Author: Sune Fischer
Date: 14:18:44 12/05/01
Go up one level in this thread
On December 05, 2001 at 17:04:36, David Rasmussen wrote: >On December 05, 2001 at 16:41:22, Sune Fischer wrote: > >>>>> >>>>>We need Bob or some other third party to make tests too I think. >>>> >>>>I have already done this. In fact, anyone that _doesn't_ do this is inviting >>>>great grief. It is essential that the hashed information match _exactly_ the >>>>computed information for every position. And I do this as part of my debugging >>>>from time to time, which will often catch a bug where I forget to stick >>>>something in the pawn hash record, but I use it later anyway. >>> >>>I agree, I am doing the same thing with Chezzz. Among a _lot_ of other debugging >>>checks as well. Chezzz has roughly the same amount of collisions as Crafty, if I >>>use 32-bit pawn hashkeys instead of 64-bit ones. >>> >>>/David >> >>I agree, but just because the hashed information doesn't match exactly the >>computed information doesn't mean you have a collision, it _could_ be something >>else. You are debugging a lot of stuff by doing this, any number of things could >>be the cause. >> >>Collisions are much easier to detect directly using the 64 bit key for a >>reference. >> > >Not at all. I am _not_ debugging a lot of stuff by doing this. What "something >else" would that be? I'm afraid you don't understand how it works then. If it >reassures you, Bob does the same thing to check for collisions. We don't find >_all_ collisions, as some positions just might happen to have the same score and >properties in Crafty as some other position that has the same hashkey. But when >we find a position that has the same pawn hashkey as some other position (as I >have done), there is by definition a collision. > >/David I got the impression he was using it for general debugging. It is a somewhat indirect way of testing for collisions, why not use a more direct approach? Besides, if both Bruce and Robert are using 32 bit keys I believe it is sound (ie. not producing many collisions). -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.