Author: Frank Phillips
Date: 11:33:16 01/11/03
Go up one level in this thread
On January 11, 2003 at 13:48:16, Christopher A. Morgan wrote: > > >Levy is president of ICGA, the organization that is hosting the match. He >states a well known fact about the relative speed of the Deep Blue hardware at >200 million nodes per second and the likely speed of whatever machine the DJ >team will use, 2-4 million nodes per second. Given that fact he has to say >something about today’s programs on the hardware it will use in the match in >comparison to the 1997 Deep Blue v Kasparov match. We should expect him to say >exactly what he said. I would not give a whole lot of credence to his >statement. > >What is fairly apparent now is that speeds of 2 million nps or so seem >sufficient to give super grandmasters, 2700 ELO plus, fits using any of the >various top programs. It seems that 200 million nps is an overkill to be >competitive with the world champ. There seems to be little discussion of this >point. That is that Deep Blue running at, say 4 million nps could have been >just as competitive as it was with GK in 1997, as it was running at 200 million >nps. > This is the point. If Junior really does have better evaluation with (what I can only imagine is) much less knowledge and like Fritz against Kramnik can hold its own against Kasparov at much lower tactical speed (depth) than Deep Blue had, then we may be closer to solving chess in the practical sense than has, perhaps, been previously thought. The results of software running on general purpose computers (Rebel, Fritz, Chessmaster, Tiger) at standardish time controls against various GMs in the past year has been surprising to me. Vincent attributes this, further down in this thread, in part to unspecified advances since Deep Blue. It would be interesting to know if there has been a genuine revolution (rather than evolution) in commercial computer chess software. I had thought that everybody other than Deep Blue was doing pretty much the same thing, but with better tuning and different emphasis on the same component parts. Frank >
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.