Bryan Whitehead wrote on Mon, Jul 29, 2002 at 11:51:27AM -0700 :
> >
> >You do realize that using this patch pretty much prevents the machine
> >from being used as a high capacity server.  It allow the kernel to be
> This is totally untrue. In fact the preemptive kernel has been added to 
> the 2.5 series and will be in 2.6 because it results in better 
> preformance and lower latency. Much of userspace is spent in kernel 
> space getting IO done. A preemptive kernel allows any process to be 
> interrupted even while in the kernel. For example, when X needs to draw 
> some stuff to the video card, that is done at a lower level in the 
> kernel. A non-premptive kernel would not be able to interupt the process 
> while it was in kernel space. The preeptive kernel can.

How is interrupting I/O to redraw the desktop a benefit in performacne?
In "perceived" performance, yes.  In raw I/O?  I don't see it.  I do
trust your opinion though Bryan, I would just like a little
clarification.  I will go read those URL's.

> Some more stuff to read:
> http://www.linuxjournal.com/article.php?sid=5833 <- kernel locking
> http://vig.pearsoned.com/samplechapter/0130224960.pdf <- Sample chapter
>       on Solaris Kernel internals
> http://www.tech9.net/rml/linux/ <- more links of how linux preemptive
>       kernel works

Looks like interesting reading.

Blue skies...           Todd
-- 
  Todd Lyons -- MandrakeSoft, Inc.   http://www.mandrakesoft.com/
UNIX was not designed to stop you from doing stupid things, because 
  that would also stop you from doing clever things. -- Doug Gwyn
   Cooker Version mandrake-release-9.0-0.2mdk Kernel 2.4.18-21mdk

Attachment: msg56619/pgp00000.pgp
Description: PGP signature

Reply via email to