Author: blass uri
Date: 13:20:12 08/03/00
Go up one level in this thread
On August 03, 2000 at 15:21:14, Terry Ripple wrote: > Using an AMD K6-2, 266Mhz, Hiarcs7.32 = 32mb hash, Fritz6 Old = 4mb hash > >An interesting note: In my last match of 306 games there was 12 games that were >overstepped by Fritz6 old, thats 3.92%. I was told that the 16mb hash i gave >Fritz6 old was too much for a 5 min./game and this was the reason for Fritz's >loses on time, but now i'am starting to wonder how true this might be! I reduced >the hash size for Fritz from 16mb down to 4mb for this Match, and so far Fritz >lost 9 games due to over-stepping the time, thats 6.52% out of 138 games played >so far! When we reach the total of 306 games as in my last Match, this will be >most interesting to see the result! I know why Hiarcs7.32 is doing much better >as the increase in hash size to 32mb from 16mb in the last Match. > >I checked the 9 games that Fritz lost on time and 8 out of 9 games were >definitly draws, but time means everything in blitz and there can be no excuses. > >Hope to hear some input from these observations! > >Regards,Terry Big hash tables can help Hiarcs. I believed that big hash tables are counter productive for Fritz in blitz(I remember from previous experiments that this is the case for Fritz5.32 but maybe it is not the case for Fritz6light and I did not check it). In my match Fritz6light lost only 3 games on time out of 144 Hiarcs was leading 53:47 after 100 games. The result was almost equal after 144 games and I prefer not to post it now because I am not sure that I have no error. 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.