Author: Vincent Diepeveen
Date: 15:27:44 04/18/01
Go up one level in this thread
On April 17, 2001 at 23:31:41, Robert Hyatt wrote: >On April 17, 2001 at 18:55:58, Chris Carson wrote: > >>Thanks for the info! :) >> >>Could you ask the participants to get a NPS bench mark on one processor and a >>NPS benchmark for the 8 cpu's. >> >>1.7 seems to be a good ratio for a 2 CPU system, but I would like to get some >>information on an 8 CPU system. Thanks. >> >>Best Regards, >>Chris Carson > > >I don't know how the others do. For crafty, a "pessimistic" formula >would be > > speedup = N - (N-1)*.3 > >for N=8 (8 cpus) this turns out to be roughly 6.0 times faster. > >That would be a _serious_ handicap for a program running on a single cpu... >And a match on a dual processor machine will show _nothing_ about how the >programs will do on an 8-way box... the biggest problem is not so much the fast machine, but that crafty also profits from deeper search in positional terms. It is tuned on outsearching opponents over the years (though work of bob continues), so it would be a real hard nut to crack if equipped with a reasonable book. The only problem would be book actually not so much whether it's on 8 processors or 16.
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.