Author: Harald Faber
Date: 05:02:18 07/05/01
Go up one level in this thread
On July 05, 2001 at 07:08:29, Fabio Barrettone wrote: >>Don't forget the differences between your testing and mine: >>1) At the moment I test the normal GambitTiger 2, not with aggressive settings, >>they will be tested later on. > >Ok, I didn't know this. > >>2) I play a different time control. 40/40 is 3x faster than 40/120 ! > >True. Tiger is a program that plays VERY slow. It is very seldom that Tiger moves under 5min. I often saw Tiger use 15-25 min for a move. That led to games in which Tiger had only 10min left for the last 5-10 moves to the time control at move #40 while the opponent had more than 40min left. >>3) Do you play on one computer? > >Yes, I have an Athlon 950mhz with Win2k server and 256 Ram. > >I understand all that you say, but your results are too much different from >mines. The difference is too huge. It is true that all the things you said taken >as a whole could maybe put a lot of difference in the results, but remember that >in yours games Shredder seems to be a *lot* stronger than GTII, where in mine >GTII they were very even, and in the last GTII won. The point here is that it SEEMS to be. 10 games are by far not enough games to have evidence. And to have evidence, one needs to compare both programs against the same opponents too, not only against each other. But this would take so much time that I am not willing to test this out nor I have the time for it. So my tournament is not set up to show if Shredder is stronger than Tiger but to be able to compare the different Tigers. And at the moment I must say that ChessTiger did better, although I might be wrong. It can also happen that GambitTiger is going to win (much) higher against the resting opponents and only Shredder 5.32 is his "Angstgegner" or had only a blackout in these 10 games. But at last I, and you too of course, will be able to compare much more than 10 games, at least 100 games of each Tiger, so this will say a bit more than this single GT2-Shr532-match. >This naturally not to say that your results are not right, only that maybe 10 >games are too little to judge a thing as this. Sure, but I repeat: look at the games. >But now I wonder also myself, and I will do some matches beetween the two >engines at tournamente time control. There are always some things to learn about >computer chess, and one is that you cannot be sure of nothing, well, at last >with the result of the test you personally made at home ;-))) I am still convinced that there is a difference between playing on one or on two computers.. Even if I am the last who is. :-) >I personally have ever thought that what Christophe said about processor speed >and time to think where not so important with engines match, but for what I see >watching play Junior and Shredder it doesn't seem so. They both seems to play >better and better the more time they have to calculate, and the seems to be true >also about Gandalf. Well... I'm really confused at the moment :-) > >Sincerely, >Fabio.
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.