Author: blass uri
Date: 00:51:42 08/04/00
Go up one level in this thread
On August 04, 2000 at 02:32:12, Terry Ripple wrote: >Using an AMD K6-2, 266Mhz, Hiarcs,32mb hash and Fritz6,4mb hash > >I want to say that Fritz over-stepped the time in "only" 1 game from this 62 >game score, so this is now a total of 10 out of 200 games that Fritz lost on >time which is down to 5% of the total games lost on time using 4mb hash. > >In my past match of 306 games Fritz lost on time 3.92%(12 games)using 16mb hash >but i still have 106 games to complete to make a better comparison with this >present match. I was told that using 16mb hash for Fritz6 old was the reason for >it losing alot of its games on time in my last match, so i lowered it to the >present 4mb. > >Any observations and comments are welcome! > >Best regards, >Terry I guess that I was wrong about fritz6light. It is the case for previous engines but probably not for fritz6light. I did some games 1 minute per game with Fritz6light(32 mbytes hash tables) and I found that in most of the games it did not lose on time. Some results in 1 minute per game: Fritz5.32(2 mnytes)-Fritz6light(2 mbytes) 7.5-2.5(both sides used 2 mbytes) Fritz5.32(2 Mbytes)-Fritz6light(32 mbytes) 6.5-3.5 Fritz6Light(32 Mbytes)-Fritz5.32(2 Mbytes) 5.5-4.5 The results are not significant but they suggest that Fritz5,32 does not like big hash tables and that Fritz6Light has the same rule as Hiarcs7.32 (as much as possible when the maximum possibel hash for Fritz6Light is 32 Mbytes). I do not know if Fritz6light has the rule as much as possible and it is also possible that hash tables above something are the same for it(the difference between 6.5-3.5 and 7.5-2.5 is very small) Uri
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.