Author: blass uri
Date: 00:41:10 02/17/99
Go up one level in this thread
On February 16, 1999 at 16:57:03, Bertil Eklund wrote: >On February 16, 1999 at 15:35:45, blass uri wrote: > >> >>On February 15, 1999 at 10:58:31, Dan Kiski wrote: >> >>>When it becomes available in the next few days, anyone have any thoughts on how >>>much stronger it will be than 10B. And if it will be as good as Fritz or Hiarcs. >> >>I think that the ssdf rating is misleading because Fritz is optimized for >>P200MMX. >> >>I also do not trust the autoplayer because Ed found that Rebel10 is weaker if it >>uses the autoplayer because of a bug in the autoplayer. >> >>This is the reason that he delayed releasing the autoplayer for Rebel10. >> >>I wonder if this is also a problem of other programs in part of their games. >> >>I think there are some candidates to be the best on relevant hardware and I have >>no idea which one is the best. >> >>I trust only the CM6000 results of the ssdf guys because the games are played >>manually. >> >>I cannot trust other results if the testers do not test the games to see if the >>moves can be reproduced manually. >> >>Uri >Hallo! > >I have played several thousands of games with Rebel7/8/9 and never seen anything >suspicious at all. I have also played a lot of games manually and it is easy to >reproduce every single move! I am happy to hear that you are playing games manually to check that there is no problem with the autoplayer. As you can read on the Rebel-board Moritz Berger >explains what is eventually wrong with Rebel10. You can also see if you use >bigger hash-tables then 28mb rebel plays slower and often at the same ply-level >play another move! This is valid for P200MMX. The manual or the program should >tell the user the size of the hash-table suitable for the choosen time-control >and machine. > >I don´t understand in which way the SSDF-rating is misleading as it is clearly >told what machine and speed that had been used. When we choosed the P200MMX it >was more or less the only realistic alternative. The point is that if fritz5 earns 75% in speed by using P400 relative to P200MMX and Junior5 earns 100% or 125% in speed by using P400 relative to P200MMX then people should know about it. You can ask volunteers to test the speed of different programs in different computers. It is an easier task then playing games between computers. 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.