Author: Robert Hyatt
Date: 15:16:53 02/07/01
Go up one level in this thread
On February 07, 2001 at 16:22:12, Severi Salminen wrote: >Hi! > >Thanks for detailed explanation. > >>1. You might start the next iteration without a PV to guide the search. >>Internal iterative deepening helps. > >Oh, a new term again: internal iterative deepening. I think this was the thing I >wondered in Crafty's search.c. Do you use IID in other places than when you are >in PV node and don't have a move to try? Has anyone ever tried to use IID in >_every_ node? If it helps in the root why wouldn't it help in other nodes as >well... I use it at _any_ PV node (node where beta != alpha+1) that does not have a hash move to search first. If you use it at non-pv nodes, you will burn way too much CPU time. > >>I don't do this in Crafty for lots of reasons. Null-move + hashing can cause >>false fail-highs on a null-window search, and if you automatically assume that >>move is better, you will make horrible mistakes every now and then. In crafty, >>if the null-window search fails high and the re-search immediately fails low, >>I pretend the fail-high never happened. > >I also think I'll skip the lazy alpha bounding and stick with the plain alpha >bounding I seem to be using. I want to keep things as simple as possible. > >Severi
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.