Author: blass uri
Date: 16:21:16 09/21/99
Go up one level in this thread
On September 21, 1999 at 13:47:43, Bradley S., Short wrote: >Eric, > >Genius 2 (without hashtables) can mate in 25 moves or less with only 5 seconds >to think per move. It seems to make no difference what the starting position >is. It always gets it done fast. >People are putting too much stock in these later versions of chess programs. >Every time a new one comes out people test them on faster and faster hardware. >I'm certain that all most no progress has been made in the algorithms. Only new >features are added and the producers are counting on hardware to make them play >better chess. >If an older version of Genius like 2.0 were played on equal hardware with any of >the new programs in a long series I'll bet it will still find itself in the top >four. >I've also been playing around with Chessmaster 4000 against 6000 and it seems no >weaker at all. Last night it drew the game. CM6000 ran on a PIII 550 and >CM4000 on a Celeron 366. Did you use the I am the next bobby fisher in chessmaster6000? Your results make me suspect that you did not do it because I read that CM6000 is strong if you tell it that you are the next bobby fisher. >It looks likely that about 90 percent of the improvement in >new programs is faster hardware and maybe 10 percent is better algorithms. Genius3 was the best program in 1994 and there was no much progress in the best program in 1994-1997 because richard lang did not work on his program when other programmers that got better were weaker. This is not the case in 1997-1999 because Genius stopped to be the best program and other programmers made progress. The fact that Genius can mate with knight and bishop when CM6000 has probelms with this position proves nothing about playing strength because this position is rare in practical games 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.