Author: Robert Hyatt
Date: 14:04:06 03/03/06
Go up one level in this thread
On March 03, 2006 at 16:22:41, Tord Romstad wrote: >On March 03, 2006 at 16:08:50, Tony Werten wrote: > >>Our thoughts are probably just a reflex reaction based on the bad experience >>from nonlocking hashtable sharing. > >In fact, I don't lock the hash table either. :-) > >I agree that I *should* lock the hash table, or use Bob's lockless hashing >trick. >So far, I have just been too lazy to implement it. I haven't noticed any bad >effects of non-locking hash table sharing so far, but of course this doesn't >mean that it hasn't happened. > >Tord based on the hash testing anthony C and I did, it isn't an issue, unless you depend on the hash move to be perfectly legal (as I do). Errors caused by interleaved updating can cause the wrong best move to be attached to an entry, and if your program can be wrecked by a bogus move from the hash table, the lockless hash trick is a life-saver. If you can handle illegal hash moves with no adverse affect, you can certainly ignore the very rare hash error you get as it is just not going to affect the final outcome at all...
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.