E3-1230 v3 cpu MHz

nxgweb

I ordered an E3-1230 v3 server today. When I run /proc/cpuinfo result show
cpu MHz : 800.000 is this correct??

[~]# cat /proc/cpuinfo

processor : 0

vendor_id : GenuineIntel

cpu family : 6

model : 60

model name : Intel(R) Xeon(R) CPU E3-1230 v3 @ 3.30GHz

stepping : 3

cpu MHz : 800.000

cache size : 8192 KB

physical id : 0

siblings : 8

core id : 0

cpu cores : 4

apicid : 0

initial apicid : 0

fpu : yes

fpu_exception : yes

cpuid level : 13

wp : yes

Awmusic12635

That is caused by hyperthreading most likely, either that or power saving mode

support123

This is because Intel SpeedStep is enabled on your server. Run this in SSH to disable it:

service cpuspeed stop

chkconfig cpuspeed off

dcc

Why would you want to disable speedstep?

skagerrak

@dcc said:
Why would you want to disable speedstep?

Most propably because he doesn't pay for the Watts consumed and wants that CPU to run on full-hertz all the time without any reason.

jar

@dcc said:
Why would you want to disable speedstep?

Clients might think they've been throttled, perception has a high value in dollars in the hosting industry.

Makkesk8

disabling speedstep will decrease the lifetime of the cpu, keep that in mind.

dcc

@jarland said:
Clients might think they've been throttled, perception has a high value in dollars in the hosting industry.

We were dealing with tons of similar tickets until we disabled cpuspeed on most nodes... By the way, I did not observe noticeable changes in power usage.

I am hoping that one day we will get to the "dd benchmark" myths... but that one will be much more difficult to debunk as things like iops and actual uptime are harder to understand, especially when "everyone else" seems to only care about dd...

dcc

@Makkesk8 said:
disabling speedstep will decrease the lifetime of the cpu, keep that in mind.

To be honest, I am yet to see a single CPU that failed because it worked "too hard".

Makkesk8

@dcc said:
To be honest, I am yet to see a single CPU that failed because it worked "too hard".

It's rather obvious really and there's been tests done.

Magiobiwan

I can't think of any real good reason to disable SpeedStep other than making it appear your CPU is always at full speed.

rapidnode

Yup that is speedstep / power saving. It should clock up when needed, I would leave it like so if it doesn't cause problems.

eLohkCalb

rapidnode said: It should clock up when needed

I have seen many cases where the frequency failed to step-up/down on virtualization servers, so I usually disable it. On laptop I do leave it on though.

Makkesk8 said: It's rather obvious really and there's been tests done.

Do you have the link to the test(s)?

Maounique

Disabling speedstep does not mean the cpu will use the full capacity all the time and spend lots of watts. You need to give it something to do for that, spending more cycles in idle state will increase a bit the power consumption compared to lowering the clock by speedstep, but not much. For a server, it is a small price to pay to avoid potential problems with speedstep failing on you.

Ruchirablog

@Maounique is right. Having speedstep turned on a server is bs plus the customer complaints.

If you have a 1000W power supply in your desktop it doesn't mean the power consumption will always be 1000 watts.

nxgweb

Thanks for the replies. I have other E3 servers with speedstep enabled. But cpu MHz not reduced at that much on all cores. First time with E3 v3 so little confused about 800 MHz on every cores.

sc754

I find the best way to disable power saving is to recompile your kernel with it disabled.

forthcloud

@sc754 said:
I find the best way to disable power saving is to recompile your kernel with it disabled.

You don't have to recompile the kernel to disable speedstep.

SpeedBus

You can change it to performance mode which will raise the CPU frequency to it's base frequency (will use more power though).

skagerrak

@Ruchirablog said:
Maounique is right. Having speedstep turned on a server is bs plus the customer complaints.

If you have a 1000W power supply in your desktop it doesn't mean the power consumption will always be 1000 watts.

I agree on the result of your answer. But that comparison is non-sense. Intel achieves the EIST by lowering the voltage of the core (C1E). And we all know that V * A = W. A power supply is working differently.

Ruchirablog

@skagerrak said:
I agree on the result of your answer. But that comparison is non-sense. Intel achieves the EIST by lowering the voltage of the core (C1E). And we all know that V * A = W. A power supply is working differently.

wrong. Speedstep works by reducing the clock speed not the voltage of the core. Lower voltage requirement is a result of lower clock. not other way around.

overclockers increase or decrease the cpu voltage to compensate for the higher or lower clocks. If processors work as you said increasing the voltage of the CPU will make the clock speed higher. It doesn't work that way

skagerrak

@Ruchirablog said:
Lower voltage requirement is a result of lower clock. not other way around.

Which is the main-reason for adjusting the frequency. The whole story behing EIST (so called "speedstep") and C1E is how you could possible achieve a lower consumption and therefore power efficiency.

EIST:

— If the target frequency is higher than the current frequency, V(_cc) is ramped up in steps to an optimized voltage. This voltage is signaled by the SVID bus to the voltage regulator. Once the voltage is established, the PLL locks on to the target frequency.
— If the target frequency is lower than the current frequency, the PLL locks to the target frequency, then transitions to a lower voltage by signaling the target voltage on SVID bus.
— The processor controls voltage ramp rates internally to ensure glitch-free transitions.

C1E:

When the processor is idle, low-power idle states (C-states) are used to save power.
Caution: Long term reliability cannot be assured unless all the Low Power Idle States are enabled.
A package C-state is automatically resolved by the processor depending on the core idle power states and the status of the platform components.
— Each core can be at a lower idle power state than the package if the platform does not grant the processor permission to enter a requested package C-state.
— The platform may allow additional power savings to be realized in the processor.
— For package C-states, the processor is not required to enter C0 before entering any other C-state.

Shoaib_A

Anyone know how to disable Powernow which is AMD equivalent of Intel's Speedstep?

Oliver

Hmm, interesting... I am about to take delivery of a few servers with these CPUs for OpenVZ nodes. I look forward to the tickets querying the CPU speed.

SpeedBus

Easy to change to performance only mode, http://murty4all.blogspot.in/2010/11/changing-cpu-frequency-scaling-governor.html