[Soekris] Harddisk Slow

David Young dyoung at pobox.com
Wed Jan 31 23:22:09 UTC 2007


On Wed, Jan 31, 2007 at 11:36:33PM +0100, Igor Sobrado wrote:
> > On Wednesday, January 31, David Young wrote:
> > > 
> > > Sorry to ask such a basic question, but when you say "the interrupt load,"
> > > do you mean that the %CPU spent processing interrupts is higher, or do you
> > > mean that the number of interrupts per work done by the IDE controller
> > > is higher?  I ask because it makes sense to me that the %CPU would be
> > > higher than on your PowerEdge 350, however, it doesn't immediately make
> > > sense to me that the number of interrupts would be higher.
> > 
> > Possibly both?  I mean, the two pieces of machinery are not even on
> > the same planet.  Different IDE controllers, different capabilities,
> > etc ad nauseum.
> 
> Hi Tobias and David!
> 
> Indeed, there are different computers and even operating systems
> in this thread.  Certainly, the PowerEdge 350 is not comparable
> to a net4801, even if the PowerEdge only requires a 75W power
> supply unit.  Both my laptop and the PowerEdge are running
> NetBSD right now, I am considering a migration to OpenBSD on some
> machines but at this time only the net4801 is running OpenBSD.
> 
> However, there is a new thread on misc@ (an OpenBSD mailing list)
> about performance problems with pciide on the sparc64 architecture.
> I suppose that this problem can be related with the one described
> here.  The thread (just started, but the problem has been nicely
> outlined by Rolf Sommerhalder) has this subject:
> 
>   High Interrupt Load cased by pciide with sparc64 on SUN V210
> 
> It can be found here:
> 
>   http://marc.theaimsgroup.com/?l=openbsd-misc&m=117027330031702&w=2
> 
> It seems that pciide0 generates interrupts at a very high rate
> (there is a permanent interrupt load between 30% and 60%, and
> the v210 does not have a low-end processor).
> 
> If there is a problem with OpenBSD, I am sure the development
> team will release a patch very soon.  They work really hard to
> fix problems in the operating system.
> 
> David, don't know if this answer is what you are looking for.
> Perhaps the information provided by Sommerhalder on his thread
> will provide a better feedback.

I was looking for.  I was hoping you could provide some evidence from
'top' or 'vmstat -i'.  Here is my question again.

On Wednesday, January 31, David Young wrote:
> 
> Sorry to ask such a basic question, but when you say "the interrupt load,"
> do you mean that the %CPU spent processing interrupts is higher, or do you
> mean that the number of interrupts per work done by the IDE controller
> is higher?

Dave

-- 
David Young             OJC Technologies
dyoung at ojctech.com      Urbana, IL * (217) 278-3933


More information about the Soekris-tech mailing list