Author: Albert Silver
Date: 12:23:18 12/19/97
Go up one level in this thread
Relax Thorsten, I'm not attacking your integrity here, nor am I supporting Moritz's conspiracy theory. All I said was that I can't reproduce your results, which is true, and was wondering if your results might not be due to a change in the setup. I also have Fritz 5, but when i ran the moves through it to the depth you played I got different results. As a matter of fact, the moves on the 37th, 38th, and 39th moves always showed up but as Fritz's second best choice. If you could send me the times at least of CSTAL (which I don't have) I'll see what I get and we can compare. As for the difference in hash tables, again I don't understand why someone claimed the hash table size could change the eval, first because this seems to be completely illogical, and second because I get consistent results in evals no matter what the hash table size. The only difference I can see is the time spent to reach a certain depth. What I believe may be the culprit for this claim is that whether Fritz has just completed 11 plies (for example) of calculations or has actually begun exploring the 12th ply, it will post an eval of 12 plies next to the played move. Albert On December 19, 1997 at 14:58:22, Thorsten Czub wrote: >>As I mentioned in a post yesterday, I have a K6/233 (so if it's a matter >>of processor type then I only have an edge in speed) but seem incapable >>of reproducing Thorsten's results. I ran the 37th, then the 38th, and > >I am using a k6/200 Mhz and used 9182 Kbyte HASH and commercial fritz5 >engine. >Time control was 40 in 120. >I saved the game with complete evals, search-depths and computing-times >and could send the compressed database to anybody who does not believe >my results. >My other 10 chess programs like the speed of my k6/200 Mhz. > > > >>finally the 39th moves where he screams murder and at no time does Fritz >>5 choose those moves. They ARE it's second choice, but most certainly >>not it's first. > >I could try if I can myself reproduce the game. It could be that Fritz5 >plays only the same garbage if you replay the complete game from >beginning. >You cannot try to replay it from the "strange" moves because the >problems started maybe somewhere else. > > > >> In all 3 instances it in fact chose Kf7. I played around >>with the hash tables (as someone mentioned that the eval changed >>according to the amount of hash tables) but acquired very consistent >>results (with the exception of how long it took to find them). Could >>someone tell me what's up? > >They say that less hash causes whatever. I think that is a very nice >excuse ! >If a chess programs needs at least 50 MB HASH to play chess... :-) > >> >>Albert
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.