Author: Robert Hyatt
Date: 19:58:28 02/11/05
Go up one level in this thread
On February 11, 2005 at 17:51:53, Uri Blass wrote: >On February 11, 2005 at 17:38:27, Martin Andersen2 wrote: > >>On February 11, 2005 at 16:23:08, Uri Blass wrote: >> >> >> >>> >>>Is it based on tests with Crafty or only on the number of ghz? >>> >> >> >>What ? They are all opterons.. > >I still do not know based on the data if the hardware is more than twice faster. >4 processors are less than 2 times faster than 2 processors but 2.2( or 2.4)>1.8 > >Uri Math goes like this roughly. dual SMP speedup is 1.7. Multiply 1.7 by 1.8ghz and you get an effective 3.06ghz. The quad SMP speedup is 3.1 (actually higher but then so is the dual, but this fits the numbers I use which are pretty conservative, see the big discussion about SMP speedup last year for real speedups that Martin computed from the raw log files I put on the ftp machine). 3.1 x 2.4 = 7.44ghz. More than twice as fast. 2.5x faster is closer...
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.