Author: Uri Blass
Date: 11:43:36 06/05/01
Go up one level in this thread
On June 05, 2001 at 14:03:03, Rajen Gupta wrote: >hi: i've been running a match between the newest programmes and the older ones, >with both programmes using their own copy of the opening books, to find out how >much better(if at all)the newer programmes are as compared to top names from >yesteryear > >hardware single celron 900 with 256 mb ram(64 hash for each programmes) > >engine vs engine match (all chessbase programmes)of at least 50 games each > >book= each engine gets the default copy of the chessbase tiger opening book with >learning on. > >time 60 moves in 90 minutes. > >current match chesstiger 14 vs junior 5 (other contestants=gambit tiger2; >fritz5.32; f6light) > >score tiger won=8 lost=7 draw =6 > >as i am currently working away from home i can't post the games till the next >weekend.based on the current performance i see no difference between the tiger >of 2001 and junior5 of 1998. The result suggests 19 elo difference for tiger but the statistical error is clearly bigger than it. It is clearly a mistake to get conclusions based on only 21 games. You need hundreds of games to prove difference that is probably between 50 and 100 elo. Uri
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.