On Sun, Jul 01, 2007 at 10:59:59PM +0900, Kazushi Marukawa wrote:
>    On Jun 29, 18:23, Manuel Bouyer wrote:
>    > Subject: Re: xen 3.1 problem (Re: xen 3.1.0 is there)
>    > 
>    > Maybe it's just a cache alignement issue with this specific application.
>    > My system is a dual-xeon 5050:
>    > (XEN) CPU: Trace cache: 12K uops, L1 D cache: 16K
>    > (XEN) CPU: L2 cache: 2048K
> 
> My system is a single Core Duo T2300.  I assumed Xen took
> care of dual core scheduing, so I've been using original
> XEN3_DOM0 and XEN3_DOMU without modification.  Should I
> configure DOM0 to support SMP?

No. the hypervisor takes care of it.

> 
> Anyway, Xen 3.1.0 is showing following.
> 
> (XEN) CPU: L1 I cache: 32K, L1 D cache: 32K
> (XEN) CPU: L2 cache: 2048K
> (XEN) CPU: Physical Processor ID: 0
> (XEN) CPU: Processor Core ID: 0
> 
> (XEN) CPU: L1 I cache: 32K, L1 D cache: 32K
> (XEN) CPU: L2 cache: 2048K
> (XEN) CPU: Physical Processor ID: 0
> (XEN) CPU: Processor Core ID: 1
> 
> Core Duo is using a single 2MB L2 cache from both core.  So,
> it is possible to happen some issue here.  No issue before,
> at least I didn't notice issues before, though.

But with cache issues, changing the size, or even just the place of
code can change results on this kind of test.

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--