Re: xen on i7-12700: works with cet-no-ibt !!

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Wed, Sep 18, 2024 at 12:48:14PM -0400, Greg Troxel wrote:
> Manuel Bouyer <bouyer%antioche.eu.org@localhost> writes:
> 
> > I did the m4 build test on a
> > cpu0: "Intel(R) Xeon(R) W-2223 CPU @ 3.60GHz"                   
> >
> > PVH
> >        47.39 real        31.13 user        13.48 sys
> >        47.19 real        31.27 user        13.54 sys
> >        47.26 real        31.23 user        13.65 sys
> >        47.65 real        32.26 user        13.01 sys
> >
> > PVHVM
> >        47.30 real        31.13 user        13.36 sys
> >        47.07 real        31.30 user        13.39 sys
> >        47.55 real        31.80 user        13.37 sys
> >        47.13 real        31.89 user        12.89 sys
> >
> > (no bare metal test here, it's a production machine).
> > I'd say the time differences are actually in the noise margin - especially
> > as the host is not idle.
> 
> Those are certainly self-consistent and with each other.   I don't
> really know what's going on, of course, but what seems clear is that
> people should compare pvh to pvhvm to decide what to do, if they have a
> choice.

It would be interesting to find why in some case PVHVM appars faster.
There may be a performance bug somewhere ...

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