Author: Uri Blass
Date: 16:33:28 10/12/02
Go up one level in this thread
On October 12, 2002 at 18:49:41, James T. Walker wrote: >On October 12, 2002 at 17:32:06, Odd Gunnar Malin wrote: > >>On October 12, 2002 at 16:40:03, Jorge Pichard wrote: >> >>>Deep Fritz peak node count (roughly positions per second) was around 3.5 >>>million. >>> >>>Take a look at the end of this Link: >>> >>>http://www.chessbase.com/columns/column.asp?pid=152 >> >>I was impressed by this: >> >>"It has surprised more than one observer that they are here without any top GM >>assistance for their openings, but of course this could be done by e-mail if >>required. And game four's Tarrasch Defense was suggested by Nigel Short, >>although Fritz had played it several times before with success." >> >>And we are trying to compare this match with the db-match :) >> >>Odd Gunnar > >The 3.5 million nps seems like a pipe dream to me. Running 8 x 900 mhz cpu >machine gives 7200 max. That's only about 5 times faster than my 1400 Mhz >Thunderbird. I don't have DF7 but my F7 runs typically 500K-750K. They would >have to be getting almost 100% efficiency out of the 8 cpus to hit 3.5M/nps. Is >that possible?? >Jim You assume that they use Deep Fritz7 for the match. I suspect that they use Deep Fritz8 when fritz8 is faster in nodes per seconds relative to fritz7 even on one processor. I remember that I read a post of GCP when he claimed that they plan to do Fritz faster in nodes per seconds relative to Fritz7 without losing knowledge by rewriting it. Uri
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.