Author: Graham Banks
Date: 12:17:07 04/20/05
Go up one level in this thread
On April 20, 2005 at 14:56:06, Ray Banks wrote: >Some of you will remember that I posted the results of my personality testing a >couple of weeks ago, but I had not tested the default setting. I've now run the >tests with the default to provide a baseline. And it is very interesting. All >personalities are much stronger than the default, and the very best ones are >STAGGERINGLY better > >My testing was as follows: > >Deep Fritz 8 GUI >40 moves in 40 min repeating, 256MB Hash >Dual Athlon MP2800+ with ponder on >3-4-5 men EGTB >All engines using 2004 Powerbook >no learning > >Each CM9000 personality played 40 games against each of: >Deep Shredder 8 >Deep Fritz 8 >Deep Junior 8 >Chess Tiger 15 >Hiarcs 9 >Ruffian 2.1.0 >Gandalf 6.0 > >That is 280 games for each personality, and 2,520 games in total. The aggregate >scores out of 280, and rankings, were: > >144.0 - Slayer >141.5 - Assassin >139.0 - R1 >139.0 - Slayer 2b >136.5 - T05 >136.0 - Judge II >135.0 - Minotaur 2 >130.0 - Alliance >107.5 - Default > >Default's total score of 107.5 compares to 144 for Slayer. That is a massive 34% >improvement !! > >Looking at it another way, 107.5/280 = 38% and 144/280 = 51%, so a 13% >improvement in the ratio. Staggering. > >Slayer will now be carried across to my CM10th testing, which is on a similar >basis to this. CM10th default will also be tested to provide a baseline in those >tests.
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.