Author: Slater Wold
Date: 11:31:05 07/09/03
Go up one level in this thread
On July 09, 2003 at 13:22:55, Russell Reagan wrote: >On July 09, 2003 at 12:41:15, Slater Wold wrote: > >>How many games were decided stricly by the books used, and not the engine? > >This is a good question. I wonder what is best, book or no book? I think it >depends on what we're trying to test. For instance, I'm sure the Chessmaster >tests included different books. My current tests did use the default book from >Hyatt's ftp. Maybe after this current one gets to a certain number of games (100 >per engine?) then I'll try it with no books and see how that goes. > >This makes me think of other kinds of tests. Maybe Crafties with no book vs. >Crafties with book. I bet if you took all combo's (no book, nunn, ftp book) and ran 100 games of each test between 4 different crafty's, added them all up, and looked, you would have no more than a 30 ELO difference. That's what I would *expect* anyway. >>(Taking for granted that >>this was run on a non-SMP machine.) > >It is a dual. Ack! Just make sure that each process has the affinty set to a certain processor. Like crafty0.exe is set to use CPU 0 & crafty1.exe is set to use CPU 1. And of course, make sure to use 'mt 1' (I am positive you are). Any kind of SMP/SMT branching between 2+ CPUs can cause some weird things (from my own personal experience with my SMP systems).
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.