Author: Robert Pope
Date: 06:42:49 08/03/99
Go up one level in this thread
On August 03, 1999 at 03:40:36, Terry Ripple wrote: >On August 02, 1999 at 21:32:56, Victor Valenzia wrote: > <snip> >---------- > Hi Victor, > Your Anology is interesting but lets face the facts that both programs have >the same "handicap" and the only difference it`s going to make is that the >quality of the games could be a little more in depth in some games played with >pondering on and other games it won`t make much difference, but in any case it`s >still a fair game as long as the handicap is affecting both programs in the >match. But by this argument, we might as well just play blitz games. The quality may suffer a little, but the handicap is affecting both programs in the match. We know this doesn't hold water. There can be significant differences in how much gain a program gets from using a variety of features. Changing any of the following could quite possibly bias results from any given program, even if we do the same to the other program: - using a different (or no) opening book - using different time controls - disabling pondering - changing hashtable size - use (or not) of tablebases I know that for my old chess program, I never got around to adding pondering at all. So engine vs. engine on one computer would be a _huge_ advantage for me, not that it would win many anyways :). But I wouldn't be surprised if different pondering strategies (e.g. instant-reply or assumed extra time) created bigger performance gaps in particular programs than in others. I don't do a significant amount of engine vs. engine, though, so my ideas aren't biased by actual results. ;) I am surprised that there isn't more engine vs. engine long-match testing done via the chess servers (or maybe there is). It seems to be ideally suited for this. I wouldn't think it would be too hard to find two CCC members with identical hardware who could autoplay a 100 game match or something similar. Rob
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.