Author: Robert Hyatt
Date: 11:24:43 09/06/00
Go up one level in this thread
On September 06, 2000 at 13:42:26, Larry Griffiths wrote: >On September 06, 2000 at 13:38:27, Robert Hyatt wrote: > >>On September 06, 2000 at 11:06:42, Philipp Claßen wrote: >> >>>Hi! >>> >>>I want to add a extra transposition table for quiescence search to my own >>>program. But I´m not sure how to avoid that important entries are overwritten by >>>less important ones. In a normal search the distance is a good criterion but in >>>a quiescence search? Perhaps the current ply? >>> >>>Phil >> >> >>I don't store q-search nodes, although a couple of years ago I did. I don't >>think it will make any difference in all in performance. The cost of the >>probes is offset by the value of the hit, almost exactly. So doing it or >>not won't make you any faster. And since not doing it didn't hurt at all, >>I got rid of it for two reason. 1. It made the q-search code a bit simpler, >>and simpler is always better. 2. For big searches with small hash tables, >>it makes much more effective use of available memory since the q-search stuff >>is the majority of the nodes searched, and they don't compete for table space >>at all. >> >>If you want to store them, just store "depth" but let it go negative into the >>q-search. the first ply of q-search would be 0, the next would be -1, etc. Or >>you can do as I did and always store 0 which is reasonable. > >I have a question Bob. > >Should all moves be generated in a q-search, or just captures? > >Larry. Depends on your q-search. In crafty, captures are _all_ I look at, and I don't pay any attention to checks or other kinds of moves. In Cray Blitz, I did all moves as the selective code could include things besides captures if it wanted to.
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.