Author: Stephen A. Boak
Date: 10:12:45 09/01/00
Go up one level in this thread
On September 01, 2000 at 11:44:52, Jonathan Lee wrote: >Statistically speaking I'd like to find how many GHZ does it match an IGM middle >game. >For example, using a 1 GHZ comp. use Kasparov's middle game (30th move until the >50th move), when he played 40 moves in 2 hours time control. >With 1 GHZ, you could let it run for 12 or 24 hours, and see if that Kasparov >move makes the same move by the computer. >You can then say (with tounge and cheek, very unofficial, 90 percent of >Kasparov's moves happened at 500 GHZ or less). > >If your 1 GHZ took 24 hours to find the move, then it matched 480 GHZ. >(24 hours divided by 3 minutes per move equals 480) >If your 1 GHZ took 12 hours to find the move, then it matched 240 GHZ. >(12 hours divided by 3 minutes per move equal 240) > >3 minutes per move came from 40 moves in 2 hours. > >This will take "hours" or "a day" to figure out, but we can figure out where >Kasparov stands on faster computers in the future. >Jonathan (59th message) Your plan will provide some data, but I'm afraid it will not be very meaningful. Do you assume that Kasparov's middle-game moves are 'perfect', say between move 30 and move 50? Do you assume that the computer/program moves are perfect? I guess that you will be lucky to find even a single game of thousands where the computer selected move is identical to a human move for 20 moves in a row in the middlegame. --Steve
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.