Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Linux vs. Windows performance.

Re: Linux vs. Windows performance.

From: Rick Denoire <100.17706_at_germanynet.de>
Date: Sun, 15 Feb 2004 16:08:37 +0100
Message-ID: <te2v205j0dgi5peusbu645vfc1l81ps3ri@4ax.com>


drak0nian_at_yahoo.com (Paul Drake) wrote:

>if the OS were to schedule execution of 2 processes on both the
>physical and logical CPUs of a single physical processor, while not
>schedule execution of a process on either the physical or logical CPU
>of the second physical processor, I would certainly not consider that
>to be "support".

And consider CPU affinity for processes being put in an out of the queue. If the OS is not aware that the logical CPU resides in the same physical CPU whose cache holds process data, the cached data will be invalidated I think.

I use the same PE2550 (2 CPUs) with Redhat Linux AS 2.1 and Oracle 9.2.0.4. At times, CPU load is short of 400%!

By the way, I consulted Oracle on this question and they did not find anything against hyperthreading in the specs, I was told.

Bye
Rick Denoire Received on Sun Feb 15 2004 - 09:08:37 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US