Author: Koundinya Veluri
Date: 20:56:27 09/08/03
Go up one level in this thread
On September 08, 2003 at 23:21:53, Russell Reagan wrote: >On September 08, 2003 at 20:30:33, Dieter Buerssner wrote: > >>On September 08, 2003 at 06:51:55, scott farrell wrote: >> >>>It [fail soft alpha-beta] really only helps with the hashtable with regards to >transpositions. >> >>It can also help in the case of a fail high in zero window search, when the >>original window is not a zero window. With fail hard, you will always get back >>alpha+1, and you will need to research. With fail soft, you can get back a value >>>= (original - not zero window) beta, and in that case, you can save the >>research. >> >>It can also help aspiration search. In fail hard with a fail high, you will >>always get back beta. For an engine, that gradually adjustes beta (say oldbeta + >>1 pawn first), it may save one (or more) adjustment cycle(s). Fail soft might >>return beta+5 pawns, making the gradual adjustment to oldbeta+1 futile. > > >These all sound like good reasons to use fail soft. What are the drawbacks to >fail soft, and the reasons to use fail hard? The only disadvantage I see with fail-soft in King of Kings is that the search is a bit less stable. I have instability with fail-hard too though, so I don't see it as a good reason to use fail-hard. Koundinya
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.