Author: Mike Byrne
Date: 17:36:06 09/28/04
Go up one level in this thread
On September 22, 2004 at 02:29:34, Joshua Haglund wrote: >On September 21, 2004 at 22:20:54, Mike Byrne wrote: > >>On September 20, 2004 at 04:12:17, Joshua Haglund wrote: >> >>>Here are my compiled Crafty's. They seem to be faster then Dann Corbit's >>>version. P4 3 Ghz, 1 GB RAM. They are also over 100kb smaller. >>> >>>The SMP version supports up to 8 CPUs. >>> >>>http://myweb.cableone.net/jhaglund/crafty/wcrafty.exe >>> >>>http://myweb.cableone.net/jhaglund/crafty/wcraftysmp.exe >>> >>>Enjoy, >>> >>>Joshua Haglund >>>toneewa@yahoo.com >> >> >>There may be something wrong with your versions - I tested the wcrafty.exe and >>it does "see" the right number of nodes under the bench command. > >What? There's something wrong because it can see the right number of nodes? >What's wrong it? It's staight out of the .zip file. > >Over my head, >:) >Joshua Haglund >toneewa@yahoo.com I left out the word "not". Sorry. I think a later poster nailed the discrpancy in nodes on the head. The "standard' crafty compiled these days is with /DFUTILITY option. It seems to be a little stronger with that option. It performs the bench command with fewer nodes than your compiles. BTW, removing /DFUTILTY does speed things up -- but it also widens the search path and Crafty is burning search nodes that by and large provide little value. So the trade off - faster nps , wider search or slower nps and smarter search. All my unscientfic tests indicate that futility adds some where between 10 to 20 ELO. Not a lot - but every little bit helps.
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.