Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Commercial Version of Deep -surely not so deep- Blue announced.

Author: Eelco de Groot

Date: 17:27:24 05/03/99

Go up one level in this thread



Thanks for your posting, Richard! This is certainly interesting news, a possible
Deep Blue for consumers. If Mr. Hsu could really produce a chip that searches 30
million pos./sec  this would be about 150 times faster than the best current
desktop PC's running commercial chessprograms. At the selectivity Deep Blue
seems to have used, 1000 nodes for a 4 ply search, (This only applies for
certain to the first 8 plies of a typical  12 ply search, so not using the
chips), 150 nodes would translate to almost a 3 ply search. Every existing
program could thus extend every endnode by three  plies, at a negligable cost in
CPU time,  in the time it used to do it's own endnode evaluation! This is
assuming Mr. Hsu would be willing to make his chip an open standard, I
personally would very much hope he does.



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.