Author: J. Wesley Cleveland
Date: 10:19:27 12/04/01
Go up one level in this thread
On December 03, 2001 at 16:29:52, David Rasmussen wrote: >On December 03, 2001 at 15:11:55, Sune Fischer wrote: > >they >>>aren't important, or that they are not happening. I feel safer with them not >>>happening, and they won't with 64-bit pawn hashkeys. Even if they do, I handle >>>that just as graceful as all other programs: I don't even detect it. >> >>How would you detect them anyway? >>All the information you have on the position that generated the old key is _in_ >>the key already. >> > >It's no problem to detect them, you just can't use the pawn hashtable at the >same time :) In a release version, where the pawn hashtable is implemented and >used for it's speed gains, you can't detect them. But in a debug version, you >can easily detect them, but you won't actually use the table values. One simple way to both detect and use them is to use 64 bit keys and report collisions on (say) the lower 32 bits.
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.