Author: David Rasmussen
Date: 06:44:44 12/04/01
Go up one level in this thread
On December 04, 2001 at 09:10:10, Severi Salminen wrote: >>Sure, as I said elsewhere, it is a design choice. I personally would give a >>great deal for correctness, if it can be achieved. That this "problem" is also >>in Crafty implies that it is not too big of a problem. But still, who knows what >>impact it has. One could put some debug code into Crafty to check when a >>collision had a big impact, that is, on the PV, or ultimately, the first move of >>the PV. > >Yep, it is a design choice and the same applies to null moves, futility pruning >and allmost all techniques used: we'll be faster but maybe less accurate. And >usually being less accurate and fast is better than slow and accurate - that's >why I favor 32-bit pawn hashing. I just wish we had 64-bit processors availabe. >Chess programs will benefit a lot from it. > >Severi That's true, but in the case of null moves and other "inaccurate" techniques, there have been extensive research to show the (limited) effects of these on accuracy. That is, null move is incorrect, but not so incorrect that it can't be used. No one knew that before it was verified by several people. And I haven't seen any arguments or research that "proves" that 32-bit pawn hash keys are "safe". /David
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.