Power Efficiency

Processor heat dissipation calculation based on cpu usage.

Dear forum,

I am working on Intel Xeon processor E5-2670 having 2.6GHZ Clock speed and no. of core 8. Its maximum TDP IS 115W. Currently cpu usage is 98%. based on this information, I have some queries as follows.

1. At the given cpu usage, what is the TDP of cpu?

 2. what is the maximum range of recommended TDP at actual operating condition.

3. is there any relation available between cpu usage and TDP?, if available please mention it.


Hello everyone,

it's surprising me that there are two different offsets for the MSR "PACKAGE_ENERGY_STATUS".

In "4th-gen-core-family-desktop-vol-2-datasheet.pdf" it says the offset should be 593Ch

In "xeon-e5-v3-datasheet-vol-2" it says the offset should be 0x90.

Both processors should be based on haswell. why there is a so big difference? 

And i can not find any information about this MSR in "64-ia-32-architectures-software-developer-manual-325462". Has not been updated?


Best Regards,


Power Efficiency tools to monitor VMs


I was looking for a way to monitor the power efficiency on a visualized server environment. I did try Power Gadget on a Windows server 2008 R2 but it didn't work. Is it a way to configure the tool in order to be able to get real consumption out of the host machine?

the main hypervisors I target are VMware ESX/ESXi and Microsoft Hyper-V 2008/2012.

Thanks in advance.

ACPI and Dynamic CPU Frequency


I have a basic question:

I'm using Linux (Ubuntu) and CPU I7.

The current CPU frequency is 1200hz, the maximum CPU frequency is 2100hz.

My Q is as follows:

Does Linux (or the INTEL CPU's)  increase the CPU frequency when the system is loaded ?

(i.e when the total CPU's are 70% loaded (for example) , Does the OS (without the user interference) may increase the CPU freq from 1200 to the max freq 2100, in order to decrease the load on the CPU's ?) (and without the user interference) ?


Actual CPU Frequency vs CPU Frequency Reported by the Linux "cpufreq" Subsystem

I performed an experiment that involves measuring instruction retirement rate and power consumption against reported CPU frequency and for on the P-state settings, the reported CPU frequency is not consistent with the results I am seeing.

Suscribirse a Power Efficiency