Author: Ron Langeveld
Date: 15:40:59 06/25/02
Go up one level in this thread
On June 25, 2002 at 16:48:08, George Sobala wrote: >On June 25, 2002 at 15:08:08, Ron Langeveld wrote: > > >>I agree. The available tournament book is a major embarrassment, as I have >>stated before on this podium. I have generated my own book for Hiarcs 8 with >>2400+ games and imported a selection of recent 2550+ games to establish the >>tournament moves. This book is still being tested, but after more than 2000 >>games Hiarcs is allmost on par with Fritz 7 and Frit7.ctg (trailing less than 30 elo points). > >To ilustrate this "opening book effect": I am in the middle of a 12-round round >robin between Hiarcs8, Fritz7, Shredder Paderborn and CMFun (CM8000 variant), >30/0 on a K6/450 but I am using Nunn2 opening positions (alternate colours) >rather than opening books (starting from number 10). In other words in round one >each program plays every other program from the same starting position, then in >round 2 the same positions with colours reversed. Round 3 moves on to a new >position etc etc. > >Current Score: > >Hiarcs8 16/23 >CMFun 12/23 >Shredder6 9/23 >Fritz7 9/23 > >* Hiarcs8 6.5-1.5 Fritz7 in their mini-match. This is no surprise to me :) > >* Fritz7 is demolishing Shredder but is losing badly to CMFun. It is my believe that one of Chessbase's major strengths with regard to releasing new engines/books is the thourough testing. They have the resources to give a new engine a crashproof testdrive of engine matches to track any glitches in the engine/book combination. I am talking about more than a handful of engine matches running simultaneously on several PC's starting at least 2 months before the release of the engine. Add a considerable effort and resources available to make the necessary adjustments to engine and book, et voila. As far as Shredder 6 is concerned: currently CB's main priority is Fritz, whether it is 7 or 9 or 10. Stephan MK probably hasn't received an equal treatment, because I also get the distinct impression that Fritz 7 performs disproportionately well against S6. Remember that the actual release date does not reflect the fact that they were able to test S6 long before F7 got out. > >I am astonished by Fritz7's (7.0.0.8) poor performance but can't identify any >fault in the settings. Indeed, CMFun is at a disadvantage because the others are >using 3/4/some 5 tablebases. CM8000 probably is not thrown in the prerelease test-pit of Fritz 7, which could explain this result. CM8000 may be fine, but I doubt whether it is able to keep up this performance over time.
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.