[uug] next cpu for virtualization

[Thomas S Hatch]

On Sat, May 12, 2012 at 8:39 AM, Aaron Toponce <aaron.toponce at gmail.com>wrote:

> On Fri, May 11, 2012 at 03:20:23PM -0600, Daniel Fussell wrote:
> > Does anyone have any experience with the new Opterons that can confirm
> > if they are DOA performance-wise?  Or does recompiling with special gcc
> > flags to optimize the multiply ordering restore AMD's standing for
> > threaded server performance?
>
> I have experience with the Intel Westmere Xeons and AMD Shanghai Opterons,
> but not with Sandy Bridge and Bulldozer. With that said, I wouldn't
> necessarily say that AMD is getting their butt handed to them on a silver
> platter. From my experience AMD is very much in the game, performance-wise,
> and this article from Phoronix confirms:
>
>
> http://www.phoronix.com/scan.php?page=article&item=ubuntu_1204_virt&num=1
>
> From the article:
>
>      Between Intel's Sandy Bridge and AMD's Bulldozer for KVM
>      virtualization, the relative performance was generally quite close
>      between these competing latest-generation architectures. If looking
>      at the harmonic mean of the over three dozen tests that were run, the
>      Intel Core i7 3960X was running at 93% the speed of bare metal with
>      KVM while the AMD FX-8150 came in at 90% the speed of the bare metal
>      Bulldozer. Alternatively, with the geometric mean of all the results,
>      the i7-3960X was at 85% the speed of bare metal while the AMD FX-8150
>      was at 88%. VirtualBox on the FX-8150 was at 85% while the
>      problematic VirtualBox-on-Sandy-E was at 59%. Xen on Sandy-E came in
>      at 94%.
>
> They had issues with their ASUS motherboard, so AMD Xen benchmarks weren't
> taken.
>
> --
> . o .   o . o   . . o   o . .   . o .
> . . o   . o o   o . o   . o o   . . o
> o o o   . o .   . o o   o o .   o o o
>
>
> /*
> PLUG: http://plug.org, #utah on irc.freenode.net
> Unsubscribe: http://plug.org/mailman/options/plug
> Don't fear the penguin.
> */
>

Something else to remember is that the cpus to use should reflect what the
virtual machines are doing. I have been in a number of situations where I
wanted lower clock speed but more cores (read "Buy AMD") for some
applications and fewer cores that deliver better single core performance
(read "Buy Intel").

In the end, I normally want more cores when doing virtualization, otherwise
I would not generally be virtualizing...

Finally, remember to look up the right benchmarks. Aaron cited a great
article because it tested the actual capabilities of the CPUS in a
virtualization environment, unlike using a windows desktop benchmark, since
a windows desktop is not something you generally virtualize...