cpuinfo, bogomips and duo core

[Gordan Bobic]

On 12/02/2013 12:17, Reindl Harald wrote:
>
>
> Am 12.02.2013 12:58, schrieb Gordan Bobic:
>> On 12/02/2013 10:46, Reindl Harald wrote:
>>> means:
>>> you buy much better hardware with more and faster CPU's
>>> for a single device as you would buy for 20 machines
>>> and most of the day one or two guests can allocate
>>> most of the ressources on their own
>>
>> Sure, but that's consolidation, assuming you have loads that tessellate nicely. If your load is capable of
>> saturating the bare metal, your performance will take a substantial hit if you virtualize. If you want to argue
>> otherwise, describe your test methodology and results. Things may have improved somewhat from Core 2 to Core i, I
>> haven't re-tested on my most recent hardware - plan to do so in a week or so.
>
> my test methodology is practical workload and not constructed crap
> i profile my own web-applications which are highly optimized

And you generated your saturation level load how, exactly? Simple 
redneck tests are the ones that hypervisor's smoke-and-mirrors cannot 
cheat, and thus the only ones that will show you the true state of 
affairs. Otherwise you might as well swallow the marketing brochure 
wholesale and stick your head in the sand.

> believe it or not, they are faster in a VMware guest as on the host
> itself in many cases, both the same Fedora version with binary
> identical packages optimized for corei7 and SSE4.2
>
> -O3 -march=corei7 -mtune=corei7 -mmmx -msse2 -msse3 -msse4.1 -msse4.2 -maes -fopenmp -mfpmath=sse

That just tells me you didn't push the machine to full saturation.

Virtualization takes resources, and you cannot go faster by adding 
overheads. The only exception to this is where you have hardware 
specifically designed for this, with a hypervisor in firmware running on 
proprietary resources that a generic bare metal OS wouldn't be able to 
access anyway (or at least it wouldn't know what to do with it). In such 
cases you have an added bonus that you can reboot the host without 
switching off the guests (and without migrating them elsewhere, either). 
But hardware like that is very proprietary and uncommon.

Gordan