Author: Anson T J
Date: 10:38:12 08/24/03
Go up one level in this thread
On August 24, 2003 at 11:59:34, ALI MIRAFZALI wrote: >On August 24, 2003 at 11:45:18, Anson T J wrote: > >>On August 24, 2003 at 11:28:29, ALI MIRAFZALI wrote: >> >>>On August 24, 2003 at 10:38:31, Anson T J wrote: >>> >>>>All PGN games are available for download and view on the following website: >>>>http://mysite.freeserve.com/intagrand/junior8test/index.html >>>> >>>> >>>>Test Configuration >>>>Hardware: Dual AMD MP 2400+ 512 MB Ram >>>>Software: Junior 8 GUI & 3-5 piece EGTB >>>> >>>>Engine Details >>>>Processor: 1 CPU each (AMD MP2400+) >>>>Hash: 128 MB hash each >>>>Ponder Move: On >>>>Books: Native Books / CM9k (F6's general.ctg) >>>>Time Control: 40/2Hrs + 20/1Hr + Rest/30 mins >>>>Rounds: 50 games vs each opponent >>>> >>>> >>>>Opponent: Junior 8's Score: >>>>Shredder 7.04 +18/-12/=20 (56%) +41 elo >>>>Fritz 8 +17/-11/=22 (56%) +41 elo >>>>Hiarcs 8 +28/-8/=14 (70%) +147 elo >>>>CM 9000 (SKR) +17/-10/=23 (57%) +48 elo >>>>CM 9000 +23/-8/=19 (65%) +107 elo >>>> >>>>This concludes my Junior 8 Summer Test. I hope people will find the games and >>>>results useful. Sorry that I will not be able to run any more test games for a >>>>while as I can no longer spare the CPU time. A fine performance by Junior 8 >>>>here. Also congrats to the "SKR creators" as the SKR personality of Chess Master >>>>9000 performed 59 elo points stronger than the default settings. >>>There is a big problem with this match: You did not use the Chessmaster opening >>>book for CM .This is like separating the head from body.Opening books are indeed >>>part of the program.Also I should note that (and I have said this many >>>times) that comp vs comp matches prove nothing about how a program will do >>>against humans:That is Program A may be stronger than program B by 100 Elo in >>>Comp vs Comp ;but Program B may be stronger by 100 Elo than program A when >>>playing only humans.Both Fritz and Junior were disasters in the matches against >>>Kasparov and Kramnik. In any case I do not see the points in these >>>Comp Matches(We need more human vs Comp) as SSDF is doing a great job pitting >>>programs against each other. >> >>a) No CM book exists for The King >>b) I didn't mention human games, nor am I concerned with human games. If you >>want human games, theres not much point looking in an engine match is there? >>c) SSDF hardware is much weaker than what I used and they don't test CM either. >I donot understand part a) CM9000 does have its own opening book(ie the default >personality)I just checked my CD the book is called CM9000. CM9000 doesn't have auto play features so I used "The King 3.23" which is the CM 9000 engine except it runs under the ChessBase GUI (which allows auto play). In the Chessbase interface Chessmaster doesn't have a book. As Kurt was one of the creators of the SKR personality, I took his advice to use the fritz 6 general.ctg. When testing CM9000 default, I used the same book so a comparison could be drawn between the personalities.
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.