Author: Tony Hedlund
Date: 09:29:32 05/05/99
Go up one level in this thread
On May 04, 1999 at 14:00:39, John Merlino wrote: >On May 04, 1999 at 13:50:51, jose hernandez wrote: > >> >>Why do you test a Strong program like CHM6000 in a P200 mmx vrs a weak program >>in a weak machine ?? Is Obvious that CHM6000 always go to win. >> >>I dont understand. > >This problem has been addressed before, but it has been a little while. You can >only determine a chess program's rating based on games with other machines that >already have an established SSDF rating ON A SPECIFIC PROCESSOR. This last part >is very important. Many SSDF testers do not have two P200 machines, so they test >one program on a P200 and another on a weaker machine. > >The fact that CM6000 is winning most of its games in this case does not >necessarily mean anything in and of itself. The other factors that go into >establishing a chess rating are the rating of the opponent AND, most >importantly, HOW OFTEN they beat their opponents. > >Obviously, CM6000 on a P200 is going to beat MCP6 on a P90 in most of the games. >The important question is: How Often? These factors all go into the equation >that determines a chess program's rating. > >If you look at the latest SSDF list, you can see that CM6000 has played >generally weaker opponents than Fritz 5.32 (which CM is currently tied with). >Fritz's opponents have averaged almost 100 points higher. BUT, CM has a much >higher winning percentage (78% to 66% for Fritz). Therefore, their ratings are >the same. > >jm A wonderful answer. Thank you jm! Tony
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.