Author: Ralf Elvsén
Date: 07:20:52 06/29/01
Some time ago I discussed with Terrapong the ratio of makemove/movegenerations. I don't think this is what he was asking for, but I got interested anyway: Given alphabeta with perfect move ordering and a branching factor = w (effective branching factor = sqrt(w) ), what is the ratio of makemove/movegenerations ? I have asymptotic expressions for "reasonably big" depths: For even search depths: the ratio is (3w + 1)/(w + 3) For odd search depths: the ratio is w*(w + 3)/(3w + 1) There is a pretty extreme odd-even effect. For e.g. w = 36 we have 2.8 (even) and 12.9 (odd). Nothing of this applies when nullmove and hashtables are considered of course... I have checked this reasonably well, but I am under a curse which doesn't allow me to get calculations right the first time... If you want to see some other numbers, I'll see what I can do :) Ralf
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.