Re: [Xen-devel] Xen 4.5 development update (July update)

[Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>]

On Tue, Aug 19, 2014 at 06:46:51PM +0200, Dario Faggioli wrote:
> On ven, 2014-08-15 at 13:23 -0400, konrad.wilk@xxxxxxxxxx wrote:
> 
> > = Open =
> 
> > == x86 == 
> 
> > *  HT enabled, virtualization overhead is high (Xen 4.4) (none)
> >    kernbench demonstrated it
> >    Looking and tracing
> >    http://www.gossamer-threads.com/lists/xen/devel/339409
> >   -  Dario Faggioli
> > 
> Mmm... I realize only now that I must have missed your last email about
> this... So, sorry.
> 
> In any case, didn't we settle on closing it? The bullet is here because
> very high perf degradation between baremetal and HVM DomU was seen when
> running `kernbench -j' (i.e., kernbench with unlimited number of build
> jobs) with HT enabled.
> 
> Both me and other people tried to reproduce the issue, and here's the
> number:
> 
> HT Enabled
> ==========
>                  Host               Guest
> kernbench -j4    31.598 (0.113666)  34.078 (0.168582)
> kernbench -j8    26.676 (0.16772)   26.672 (0.0432435)
> kernbench -j     27.674 (0.933584)  27.49 (0.364897)
> 
> To me, this means we don't have an issue of high virtualization overhead
> when HT is enabled (look at the last row!).

Ok, so false alarm. Thank you!
> 
> All the other things that came up in previous threads are not to be
> ignored or forgot, of course, and I won't, but they don't have anything
> to do with this.
> 
> Let me know,
> Dario
> 
> -- 
> <<This happens because I choose it to happen!>> (Raistlin Majere)
> -----------------------------------------------------------------
> Dario Faggioli, Ph.D, http://about.me/dario.faggioli
> Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)
> 



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel