Author: Dann Corbit
Date: 12:13:13 05/05/00
Go up one level in this thread
On May 05, 2000 at 14:49:07, Peter Kappler wrote: >On May 05, 2000 at 13:54:39, Dann Corbit wrote: >>There is no real preference for any of the 4 possible choices. If your program >>chooses any of them, it is doing well. Less than 1/2 pawn separates all of the >>choices which is "well within experimental error." >> ><snip> >> >>From crafty's opinion, they would rank Bh6, Rxe6, Qd3, Nxe6 in order of >>preference, but it would be silly to tag any of them 'wrong' at this point. >> >>My opinion: >>It's not a very good test position. There may be additional moves that are just >>as good, since it is a very quiet position. I think that test positions should >>be decisive. > >Dann, > >How long did you let Crafty search each position? Several hours on a 500 MHz PIII. >Also, I have some questions about how you post search results: > >What are acs, acd, acn? acs is average count in seconds. acd is (non-standard) depth in plies. acn is average count of nodes. The PGN standard by Steven J. Edwards explains the meanings in detail (except for acd which is a very useful non-standard extension). >I assume ce is centipawn-evaluation, and I assume the large number at the >beginning is # of nodes searched, though I don't understand the reason for the >negative sign. The negative sign is due to the count of nodes going beyond what a 4 byte integer can hold (the number wrapped back around due to integer overflow).
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.