Author: Thorsten Czub
Date: 11:38:43 02/23/00
Go up one level in this thread
On February 22, 2000 at 15:41:00, Jari Huikari wrote: >On February 22, 2000 at 08:24:52, Thorsten Czub wrote: > >>I am sure when i play 10 games manually >>(same machines and both 40/120), >>i don't get these strange >>results harald gets. >> >>what is the reason ? i have no idea. > >It'll need much more games that one can get trustworthy results. A few >games can end 6-0, 0-6 or whatever. Yesterday I tried my Nero against >itself giving different time controls for each side. It played more than >20 games and the side having 50% more thinking time losed more games than >the other side with shorter timecontrols. > > Jari i see no reason to untrust my own result more than haralds results. if you want i can start a match junior vs. cstal 20 games manually. or shredder versus cstal 20 games. i have no problems with that. i don't really understand what harald wants to show. we were the same opinion that a) shredder4 b) tiger12 c) junior6a are the strongest programs in the moment. harald tested shredder4 against cstal. i said: no wonder that cstal loses, shredder is the strongest program in the moment. harald said: it would also lose to others. i said: it would lose to other that are as strong as shredder4 is. now harald plays against junior. but junior is nearly as strong as shredder4. this has been shown in tournaments, in my own tournament e.g. shredder + junior are very close, in paderborn1999, in paderborn2000. i guess harald would get similar results testing cstal vs. tiger. what is the sense ? all 3 programs have the same level. and we don't know if the autoplayer changes results. i would say: if you really want a test, let me play against the same opponents manually and we correlate the results. IMO harald should better test on 2 machines the same speed, otherwise we cannot relate the results so good since cstal has to fight against similar hardware on my machines too. we should better relate our results. i could e.g. play the games manually from the same positions they left the book in haralds 20 shredder-cstal games. if i am right, and the autoplayer changes results, i shall get better results although cstal has no hardware advantage on MY system. I get BETTER results although tal has no hardware advantage relatively to haralds setting. what does this tell us ? shall we try out ? i could e.g. post the game each day here. it would take at least 20 days, because i have no autoplayer and 1 game per day is enough, but we would have data that can be related to each other.
This page took 0.01 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.