Author: Bert van den Akker
Date: 12:36:28 11/22/99
Go up one level in this thread
On November 21, 1999 at 21:12:09, James Robertson wrote: >On November 21, 1999 at 18:57:56, Bert van den Akker wrote: > >>On November 21, 1999 at 16:37:52, Daniel Clausen wrote: >> >>>Hi >>> >>>On November 21, 1999 at 16:04:39, Peter Kappler wrote: >>> >>>>On November 21, 1999 at 14:59:53, Bert van den Akker wrote: >>>> >>>>>From the start position I get a branche factor of about 14% >>>>> >>>>>This is in a 10 ply search from the start position. >>>>> >>>>>The branche factor is defined as: >>>>> >>>>>(total_number_of_moves_looked_at_in_a_node_before_cutoff / >>>>>total_number_of_generates_moves) * 100 >>>>> >>>> >>>>This is how I used to compute branching factor, too. I still think it's the >>>>best way to measure the quality of your move sorting. >>> >>>I've no "experience" in computing a branching factor, but this formula looks a >>>bit >>>strange to me. I would have expected an absolute number, not a percentage. >>> >>>And consider the following: Let's say you try the move in the hashtable first >>>and you >>>get a cuttoff. (That's what you want in most cases.) According to your formula >>>you >>>get a branching factor of 100% - I think you'd like to get a small percentage >>>though! >>> >>>>But, I think most programmers measure branching factor by dividing the total >>>>number of nodes search at depth n by the number searched at depth n-1. >>> >>>This formula seems to be very intuitive to me. >>> >>>Kind regards, >>> -sargon >> >> >>Before I try the hash move I first generate all moves. >>So the percentage will be ok. >> >>In my example on ply 10 >>27632 moves from the hashtable were causing a cutoff. >> >>Not generating all moves would save me around 0.15 second. >> >>My move generator can generate 200000 times all moves in the start >>position. >> >>BvdA > >It is a much better plan to try the hash move before generating all moves. :) > >James Why? I have to check if the hash move is correct. Now I can look in my generated moves and this cost only 0.15 seconds. Otherwise I must write seperate code to check if the hash move is legal. BvdA
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.