Author: Dann Corbit
Date: 10:22:00 03/22/05
Go up one level in this thread
On March 22, 2005 at 06:38:10, Roger D Davis wrote: >On March 21, 2005 at 11:32:38, Vincent Lejeune wrote: > >>http://hydrachess.com/ >> >>"The developers expect to break the 3000 Elo barrier till the end of this year. >>" is quite clear. >>A video (19 min duration) with a quick history of computer chess and history of >>hydra's games. >>hydra games + some pictures + hydra informations >> >>great ! > >If with 32 parallelized FPGA cards the Hydra team can break 3000, and if >Shredder is rated 2800, then is it fair to say that each card gives Hydra an >approximately 6 ELO advantage over the best program on a fast AMD-based >microcomputer? I am pretty sure it won't be anything close to linear for Elo improvement per card. Look at the Elo of a single Deep Blue chip, and then consider a cluster of 480 chips. Probably going from 2500 to 2900 Elo, for 480 times as much horsepower. Since chess is exponential, and since there are sure to be SMP losses, you can expect to throw a bigger mountain of power for every new Elo.
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.