Author: Robert Hyatt
Date: 20:57:31 04/21/01
Go up one level in this thread
On April 21, 2001 at 05:54:39, guy haworth wrote: > >ICGA_J v23.4 (Dec. 2000) gives a clue in two articles by Frederic Friedel. > >In June 2000, Fritz competed in Frankfurt on a Fujitsu Siemens Primergy N800 >computer with 8 processors @ 700MHz each. Fritz was working at 2.8M nps. >Fritz's Performance Rating in the tournament (after a 5-5 score) was ELO 2750. > >The following month, Junior 6 competed in Dortmund under 40/2 conditions on an >8-processor Fujitsu Siemens Primergy machine at 2.8M nps - perhaps the same >machine. > >Junior 6 also scored 50% for a tournament rating of 2703. > >Ken Thompson in an article way back said that 6x more power was required to gain >1 ply in search depth. I put this '6' alongside an assumed fan-out rate of 36 >and the square-root factor provided by a search method using alpha-beta cutoffs >to the maximum. > >Others may have views about the '36' and the alpha-beta contribution. > >What are others' current figures equating nps and search-depth? > >G Rather than 6, it is roughly 3 today. In my case, this is caused by the null-move search. 3x faster = 1 more ply. 9x = 2 plies, etc...
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.