Author: Melvin S. Schwartz
Date: 14:32:24 07/16/99
Go up one level in this thread
On July 15, 1999 at 21:04:39, Chuck wrote: >On July 15, 1999 at 20:52:37, Melvin S. Schwartz wrote: > >[snip] > >> >>According to the formula for Fritz, the amount of megs for hash tables should be >>144 at 40/2 and on a 400MHz processor. This formula is stated in the manual for >>Fritz, and I would say that 64 megs is quite insufficient at your time control >>and processor speed. >> >>Regards, >>Mel > >So what? It is not the publics' fault that Fritz was programmed with such >specific hardware limitations (that it's search is so affected by less than >gargantuan hash tables). IMO, to fairly test 2 chess engines they should be >running on equal hardware, i.e., uniform platforms. And ____________________________________________ I also like to see testing on equal hardware. However, I believe that such fairness should extend to whatever the program needs. If it needs large hash tables and you give it an insufficient amount against another program with a proper setting, then you are not using the equal hardware, etc. principle to all conditions. This is in my opinion - very contradictory. Mel _________________________________________ the hardware should be >relatively top-notch but commercially available and reasonably affordable. It >should NOT cater to the specific desires of 1 programmer so he can tweak his >engine for the match. This is exactly why Fritz is not the best choice for most >chess program purchasers, because it slows way down when it runs out of >megabytes of memory. > >Chuck
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.