Author: Reinhard Scharnagl
Date: 02:34:10 10/29/03
Go up one level in this thread
On October 28, 2003 at 21:29:18, Vincent Diepeveen wrote: >On October 28, 2003 at 17:55:14, Reinhard Scharnagl wrote: >>because 64-bit random numbers mostly will be split into an address-part >>and a key part, I do not see a need for 64-Bit numbers but concatenate >>two 32-Bit numbers, using only special generated numbers with an equal >>bit distribution (because of a better quasi hamming-distance). >Wrong, i had very bad experiences combining 2 numbers of 32 bits when i had >programmed in a check how many collissions that gave, it exponentially became >more collissions each x million nodes when i did this very buggy combination. One should always see the _combination_ of both: a) concatenating and b) selecting special random numbers (having equal distibuted bits). >I might be running again a collission test at the supercomputer at 64 processors There nothing is standing against such tests. Regards, Reinhard.
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.