Xeon/P4 hyperthreading

William Park opengeometry-FFYn/CNdgSA at public.gmane.org
Sun Oct 12 23:33:40 UTC 2003


On Sun, Oct 12, 2003 at 07:12:03PM -0400, cbbrowne-HInyCGIudOg at public.gmane.org wrote:
> > If there's anyone here with P4 or Xeon with hyperthreading and SMP
> > kernel, can you compile kernel
> >     make bzImage > /dev/null
> >     time make -j 3 bzImage > /dev/null
> >     time make bzImage > /dev/null		
> > and post the elapsed time for the last 2 commands.
> > 
> > Thanks.  It's time for upgrade, and I would like to get some feel for
> > hyperthreading in P4 and Xeon.  
> 
> I have one of these to "play with;" the problem with your benchmark is
> that it doesn't quite measure what you think it does.
> 
> In the case of the box I have, I'd have to use "make -j 12" in order to
> get the result you are thinking of, because the box has 4 CPUs,
> pretending to be 8, and you need to "flood" them to get a good feel for
> things.
> 
> I haven't been able to test the box for how much difference (positive or
> negative) HyperThreading makes.  The box is generally stinkin' fast...
> 
> 12 UW SCSI drives with a cacheing RAID controller helps too :-).

'-j 3' is for those who have single P4 or Xeon.  Single CPU result will
give me better feel of things.

-- 
William Park, Open Geometry Consulting, <opengeometry-FFYn/CNdgSA at public.gmane.org>
Linux solution for data management and processing. 
--
The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml





More information about the Legacy mailing list