Author: Vincent Diepeveen
Date: 19:07:10 08/10/99
Go up one level in this thread
On August 09, 1999 at 21:11:55, Dave Gomboc wrote: >On August 09, 1999 at 15:43:58, Vincent Diepeveen wrote: > >>Also that 100% can't be correct. It's seeminglessly just dividing >>the number of hashtable writes nodes by the number of tuples and if that's >>>= 1 then it puts it at 100%. >> >>However practically spoken this is impossible because of chaining. >>So it should be more like 90% for a long period of time and then slowly >>get up. 100% should take up to an hour sometimes if hashsize is huge. > >It really depends on what's being reported. A moving average of the collision >rate of the last N attempted inserts would reach 100% sooner than your measure >would. Such a statistic is a valid and useful measure of hash permeability. well it's not reported the way the user *thinks* it is reported. it's just an 'if then else' now. and not a true fill. a true fill is easily measurable by counting how many tuples are left empty. >Dave
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.