Author: José Carlos
Date: 17:47:04 01/27/02
Go up one level in this thread
On January 27, 2002 at 17:47:12, Uri Blass wrote: >On January 27, 2002 at 16:08:21, walter irvin wrote: > >>what is depth of search worth in elo per ply ?????? i mean is the jump from 5 >>ply to 6 ply worth the same as 14 ply to 15 ply ????? i wonder if there is a >>relationship between the different plys and the percentage of times that the >>computer changes its mind on the moves at diferent plys .i mean if the program >>only changes its mind .01 % of the time between 14 ply and 15 ply and it changes >>its mind 14 % between 5 ply and 6 ply then how can the gain in ply be worth the >>same elo gain ??? and if this indeed the case have we just about gotten all we >>are going to get out of hardware speed up ???????? > >There is some investigation of the question how much program change their mind >and programs often change their mind even between 14 and 15 plies(my experience >say that they may change their mind even at 19 plies to a new move that they did >not consider before. > >Uri But even if this is correct (which I don't know) it seems reasonable that, at some point, the curve of strength increase versus depth must start to flatten and eventually become flat, because its obvious that searching 1001 doesn't give much compared to searching 1000 plies, and I don't think there's a step at certain depth, so it must be a progressive flattening. Just my opinion. José C.
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.