Author: Robert Hyatt
Date: 10:41:05 05/23/01
Go up one level in this thread
On May 23, 2001 at 02:17:50, K. Burcham wrote: There is one problem with your "class" idea. At the drag strip (and yes my son and I drag-race his mustang regularly) there are enough entries so that multiple classes work well to even out the competition and money factor, and let the 'little guys' avoid the 'big guys'. But in chess, there are maybe 100 programs around the world. If you divide them into classes, each class will have just a couple of players in it as at any event you are lucky to get 32 participants... > >when we race, we dyno first. yes we have variables in our computer printout, >barometric pressure, ambient temperature, operator error, slight varibilities in >software parameter settings. > >all of these effect our "horsepower numbers" that the pc prints out. > >but they will be close. and this tells us that we are ready to race with the >best. that we have done all we can with the parts that are available today. >we tune this hardware to get the highest numbers we can. only when you do this >can your total package compete against another competitor that has also put his >deal together on this level. this is standard at the top. > >when i see someone post here about a programs results, running on a 550 pentium >lll, >or really even an 800 amd, in a tournament----it just means to me that with that >operator, and with that program, it had those results with a certain opening, >and with outdated hardware. > >i am not saying that i dont respect tiger or fritz running on a 900 amd, >and sufficient ram for a 900. this will be a very strong game. >and few humans can play on this level. > >what i am saying is that if we have tournaments in the year 2001, then any >up-to-date results will have to be on up-to-date hardware. > >if someone posts here, such as sarah, and uses two 950 amds to play fritz >against tiger. then i feel that is an accurate result to judge the strength >of a program, if enough games are played in the match. > >but if match or tournament results are posted with inferior hardware, then the >results have inferior meaning. > >for example tiger on 1000 mhz vs deep fritz on a dual 1 gig. >tiger on 1500 mhz vs junior on 1 gig. > >we all know mhz isnt everything, but i dont see how some of the results posted >here can be an accurate indication of a programs strength. > >and i am not sure how close this mhz rule has to be for programs in a >tournament. >with some showing up on quads, and some on dual one gigs, it seems the results >are not true. > >we all rate our humans. we think that a master with all his knowledge and chess >skills, cannot beat a supergrandmaster, even when the super gm is having a bad >day. i watch game after game at chess.net when a gm plays several masters in a >row. it almost seems like the gm plays with these guys. it is another level that >very few masters will ever achieve. > >i think it is the same way with our hardware. we must use the most advanced >hardware to really know what level these programs are capable of playing on. > >if crafty shows up at a tournament on two of the newly released 1700 mhz xeon >processors, and mr hyatt has sufficient ram, and his opening book is well >prepared, and with a little luck, and if everything else goes as planned---he >wins a tournament against the worlds best on lesser hardware. what would this >mean to me. it just means that on that day with that hardware he had the best >results. > >until we start setting hardware limitations, and have defined classes, then >these results will be inaccurate. > >the tournament coming up at icc. >it seems the hardware is not equal. >i know this is not possible. > >i will still enjoy reading about the results of this tournament > >i always wonder when i read results of some top commercial program on top >hardware, vs a freeware program on small hardware, what the results would have >been on equal hardware. instead of being 40 moves to mate, maybe it would have >been 65 moves to mate, or maybe even a draw. > >i would like to see classes at each tournament: > single processors > 1000 mhz only > 1500 mhz only > 2000 mhz only > > dual processors > 1000 mhz only > 1500 mhz only > 2000 mhz only > > quad processors > 1000 mhz only > 1500 mhz only > 2000 mhz only > >also some rules on time management. >also each class has a limitation on ram. >also each program must run on the same operating system. >also each program must use the same opening book given to them on tournament >day. > >until some rules similiar to these are excepted, our tournament results will >seem to the outside world as unprofessional. > >i want to see shredder5, fritz6, junior6, nimzo8, cm8000, rebel century 3.0, >hiarcs 7.32, gambit tiger, tiger 14, crafty, sos, all the deeps, and some >others, on the exact same mhz, with the same book and same ram. > >but i think this is asking too much. too difficult to arrange for a tournament. > >but enjoyable to think about the possibilities and wonder what the results would >be. > > >thanks
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.