Author: Robert Hyatt
Date: 21:16:53 07/18/00
Go up one level in this thread
On July 18, 2000 at 23:32:41, Albert Silver wrote: >On July 18, 2000 at 17:48:32, Robert Hyatt wrote: > >>On July 18, 2000 at 14:51:41, Albert Silver wrote: >> >>>On July 18, 2000 at 14:36:00, Bruce Moreland wrote: >>> >>>>On July 18, 2000 at 13:39:01, Jerry Adams wrote: >>>> >>>>> >>>>> I doubt if DeepBlue with all it's billions of calculations per second could >>>>>score much better than DeepJunior at Dortmund. Seem it is a bad day for the >>>>>Advocates of "Hardware is everything" Theory. Deepblue could probally Easily >>>>>Defeat DeepJr in a Match, but against humans the story is different. I hope >>>>>programmers Continue to Develope Software and not sit back lazily waiting for >>>>>Hardware to do all the work. >>>> >>>>When I saw the thread title I thought you might have gotten the title backwards, >>>>since Junior had huge hardware at Dortmund. But then I saw the reference to DB >>>>and I said, "Oh, that." >>>> >>>>I think all that was proven here is that a computer can do well in a tournament >>>>with GM's. I haven't looked at all the games, but from the results it seems >>>>like the computer belonged in the tournament. That's a big thing. >>>> >>>>Nothing about DB is proven, because the DB matches were short, Kasparov was >>>>stressed, and the computer arrived out of nowhere and vanished without a trace. >>>>It could be a lot better or worse than we perceive it now, and we have no >>>>practical way of knowing which. >>>> >>>>Junior at least seems to be willing to come out and play, and I commend Amir for >>>>this. >>>> >>>>The fact is though that Junior was on some great hardware, so I don't know if >>>>you are proving your point by bringing up the issue of hardware, or disproving >>>>it. >>> >>>One way, though one would have to have the accurate move times to do so, would >>>be to simply take an inferior machine and see if it is capable of finding the >>>moves played in the games. One wouldn't be able to calculate any Elo ratings >>>with this, but it would be interesting to see how many moves were made possible >>>due to the hardware. If Amir has this available, I think it would be wonderful >>>if he could post this (the time per move for both sides). >>> >>> Albert Silver >>> >> >> >>Anecdotal evidence also suggests that hardware is very important. Each time >>I upgraded on ICC, from the P6/200, to the 4 X p6/200, to the quad xeon/450, >>to the quad xeon/550... GM players would comment on its newfound skill... >>IE, "what did you do to it recently? It has really improved... etc..." >> >>And hardware would be the only change of any significance, and they could "feel >>the difference..." >> >>I'd think an 8x700mhz machine would be very strong feeling... > >I have no doubt, but they are commenting on Blitz play for the most part. How >big a difference would this represent in 40/2 as was the case for DJ? I'm sure >it is _much_ stronger but I think that human players do a heck of a lot more >catching up with more time on the clock. > > Albert Silver A couple of the ones I talk to regularly play some 15 15 and even 30 30 games when no one is noticing. Faster hardware just ramps up the tactical acuity a good bit, and occasionally it will see something positional that it wouldn't have seen without the extra depth... > >> >> >> >> >>>> >>>>bruce
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.