Author: Vincent Diepeveen
Date: 06:32:43 10/11/02
Go up one level in this thread
On October 10, 2002 at 04:22:01, Gregor Overney wrote: >On October 09, 2002 at 03:52:14, Georg v. Zimmermann wrote: > >>Fritz might well be stronger than DeepBlue. >> >>So you are an expert. What programs have you written ? >> >> >>Georg v. Zimmermann > >Speech recognizer, image processing software, local density approximation (LDA) >code for Cray systems, firmware for mass spectrometers, a quantum chemistry >program dealing with empirical potentials, ..... it's really difficult to >remember all programs I have written. BTW, why do you ask? > >I know the quality behind an implementation to perform evaluation and searches >when I read a paper about a chess playing system, such as DB or look through the >Crafty code. > >What papers do you read? > >Gregor But if you ask an IM to judge the games db - kasparov 1997, then you would get back as answer: NM level. Which for non insiders stands for National Master level. Further the number of computerchess experts which are capable of judging how inefficient the parallel search from DB was, you can count them on like 2 hands at most. The whole aim of the machine was to get as many nodes a second as possible, that's what the whole project clearly shows. This job Hsu did magnificent. His thing got an estimated 126MLN nodes a second on average in the match versus kasparov. All other things were irrelevant obviously. Such as search depth, effective parallel speedup, evaluation quality and quality of hand tuned openingsbook (just 4000 moves hand tuned, that's not even covering the najdorf mainlines... ).
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.