell Reagan wrote: > >>If I have one legal move array, will that work for searching recursively? For >>example, in AlphaBeta() where it says GenerateLegalMoves(), if I have a global >>variable that won't work, at least it doesn't seem like it would, because you >>would be writing over the moves from previous plys. So it would seem like you >>would need a local variable for this. But in TSCP it looks like Tom used a >>global legal move list in gen_dat[] but he also has a first_move[] array that >>could do something to make this work. I think I get the general idea, but I'm >>not very clear on it. >> >>It seems like a simpler solution would be to use a local variable inside >>AlphaBeta() to store the legal moves, then they would be saved on the stack >>during recursive calls. Is there any downside to this method? >> >>Thanks, >>Russell > > >You can use a global array, but for each ply of search, you need an index into >this table that is different. IE ply1 could go from entry 0 to 41, ply 2 could >then go from entry 42 to 100, etc... > >Not a good idea if you plan on doing a threaded (SMP) search, of course... You could have one of these structures per thread and pass the address in a pointer. If all the global structures needed in a program are in one big structure only one pointer will be needed to pass. I am not doing SMP, but in the multithreaded approach I am using (one thread is the interface, the other is the engine) that is what I do. The interface has their own board to check legality etc. and the the engine has their own board to "think". I thought that this may work in an SMP program. Is there any serious drawback? Miguel
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.