Author: Sune Fischer
Date: 15:33:21 12/05/01
Go up one level in this thread
On December 05, 2001 at 18:12:27, David Rasmussen wrote: >>It is a somewhat indirect way of testing for collisions, why not use a more >>direct approach? >> > >No, it is a very direct approach. The hashtable is just a means to an end. What >we want is to save time by saving some results for reuse. What I am checking >directly is not collisions the technical hashing theoretic sense, I am checking >whether the hashtable returns the right values for reuse. Incidentally, this is >the same as finding _some_ collisions, not all, but some. Not none, as this >couldn't happen if there wasn't collisions. But it could happen, and apparently it does in your case. >>Besides, if both Bruce and Robert are using 32 bit keys I believe it is sound >>(ie. not producing many collisions). >> > >What a scientific argument :) I didn't say it was ;) >There's a difference between saying "32-bit pawn hashing 'works'" and "32-bit >pawn hashing has no collisions". Well for pawn tables I believe this is the same. Since there are so few different posistions, even one collision could mean a thousand collisions because it is reused so many times. > Maybe the collisions that do happen, doesn't >matter. So it 'works'. But there are collisions. I am only saying one thing: >There are collisions (nobody has disproved this yet, Huh? Neither me or Hyatt can confirm your findings, and since Bruce is also using them I doubt he's getting collisions. This is what I would call disproving. > on the other hand, I have >proved it). If you say so ;) >Do they matter? I would really like to know. > >/David -S.
This page took 0.01 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.