Author: Mark Young
Date: 14:01:11 05/31/03
Go up one level in this thread
On May 31, 2003 at 10:31:21, Uri Blass wrote: >On May 31, 2003 at 10:16:21, Jorge Pichard wrote: > >>On May 31, 2003 at 04:13:54, Sandro Necchi wrote: >> >>>I am glad that SSDF decited to test Shredder 7.04 under UCI gui. >>> >>>The reason is that this program may be the first one to cross the 1800 Elo >>>barrier on the SSDF list. >>> >>>Some time ago I stated that 7.04 was about 20 points stronger that 7.0, but this >>>was based looking some game of a previous version. In reality, and testing made >>>by many chess funs, gives a little more: >>>betwen 22 to 30 points. >>> >>>So, considering that the UCI gui with tournament book selection should give at >>>least 15 points more, it possible to get a rating on a K7 1200 of 2805 to 2810 >>>Elo points. >>> >>>We'll see if these considerations will be confirmed. >>> >>>About the SSDF not willing to test CM9000 I think people that do not know SSDF >>>long enough like me, are saing unfair things on them. >>> >>>First of all they do this as a hobby and because they love computer chess. To >>>claim that their are sponsored by ChessBase is so “non sense” and does not need >>>any reply. >>> >>>What a programmer should do to have its program tested is to send them 3 copies >>>of the program with clear instructions for settings and how to run it. >>>Also that program MUST include the autoplayer as it would take TOO MANY human >>>resources to test it in the normal way, like when the autoplayer was not >>>available. >>> >>>The point is that if they have TO USE ALL THEIR TESTERS to test a program in >>>the “old” way, this would probably not be acceptable to them as other >>>programmers would complain. >>> >>>I do not know how many testers they have now because the latest contacts I had >>>with them were about 10 year ago. >>> >>>I personally DO NOT and WILL NOT BUY or run ANY CHESS PROGRAM, even if 4000 Elo >>>strong if does not have the autoplayer. >> >>I agree with you, I bought all the previous version of CM from CM2100 back in >>1989 to CM 8000. But since they never heard our request to add an autoplayer I >>decided to not buy any more CM programs. >>> >>>I also believe that using a different book for running a program is no sense as >>>the opening book is part of the program. >>> >>>It is like to drive a Ferrari with another engine. It is not the same and to me >>>means nothing. >> >>I have to congratulate you for making such an excellent Opening book which of >>course is part of the program. Just to give you an idea of how much your Opening >>book constribute to the high SSDF Elo of Shredder being on first place, I >>decided to test Shredder 7 vs Friz 8.008 both without opening Books and Fritz >>after 12 games at G/60 is leading by two points so far W6 D2 L4. >> >>Jorge > >I do not care about shredder7 because 7.04 is the best version and a result of >7-5 in 12 games are too small to get conclusions. > >Uri The conclusions that Shredder is best because of books or positional learning is without merit. I always test without the programs book, I always test my programs with a common book, same hash, and tablebase. Here are my results of a current test being run as we speak. Game 60' , Shredder 7.04 Vs. Fritz 8.0.0.8. Chess Tiger 14 book was used by both programs. 256MB of hash and all 5 stone and below tablebases. P4 2.53 Ghz The match will be to 100 games. Shredder has never been behind in the match with over 40 games played. 1234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890 1 Shredder 7.04 ½½1½11011001110100½½1½½101½½1½11½0½½½½½½½00????????????????????????????????????????????????????????? 24.0/43 2 Fritz 8 ½½0½00100110001011½½0½½010½½0½00½1½½½½½½½11????????????????????????????????????????????????????????? 19.0/43
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.