on Thu, May 24, 2001 at 08:50:33PM -0700, Mike Egglestone ([EMAIL PROTECTED]) wrote: > Thanks for the input guys... > I took it upon myself to do some VERY basic testing... > and I'll share my results if anyone's interested..... > > The test: untarring a 88MB file full of .gifs and .html files > > Imac 400 Mhz 128 Ram IDE drive > time= 53 seconds > > Celeron 700 Mhz 128 Ram IDE drive > time= 80 seconds > > Celeron 700 Mhz 128 Ram SCSI drive > time=21 seconds... > > Weird... seems like hard drive performance is > very important too....:)
That's a piss-poor test for evaluating processor speed. It is a reasonable proxy for total system performance. If you want to get a feel for what processor can do for your system, try a processor-bound process. Several of the distributed clients are probably good options, as are graphic image manipulations and other raw computational tasks. Reality is often quite different. My former life was as a SAS analyst, usually on Unix boxen. SAS is a high-end data analysis, statistics, and reporting tool, used by 98% of the Fortune 500, on data sets ranging from several MB to multipble TB. I have a standard rant for people who ask for new box specs. My emphases for SAS are disk, including multiple IO channels, striping, and RAID, as appropriate, RAM, and processor. If all else is feeding smoothly, CPU will help you out. But swapping or waiting for disk I/O will trip you up far more than any incremental increase in CPU -- we're talking multiple orders of magnitude (typically 100-1000 X) rather than a fraction, or even factor, of two or so. Real-life tasks are a mix of task switching, pulling data from disk, and trying to avoid swapping active processes. -- Karsten M. Self <kmself@ix.netcom.com> http://kmself.home.netcom.com/ What part of "Gestalt" don't you understand? There is no K5 cabal http://gestalt-system.sourceforge.net/ http://www.kuro5hin.org Disclaimer: http://www.goldmark.org/jeff/stupid-disclaimers/
pgp7gSL237T6H.pgp
Description: PGP signature
