Virtualization Performance: Intel vs. AMD

Virtualization Performance: Intel vs. AMD

[Andreas Winkelbauer]

Hi,

I am currently in the process of setting up the configuration for some new
workstations. I don't yet know if I'll use an Intel or an AMD plattform.

Now my question is: Are there any important differences in terms of
virtualization performance and/or features between the current Intel/AMD CPUs
(e.g. Core i7 "Lynnfield" and Phenom II X4 "Deneb")?

Best regards,
Andreas

RE: Virtualization Performance: Intel vs. AMD

[Neil Aggarwal]

Andreas:

> Now my question is: Are there any important differences in terms of
> virtualization performance and/or features between the 
> current Intel/AMD CPUs

I prefer AMD CPUs, they give you a better bang for the buck.
Besides that, I don't think they would be any technical
differences, they are supposed to be completely compatible.
I have seen no evidence to the contrary.

	Neil

--
Neil Aggarwal, (281)846-8957, http://UnmeteredVPS.net
CentOS 5.4 VPS with unmetered bandwidth only $25/month!
7 day no risk trial, Google Checkout accepted 

Re: Virtualization Performance: Intel vs. AMD

[Thomas Treutner]

On Sunday 15 November 2009 14:05:52 Neil Aggarwal wrote:
> I prefer AMD CPUs, they give you a better bang for the buck.
> Besides that, I don't think they would be any technical
> differences, they are supposed to be completely compatible.
> I have seen no evidence to the contrary.

Isn't AMD the only one who has hardware support for nested virtualization? Or 
isn't that true any longer? Anyways, I'm just curious, as this feature is 
primarily interesting for development, IMHO.


-t

Re: Virtualization Performance: Intel vs. AMD

[Avi Kivity]

On 11/15/2009 05:55 PM, Thomas Treutner wrote:
> On Sunday 15 November 2009 14:05:52 Neil Aggarwal wrote:
>    
>> I prefer AMD CPUs, they give you a better bang for the buck.
>> Besides that, I don't think they would be any technical
>> differences, they are supposed to be completely compatible.
>> I have seen no evidence to the contrary.
>>      
> Isn't AMD the only one who has hardware support for nested virtualization? Or
> isn't that true any longer?

No, the Core i7 has ept which is the Intel equivalent.

> Anyways, I'm just curious, as this feature is
> primarily interesting for development, IMHO.
>    

No, it's primarily interesting for performance.

-- 
error compiling committee.c: too many arguments to function

Re: Virtualization Performance: Intel vs. AMD

[Thomas Treutner]

On Monday 16 November 2009 11:12:19 Avi Kivity wrote:
> > Anyways, I'm just curious, as this feature is
> > primarily interesting for development, IMHO.
>
> No, it's primarily interesting for performance.

I think I confused NPT with support for nested virtualization (which I think 
no one except devs would have a use case for). Thanks for the pointer.


-t

Re: Virtualization Performance: Intel vs. AMD

[Thomas Treutner]

On Sunday 15 November 2009 13:22:36 Andreas Winkelbauer wrote:
> Now my question is: Are there any important differences in terms of
> virtualization performance and/or features between the current Intel/AMD
> CPUs (e.g. Core i7 "Lynnfield" and Phenom II X4 "Deneb")?

The Core i7 has hyperthreading, so you see 8 logical CPUs. But don't ask me if 
that's really an advantage for doing virtualization ;-)


-t

RE: Virtualization Performance: Intel vs. AMD

[Neil Aggarwal]

> The Core i7 has hyperthreading, so you see 8 logical CPUs. 

Are you saying the AMD processors do not have hyperthreading?

I have a machine with two six-core AMD Opterons.
top shows me 12 logical CPUs.

	Neil


--
Neil Aggarwal, (281)846-8957, http://UnmeteredVPS.net
CentOS 5.4 VPS with unmetered bandwidth only $25/month!
7 day no risk trial, Google Checkout accepted 

Re: Virtualization Performance: Intel vs. AMD

[Thomas Fjellstrom]

On Sun November 15 2009, Neil Aggarwal wrote:
> > The Core i7 has hyperthreading, so you see 8 logical CPUs.
> 
> Are you saying the AMD processors do not have hyperthreading?

Course not. Hyperthreading is dubious at best.

> I have a machine with two six-core AMD Opterons.
> top shows me 12 logical CPUs.

If it had Hyperthreading, you'd see 24 logical cpus.
6 + 6 == 12 * 2(ht) == 24.

Those six cores in each cpu are actual physcial cores. Not fake logical 
cores.

> 	Neil
> 
> 
> --
> Neil Aggarwal, (281)846-8957, http://UnmeteredVPS.net
> CentOS 5.4 VPS with unmetered bandwidth only $25/month!
> 7 day no risk trial, Google Checkout accepted
> 
> --
> To unsubscribe from this list: send the line "unsubscribe kvm" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 


-- 
Thomas Fjellstrom
tfjellstrom@shaw.ca

RE: Virtualization Performance: Intel vs. AMD

[Neil Aggarwal]

> Those six cores in each cpu are actual physcial cores. Not 
> fake logical 
> cores.

OK, I see what you are saying now.
Thanks for the clarification.

	Neil

--
Neil Aggarwal, (281)846-8957, http://UnmeteredVPS.net
CentOS 5.4 VPS with unmetered bandwidth only $25/month!
7 day no risk trial, Google Checkout accepted 

Re: Virtualization Performance: Intel vs. AMD

[Gordan Bobic]

Thomas Fjellstrom wrote:
> On Sun November 15 2009, Neil Aggarwal wrote:
>>> The Core i7 has hyperthreading, so you see 8 logical CPUs.
>> Are you saying the AMD processors do not have hyperthreading?
> 
> Course not. Hyperthreading is dubious at best.

That's a rather questionable answer to a rather broad issue. SMT is 
useful, especially on processors with deep pipelines (think Pentium 4 - 
and in general, deeper pipelines tend to be required for higher clock 
speeds), because it reduces the number of context switches. Context 
switches are certainly one of the most expensive operations if not the 
most expensive operation you can do on a processor, and typically 
requires flushing the pipelines. Double the number of hardware threads, 
and you halve the number of context switches.

This typically isn't useful if your CPU is processing one 
single-threaded application 99% of the time, but on a loaded server it 
can make a significant difference to throughput.

Gordan

Re: Virtualization Performance: Intel vs. AMD

[Thomas Fjellstrom]

On Sun November 15 2009, Gordan Bobic wrote:
> Thomas Fjellstrom wrote:
> > On Sun November 15 2009, Neil Aggarwal wrote:
> >>> The Core i7 has hyperthreading, so you see 8 logical CPUs.
> >>
> >> Are you saying the AMD processors do not have hyperthreading?
> >
> > Course not. Hyperthreading is dubious at best.
> 
> That's a rather questionable answer to a rather broad issue. SMT is
> useful, especially on processors with deep pipelines (think Pentium 4 -
> and in general, deeper pipelines tend to be required for higher clock
> speeds), because it reduces the number of context switches. Context
> switches are certainly one of the most expensive operations if not the
> most expensive operation you can do on a processor, and typically
> requires flushing the pipelines. Double the number of hardware threads,
> and you halve the number of context switches.

Hardware context switches aren't free either. And while it really has 
nothing to do with this discussion, the P4 arch was far from perfect (many 
would say, far from GOOD).

> This typically isn't useful if your CPU is processing one
> single-threaded application 99% of the time, but on a loaded server it
> can make a significant difference to throughput.

I'll buy that. Though you'll have to agree that the initial Hyperthread 
implementation in intel cpus was really bad. I hear good things about the 
latest version though.

But hey, if you can stick more cores in, or do what AMD is doing with its 
upcoming line, why not do that? Hyperthreading seems like more of a gimmick 
than anything. What seems to help the most with the new Intel arch is the 
auto overclocking when some cores are idle. Far more of a performance 
improvement than Hyperthreading will ever be it seems.

But maybe that's just me.

> Gordan
> --
> To unsubscribe from this list: send the line "unsubscribe kvm" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 


-- 
Thomas Fjellstrom
tfjellstrom@shaw.ca

Re: Virtualization Performance: Intel vs. AMD

[Gordan Bobic]

Thomas Fjellstrom wrote:

>>>>> The Core i7 has hyperthreading, so you see 8 logical CPUs.
>>>> Are you saying the AMD processors do not have hyperthreading?
>>> Course not. Hyperthreading is dubious at best.
>> That's a rather questionable answer to a rather broad issue. SMT is
>> useful, especially on processors with deep pipelines (think Pentium 4 -
>> and in general, deeper pipelines tend to be required for higher clock
>> speeds), because it reduces the number of context switches. Context
>> switches are certainly one of the most expensive operations if not the
>> most expensive operation you can do on a processor, and typically
>> requires flushing the pipelines. Double the number of hardware threads,
>> and you halve the number of context switches.
> 
> Hardware context switches aren't free either. And while it really has 
> nothing to do with this discussion, the P4 arch was far from perfect (many 
> would say, far from GOOD).

I actually disagree with a lot of criticism of P4. The reason why it's 
performance _appeared_ to be poor was because it was more reliant on 
compilers doing their job well. Unfortunately, most compilers generate 
very poor code, and most programmers aren't even aware of the 
improvements that can be had in this area with a bit of extra work and a 
decent compiler. Performance differences of 7+ times (700%) aren't 
unheard of on Pentium 4 between, say, ICC and GCC generated code.

P4 wasn't a bad design - the compilers just weren't good enough to 
leverage it to anywhere near it's potential.

>> This typically isn't useful if your CPU is processing one
>> single-threaded application 99% of the time, but on a loaded server it
>> can make a significant difference to throughput.
> 
> I'll buy that. Though you'll have to agree that the initial Hyperthread 
> implementation in intel cpus was really bad. I hear good things about the 
> latest version though.

As measured by what? A single-threaded desktop benchmark?

> But hey, if you can stick more cores in, or do what AMD is doing with its 
> upcoming line, why not do that? Hyperthreading seems like more of a gimmick 
> than anything.

If there weren't clear and quantifiable benefits then IBM wouldn't be 
putting it in it's Power series of high end processors, it wouldn't be 
in the X-Box 360's Xenon (PPC970 variant), and Sun wouldn't be going 
massively SMT in the Niagara SPARCs. Silicon die space is _expensive_ - 
it wouldn't be getting wasted on gimmicks.

> What seems to help the most with the new Intel arch is the 
> auto overclocking when some cores are idle. Far more of a performance 
> improvement than Hyperthreading will ever be it seems.

Which is targeted at gamers and desktop enthusiasts who think that FPS 
in Crysis is a meaningful measure of performance for most applications. 
Server load profile is a whole different ball game.

Anyway, let's get this back on topic for the list before we get told off 
(of course, I'm more than happy to continue the discussion off list).

Gordan

Re: Virtualization Performance: Intel vs. AMD

[Andi Kleen]

Thomas Fjellstrom <tfjellstrom@shaw.ca> writes:

> Hardware context switches aren't free either. 

FWIW, SMT has no "hardware context switches", the 'S' stands for
simultaneous: the operations from the different threads are travelling
simultaneously through the CPU's pipeline.

You seem to confuse it with 'CMT' (Coarse-grained Multi Threading),
which has context switches.

-Andi

Re: Virtualization Performance: Intel vs. AMD

[Avi Kivity]

On 11/16/2009 12:29 AM, Gordan Bobic wrote:
> Thomas Fjellstrom wrote:
>> On Sun November 15 2009, Neil Aggarwal wrote:
>>>> The Core i7 has hyperthreading, so you see 8 logical CPUs.
>>> Are you saying the AMD processors do not have hyperthreading?
>>
>> Course not. Hyperthreading is dubious at best.
>
> That's a rather questionable answer to a rather broad issue. SMT is 
> useful, especially on processors with deep pipelines (think Pentium 4 
> - and in general, deeper pipelines tend to be required for higher 
> clock speeds), because it reduces the number of context switches. 
> Context switches are certainly one of the most expensive operations if 
> not the most expensive operation you can do on a processor, and 
> typically requires flushing the pipelines. Double the number of 
> hardware threads, and you halve the number of context switches.
>

The real win is in parallelizing memory access.  If a cache miss costs 
200 cycles, no amount of pipelining and out-of-order execution will hide 
this cost.  Running two threads in parallel will at best hide the cost 
by letting another thread execute, or at least issue two memory accesses 
in parallel instead of just one.

> This typically isn't useful if your CPU is processing one 
> single-threaded application 99% of the time, but on a loaded server it 
> can make a significant difference to throughput.

If you are able to saturate the multiple threads (typically easier with 
many small guests rather than a few large ones) then hyperthreading is 
likely a win.

-- 
error compiling committee.c: too many arguments to function