Author: Dann Corbit
Date: 13:34:52 01/31/02
Go up one level in this thread
On January 31, 2002 at 16:19:51, Uri Blass wrote: >On January 31, 2002 at 14:36:04, Dann Corbit wrote: > >>On January 31, 2002 at 14:31:38, David Rasmussen wrote: >> >>>On January 31, 2002 at 13:13:57, Dann Corbit wrote: >>> >>>>On January 31, 2002 at 06:58:28, David Rasmussen wrote: >>>>[snip] >>>>>>Just like "ply" means 20 different things to 20 different programmers. Even >>>>>>"nodes" does not always mean the same thing. >>>>>> >>>>> >>>>>Agree with you on nodes, but ply? Ply is pretty well defined, I think. >>>> >>>>For sure it is not. In fact, even when we agree, we disagree. Every single >>>>chess program will have a tree of a different shape. So even when we count >>>>plies the same way, the actual search can be incredibly different (with the >>>>number of nodes visited differing by several orders of magnitude). Compare, for >>>>instance, Mchess with Goliath. >>>> >>>>Junior [for instance] does not count plies the same way as other programs. >>>> >>>>Ply is ill defined. In fact, I think it is actually impossible to define it >>>>accuracy, except in the brute force sense. And absolutely nobody exhausts a ply >>>>when doing chess games with an engine. >>>>[snip] >>> >>>That is bull. A ply is a half move and that is that. What you are talking about >>>is: what does it mean when we say a program searches 8 ply? Of course if that is >>>the question, the answers are as many as there are programs. But in this thread >>>we are talking about a full width alpha-beta tree (at least, since DB had >>>singular extensions, which requires extra searches), of some fixed depth search >>>in a given program (that is, move ordering and evaluation forms the tree). >> >>I suggest that you lookup with the CCC search engine the dozens of posts that >>shot *ME* down when I tried to insist that a ply is a ply is a ply. >> >>The thorough and convincing arguements against it showed that I was clearly >>wrong. >> >>As you are clearly, unmistakeably, and totally wrong right now. > >He is not wrong if we talk about full width alpha beta. Which is the same as saying "If we all wore giant balsa-wood shoes, we could walk on water." Nobody wears them, as far as I know. There is no chess program on the planet that does that. (Except -- of course -- for the rankest sort of amateur program). Beowulf will often spend 80% of its nodes in quiescense. What about programs that have a very sophisticated SEE at the end of a ply? What about the way that Junior counts plies? Some related threads: http://www.it.ro/ccc_search/ccc.php?find_thread=118884 http://www.it.ro/ccc_search/ccc.php?find_thread=120838 http://www.it.ro/ccc_search/ccc.php?find_thread=13258 http://www.it.ro/ccc_search/ccc.php?find_thread=71072 http://www.it.ro/ccc_search/ccc.php?find_thread=78442 http://www.it.ro/ccc_search/ccc.php?find_thread=109958 I could not find the thread where I was demanding that a ply is at least a good measure to compare one program with the next as far as searching [it's not -- I was wrong]. If I recall correctly, you were amongst those who gave me correction! ;-)
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.