Author: Uri Blass
Date: 10:36:03 05/31/02
Go up one level in this thread
On May 31, 2002 at 12:35:18, pavel wrote: >On May 31, 2002 at 06:50:56, stuart taylor wrote: > >>For me, the experiment is over. Hiarcs does not look especially great at higher >>time controls. Even if it would win a match like this (although it is looking a >>bit doubtful) it is clear that it does not excell at this time allowance. >> Maybe at an hour per move it would, but the indications for that are also >>nothing special, as this is not showing any movement in that direction either. >> >>So, in spite of the one or two brilliances which Hiarcs 8.0 might be capable >>of, as written about in Uniaks article, I don't know if I want to buy it for >>that alone. >>S.Taylor > >Obviously IMO the thought about Hiarcs being better at long time control is very >wrong. Its not about long time control, if its about not searching the same >depth as it opponent, then I dont see how Hiarcs will achieve it even on the >most fastest hadware with the same tme control, beacause Hiarcs in almost all >occasions got outsearched by Fritz, hiarcs is just too slow. But I dont think >this is a prblem, because according to the author Hiarcs is a knowledge based >program, so I am assuming that it doesnt necessarily need to search deep enough >to "understand' any position, unlike most search-based program. I do not agree. The nominal depth means nothing. Hiarcs saw the mate against itself some moves before Fritz saw the mate for itself. I do not think that Fritz's search is clearly superior because the data does not suggest that Fritz fails high when Hiarcs some plies after it fails low. I do not believe that Hiarcs's evaluation is better than Fritz otherwise Hiarcs could beat Fritz because Fritz has no big advantage in it's search algorithm and sometimes even hiarcs can see things faster than Fritz. 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.