Author: Albert Silver
Date: 17:02:53 12/18/97
Go up one level in this thread
On December 18, 1997 at 15:41:07, Thorsten Czub wrote: >>Thorsten ***knows*** that Fritz will get an hash table overflow with his >>settings in about 10-15 seconds. It will then get 10 fold hash >>collisions at tournament time controls and he can comment on "misguided >>search and evaluations". > >Is this my problem. I am not the programmer. I am a customer ! >I bought it. Thats enough. I don't want to study physics to get a strong >chess program. > >>Of course he's running Fritz on an AMD K6 >>processor which is the worst available environment for Fritz you can >>imagine (Pentium MMX should be about 75 percent faster for Fritz). > >Bullshit ! And also not my problem. If they tune on Intel CPU, is that a >problem of a customer who buys a software of the year 1997 ? >Maybe you should tell the people WHEN you call it a software of the year >1997 >not to buy 200,- DM Fritz5 but also 200 ,- DM for RAM and another 500,- >DM to get an INTEL instead of AMD. >The AMD CPU works perfectly well for computerchess as all other >benchmarking with all my other programs has shown. I have no fault when >Frans tunes assembler on INTEL chip ! Is it my problem when ChessBase >AND Morsch gave their best to castrate their SOFTWARE of the YEAR 1997 >that much ?? >And - how do you know ? >In Paris Fritz5 got a Pentium300, an optimized book, enough RAM and >Frans was sitting next to it, and the results of Fritz5 AND the games >were bad. >So: Fritz works maybe at your home. Brilliant. But what will the other >million customers in the world think about this SOFTWARE OF THE YEAR >when they cannot reproduce the advertising results of the ChessBase >home-page ?? > > > >> >>Fritz plays only some of the commented dubious moves from Thorsten's >>game on my P233MMX with 100MB hash tables, in each case EVAL is >>different at same search depths. > >And ? ! I am sure on my 486 Fritz5 would play also a different move. >What can we learn from this ? >Nothing. > > >> >>Thorsten is on a personal crusade. > >Why ? I am just studying a software you want to give the title: >Software of the year. >> >>Whenever he attacks "fast and stupid" programs, he will pick Fritz. > >Oh - thats not my idea. But this is the program you want to give a >prize. WHY? >The customer has to buy an expensive Intel, amazing big Hash, what do >you need in addition ? Full moon ? Candle-light in the windows ? > >>Whenever he attacks Fritz, he means Chessbase. > >I am attacking a program for the moves it plays. >Why should I attack chessbase. They have done nothing to me. > > >>When he attacks Chessbase he's angry because of one person that also >>plays a certain role in CSS magazine. > > >Why should I be angry about Steinwender or Friedel. They have not >invented Fritz5. >I am talking about the product KK wants to call product of the year >1997. > >> >>So he really wants to hurt his old enemy at Chessbase and fight the >>shadows of the past. > >Bullshit. How can you talk such a bullshit ? >Friedel nor Steinwender are not my enemies. They have a very weak >magazine. Thats all. Their decreasing level of quality is NOT MY >PROBLEM. > >I don't see any shadows. Concentrate on the facts instead of searching >for fog. > > >>To me, this "cascaded attack" (you might call it an "xray" attack) is >>not acceptable. > >It is not an attack. It is just a game of chess. And the fact that you >think it is MY fault as a customer that I have bought your product of >the year shows me that I am right. A product of the year that fails at >the customers home... > >>If he wants to fight Friedel, he may well do so. > >Why should I fight Friedel ? He was always very nice to me. I met him in >Paris too and we had some nice chat. I took some nice photographs of him >and Dieter was also there and we had some nice chats. Why should I be >angry towards them ? >The atmosphere in Paris was very cool. No attacks. Very pleasant nice >atmosphere between the participants. I had no problems with Friedel, not >with Levy, not with Marsland, not with van den Herik. But - of course >you know it better Moritz, because you were not there. It is always >easier to explain when somebody is dreaming without having any data like >you ! :-) > >>But >>then not pretend that it's fast chess engines in general or Fritz or >>Chessbase in particular. I appreciate constructive comments from >>Thorsten about almost every other computer chess matter, but when it >>comes to Fritz, his statements are somehow from another galaxy. > >Bullshit. You don't want to register that you PROGRAM of the year can be >the disaster of the year for a normal german customer who feels betrayed >after he had bought it. I bought it ! I did not review a beta-copy. I >bought it. >And it fails. So what ? What has this to do with Friedel ? >Maybe you should check if your results are "from another galaxy". > >>Moritz I realize that this may seem a moot point, but I have a K6/233 and I seem to be absolutely incapable of reproducing Thorsten's results. I have tried various combinations of hash tables (my machine only has 24Megs) all of which produce CONSISTENT results (so I honestly don't understand why the amount of hash tables would influence the eval) and Fritz here never plays the alleged moves. In fact the program seem to desperately want to play Kf7 on the 37th, 38th, and the 39th moves (it won't even play 39...e3+). the only thing I have noticed about the eval at a depth of say 12 plies is that Fritz does not make the difference in whether it has begun exploring the 12th ply and when it has only completed 11 and has yet to give an eval of the 12th. In both cases playing the move will have Fritz post a depth of 12 plies). Albert Silver
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.