Author: Bruce Moreland
Date: 23:26:49 02/04/99
Go up one level in this thread
On February 05, 1999 at 00:27:05, James Robertson wrote: >You must be waving magic dust on your program. Here are some results from my >program on the 11 positions in the nolot test. The null move doesn't seem to >have tremendous advantage, although in several cases it was substantially >better. Thrice the null move took more time than without. > >I ran each position for one minute on my P233, and recorded the times for the >last completed ply. No hash tables, but pawn hash was 1MB. I looked at your times and you have to be doing something totally wrong. You get into a node, you check the hash and cut off if necessary, and if you aren't in check you do a search that's two plies shallower than you'd ordinarily go here (go into quiescence if appropriate), and if you come back >= beta, you cut off. This should be a very common case. If you don't cut off, you just generate moves and search them. Works like a charm right away. bruce
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.