Author: Chuck
Date: 15:43:38 12/15/99
Go up one level in this thread
On December 15, 1999 at 13:45:51, Dann Corbit wrote: >On December 15, 1999 at 08:30:36, James T. Walker wrote: >[snip] >>This sounds good to me but isn't it true that if you sort your moves so you >>search the most promising moves first then a 15.5 ply search may be just as good >>as a 16 ply searcch. That is, the moves at the end of the list are usually just >>"Give away" moves or wild sac's. I'm obviously not a programmer so excuse my >>ignorance on the subject. >You make a good point. I think you may still have some advantage. However, at >some point, you will hit the wall on the first move of the next ply. It's >exponential, after all. So this brings up a further point, or hypothesis which we continually debate. Assume you hit this wall, as in this example: Program A is fast and "dumb" and completes ply 14 in 3 minutes. The time control prevents it from completing the main branch another ply. Program B is slower, but "smarter", takes 5 minutes to get to 14 ply, but evaluates all but the last few branches at 14 ply. So in general I would tend to believe that the "smarter" program would have a big advantage having "understood" more and searched to "essentially" the same depth. Of course, this assumes a factor we're never sure of, that Program B is slower due to evaluating more factors, etc., but that it is otherwise just as efficient. Also, due to extensions, you can never be sure they searched to the same depth even if they both searched the same number of ply.
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.