Author: Kai Lübke
Date: 03:40:16 11/10/97
Go up one level in this thread
On November 07, 1997 at 09:58:25, Bruce Moreland wrote: >There are other reasons to beware of single-machine testing. > >If you test with "10 seconds per move" or something like that, you >should check to see that both programs are actually thinking for ten >seconds per move. A better way would be to do "game in 15 minutes", or >something like that, I think. I've switched to "game in XX minutes" anyway since it allows me to plan the length of a game better (40/2 tends to take 8-10 hours per game sometime). Unfortunately, not all programs support all "game in NN" possibilities; Rebel does not offer anything but 2/5/15/30/60/120 min/game, but not 90 or 180 which I like to use. :-( >Also, you should try to see if one program is impeding the other one. >There should be no difference in node rate of either program when the >other one is also booted. Yes, I always do that. Rebel, MChess, Crafty, Comet, Arasan, The Crazy Bishop, GnuChess, Gromit are all very friendly in that respect. Only Chessmaster hogs a little more, but less than 10% as far as I've seen. --- Shep
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.