Author: Jeremiah Penery
Date: 17:27:03 01/27/00
Go up one level in this thread
On January 27, 2000 at 20:01:55, Dann Corbit wrote: >On January 27, 2000 at 19:25:40, Jeremiah Penery wrote: >[snip] >>This is what you originally wrote: >>------ >>Do you know how Deep Blue searched? How will we reach those fantastic speeds? >>Faster hardware does not inflate ratings. It raises it by a logarithmic amount >>for each increase. It is estimated about 50 ELO. So, suppose we have a program >>with 2500 ELO and we want to get to 2800. >>2500 + 50ELO/doubling * 6 doublings = 2800. Sounds pretty simple, right? >> >>But if it takes 500 Mhz to get 2500 ELO, then it will take: >> >>500 * 2 * 2 * 2 * 2 * 2 * 2 = 32 GHz to achieve it (roughly speaking). >> >>IOW, more horsepower is a tough way to make chess programs play better. There >>is also evidence (according to some) that the increase in speed has >>*diminishing* returns. Hence, it may take a terahertz to get there. Don't know >>of any material that could do that, not even a Josephson Junction. >>------------ >> >>When you're talking about increasing Elo, I assume that's in games with some >>time control. When you said that the increase in speed has diminishing returns, >>I figured this was because of the exponential nature of the chess tree. I.e., >>at some point, the increase in speed to see any gain will be prohibitively high. > >I don't remember where I read it, and I can't find the article now. It may even >have been discussions on this board. As far as diminishing returns, I meant >that you don't always get 50 ELO for the next ply. At some point, maybe you get >40, then later 30, etc. I was taking your wording literally that *speed*, not *depth*, has diminishing returns. Because the speed has to go up exponentially to see more depth. And seeing more depth is the only way to get a better rating (using the same program, of course).
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.