Author: Bruce Moreland
Date: 12:36:22 02/27/99
Go up one level in this thread
On February 27, 1999 at 14:04:32, Matt Frank wrote: >I would like to submit the argument that the top micro programs (e.g., Hiarcs 7, >Rebel 10 EOC Anti-Grand Master, and Fritz 5.32) have features that Deeper Blue >did not have; assuming the information on Deeper Blue is reliable. For example, >Deeper Blue did not have a learning function to my knowledge. Even if it did, it >relied on a huge multi-processing configuration to do the calcuations, and this >feature alone introduces software technical issues that the present single CPU >micro programs don't have to deal with. My major point is this; Deeper Blue >(circa 1997) was not as evolved a software system (please note: SOFTWARE SYSTEM) >as todays micro based programs. And consequently, was not able to squeeze as >much chess playing skill out of its enourmous computational resouces compared to >todays micro/software configs. In fact if Deeper Blue's software could be made >to play as slow as on a typical Pentium 450 machine it would get beaten severely >by todays programs, IMHO. A learning function has nothing to do with how "evolved" a chess program is. Beyond a doubt it is a good thing to have in a program, but it says nothing about the state of the rest of the program. I'm sure it would have had a learning function if it had to compete in an environment where a learning function would have added any strength. The strength comparison on a P2/450 is meaningless, it's like saying that a Volkswagen Beetle is better than a dragster because if you put a Volkswagen engine in the dragster, the Volkswagen would do a faster quarter mile. 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.