Author: Drexel,Michael
Date: 01:33:25 02/16/03
Go up one level in this thread
On February 15, 2003 at 20:52:09, GuyHaworth wrote: > >Rob Hyatt's 1 + (N-1)*0.7 formula suggests that 4x 1.9GHz ~ 5.89GHz, say 6.0GHz > >I believe that an extra ply of search requires 3x the power. > >Ernst Heinz' self-play stats suggest that 1 ply is worth about 50 ELO at this >level. > >Therefore, DJ's TPR of 2847 'maps' to 2797 on 2GHz and 2747 on 0.67GHz. > >Thus, 2767 on 1.2GHz does not seem unreasonable in this context. > >As Uri points out, there would be quite a wide variance to DJ's ELO expectation, >based on only 6 games. > >So it's always worth remembering that we are dealing in means of 'expected ELO' >distributions. So, you might conclude that the probability of DJ's 2768 is x% >... but 'x' might be less than 50. > >g Nonsense calculations. 1 ply might be worth about 50 ELO in comp-comp games. That doesnt mean a thing in comp-human games. The gain is probably less than 15 ELO there. Computers are already far stronger in open positions, so all (strong) humans try to avoid complications unless they see something which is far beyond the horizon of the programs. Furthermore, no one knows if ~3Mn/s are really necessary for a top program to draw against Kasparov. Kasparov lost a 2 game rapid match against Genius on a P90 and many blitz games against Fritz 3 long ago. He is not one of the top players against computers. ~1Mn/s would have been sufficient to draw this match. My humble AMD is everything you need to draw against this Kasparov. Just let everyone know it is a 8 x 1.6 GHz :)
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.