Re: [Publicity] [blog DRAFT] How fast is Xen on ARM, really?

[Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>]

On Sat, 24 May 2014, Anil Madhavapeddy wrote:
> On 23 May 2014, at 17:27, Stefano Stabellini 
> <stefano.stabellini@xxxxxxxxxxxxx> wrote:
> > 
> > 
> >> The postgres benchmark left me wondering why x86 is so much faster -- is 
> >> that because it just had a better disk?
> > 
> > I don't think that results such as 0.4% or 0.1% are very significant.
> > As I wrote in the article, the overhead is so low in the disk benchmarks
> > that they only manage to demonstrate that one needs a fast SSD to really
> > show the virtualization overhead.
> 
> you should generally scale the Y axis so that the threshold of significance
> is obvious to the reader.

You are right, that is a good idea, I'll make the change.


> In this case, it's at 0.8 (0.8 what? percent? relative to?)

Percentage of native performance, it is written right at the beginning
of the benchmark chapter.


> If it's not a significant result, then also explain that somewhere in the 
> text so the reader knows the implication.  But it's better to fix the graph 
> itself so that it's self-evident...

It is written right before "FIO".

_______________________________________________
Publicity mailing list
Publicity@xxxxxxxxxxxxxxxxxxxx
http://lists.xenproject.org/cgi-bin/mailman/listinfo/publicity