Xen users - Aug 2009 - xen benchmark results on nehalem

Hi,

I made a few xen benchmark tests on Nehalems and Amd Opterons.
Results can be found here ;

http://wp.me/pBob2-B

best wishes,
Oz


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Opteron vs Nehalem tests can be found here; http://wp.me/pBob2-3


On Fri, Aug 14, 2009 at 10:49 PM, Ozgur Akan <ozgurakan@gmail.com> wrote:
> Hi,
>
> I made a few xen benchmark tests on Nehalems and Amd Opterons.
> Results can be found here ;
>
> http://wp.me/pBob2-B
>
> best wishes,
> Oz
>

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Ozgur Akan <ozgurakan@gmail.com> writes:
> Opteron vs Nehalem tests can be found here; http://wp.me/pBob2-3

I''ve found in most webapp workloads, the amount of ram matters more
than anything else; I mean, until you cache all your regularly accessed 
data, and my experience has been that most people tend to overbuy CPU
and skimp on the ram.  

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Sun, Aug 16, 2009 at 12:56 PM, Luke S Crawford <lsc@prgmr.com> wrote:
> Ozgur Akan <ozgurakan@gmail.com> writes:
>
> > Opteron vs Nehalem tests can be found here; http://wp.me/pBob2-3
>
>
> I''ve found in most webapp workloads, the amount of ram matters
more
> than anything else; I mean, until you cache all your regularly accessed
> data, and my experience has been that most people tend to overbuy CPU
> and skimp on the ram.
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
>
According to these tests having more ram was even slower. Anyone want to
comment as to why?


Grant McWilliams


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Grant McWilliams <grantmasterflash@gmail.com> writes:
> According to these tests having more ram was even slower. Anyone want to
> comment as to why?
Because unixbench does not mirror webapp workloads very well?  

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

It depends on what you are trying to achieve. If you want to support  
as many users as possible on a server then yes, ram is often the  
hotspot.

If, however, you have a slow application (which is sadly very common  
even today) and you want to optimize response time then straight line  
CPU power might be the variable you need to tweak. Many people don''t   
realize that a 2GHz core two duo can be faster than a 3.8 GHz P4 or  
that one 2GHz core two duo can be 25% faster than a 2.33 GHz core two  
duo.

Sent from my iPhone

On Aug 16, 2009, at 3:56 PM, Luke S Crawford <lsc@prgmr.com> wrote:
> Ozgur Akan <ozgurakan@gmail.com> writes:
>
>> Opteron vs Nehalem tests can be found here; http://wp.me/pBob2-3
>
>
> I''ve found in most webapp workloads, the amount of ram matters
more
> than anything else; I mean, until you cache all your regularly  
> accessed
> data, and my experience has been that most people tend to overbuy CPU
> and skimp on the ram.
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Sun, Aug 16, 2009 at 10:17 PM, Peter Booth <peter_booth@me.com> wrote:
> It depends on what you are trying to achieve. If you want to support as
> many users as possible on a server then yes, ram is often the hotspot.
>
> If, however, you have a slow application (which is sadly very common even
> today) and you want to optimize response time then straight line CPU power
> might be the variable you need to tweak. Many people don''t 
realize that a
> 2GHz core two duo can be faster than a 3.8 GHz P4 or that one 2GHz core two
> duo can be 25% faster than a 2.33 GHz core two duo.
>
> Sent from my iPhone
>
>
> On Aug 16, 2009, at 3:56 PM, Luke S Crawford <lsc@prgmr.com> wrote:
>
>  Ozgur Akan <ozgurakan@gmail.com> writes:
>>
>>  Opteron vs Nehalem tests can be found here; http://wp.me/pBob2-3
>>>
>>
>>
>> I''ve found in most webapp workloads, the amount of ram matters
more
>> than anything else; I mean, until you cache all your regularly accessed
>> data, and my experience has been that most people tend to overbuy CPU
>> and skimp on the ram.
>>
>>
But would having MORE ram every make a machine slower? I can understand that
it might not make it faster but hard to believe it''s going to slow down
the
machine!

Grant McWilliams.


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Peter Booth <peter_booth@me.com> writes:
> If, however, you have a slow application (which is sadly very common
> even today) and you want to optimize response time then straight line
> CPU power might be the variable you need to tweak. Many people
don''t
> realize that a 2GHz core two duo can be faster than a 3.8 GHz P4 or
> that one 2GHz core two duo can be 25% faster than a 2.33 GHz core two
> duo.
my experience has been that often people try to upgrade the CPU while the 
application is still touching disk to read data.  This is obviously
silly, because instead of being 25% faster or what have you, disk is millions
of times slower than ram, so you get this ''fall off a cliff''
effect whenever
you hit disk.


Sure, once your dataset is cached to ram, then yeah, you are right, 
upgrading the cpu and/or ram speed makes a lot of sense.   And there are
some workloads that work like that;  from what I understand most HPC
applications are designed with the knowledge that disk is so many orders
of magnitude slower than ram, so it doesn''t hit disk much, and you get
a
real boost out of upgrading cpu or ram speeds.

But most web applications (and *NIX applications in general)  are still
hitting disk, so it makes sense to have enough pagecache for all your 
commonly accessed data before you start worrying about cpu speed.  
As long as you are still falling off the cliff and hitting disk, a cpu
that is 2x or 3x faster isn''t going to make a bit of difference.  

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Grant McWilliams <grantmasterflash@gmail.com> writes:
> But would having MORE ram every make a machine slower? I can understand
that
> it might not make it faster but hard to believe it''s going to slow
down the
> machine!
Quite often, especially with registered ecc ddr2 and registered ecc ddr3
there is a trade off between the number of modules (which effects how much
ram you can have at a given price point)  and the speed of those modules.
I am given to understand that the monumental failure that was FBDIMM technology
mitigated that somewhat, of course at the cost of a ridiculous amount
of power.  

The dell reps trying to sell to one of my clients the other day were explaining
that the ram would only run at 1333mhz if we only filled a few of the slots.
If we filled all 18 slots, the ram would only run at 800mhz.  I tried to
explain to my clients that ram speed didn''t really matter for what they
were
doing, because they hadn''t gotten to the point yet where all their
frequently
used data was cached, and that their biggest performance win would be buying
as much 800mhz ram as they could afford.   

So there is a trade off.   Like I said, it''s nothing like the speed
difference
between slow ram and disk, but it is something to worry about if you are 
running HPC apps or other stuff that was designed so it doesn''t hit
disk
hardly at all.

If you care about memory speed, you might want to check out the STREAM 
benchmark.  See: http://www.cs.virginia.edu/stream/  


(And yea, I understand all the cs guys are freaking out ''cause ram
isn''t getting
faster at the speed CPUS are getting faster.   But in my world, that 
won''t matter at all until we are entirely off of ram, or at least until
SSDs finish crushing hard drives.)  


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users