6
  • Which core's frequency and temperature are shown by the Intel Power Gadget tool?

  • How can I observe the frequency of all the independent cores on my MacBook?

  • Can the Intel Power Gadget be configured to do so? (I was not able to find the required setting/extension)

There's already a discussion on observing CPU frequencies using the Terminal here:

However, I cannot draw any conclusion/method of achieving this from the above. An equivalent for cpu-frequtils would have been nice to have on macOS.

1 Answer 1

8

I'm the owner of Intel Power Gadget on macOS.

Which core's frequency and temperature are shown by the Intel Power Gadget tool?

On any Intel processor prior to Icelake, all cores run at the same frequency. In the current version of Power Gadget, frequency is read from a package scope register, so it's not reading a specific core's frequency. A future version of Power Gadget will measure frequency on all cores.

At present Power Gadget measures temperature at the package scope, although there are core scoped temperature readings provided by the hardware. I haven't seen any particular need for core scope temperature readings, but let me know if you have a justification for that.

How can I observe the frequency of all the independent cores on my MacBook?

Because you are running on a pre-Icelake system where all cores are running at the same frequency, there is no need to observe the frequency of cores independently.

Can the Intel Power Gadget be configured to do so?

In the current and older versions of Power Gadget there is no mechanism for reading frequency on all cores, even if you were to use the public API. However, a future version of Power Gadget will provide this capability.

3
  • On my Linux machine, cpu-frequtils reported different frequencies on different cores for a 7th Gen Intel Core i5 processor. As far as I remember, the frequencies did not differ by a huge margin. If the processor is configured as you mentioned, shouldn't the frequencies be exactly the same? Commented Jul 16, 2019 at 5:47
  • 1
    It would just depend on how you measure frequency. If you read the instantaneous frequency on all cores at one time, then it would identical on all cores. But the frequency is constantly changing, so if you read at different times, or if you use counters to measure frequency over time, then it will be a bit different on different cores. Commented Jul 16, 2019 at 15:31
  • No, the frequencies definitely vary across cores, even at the same time. Dynamic acceleration dates back to around 2007, when a similar feature named IDA was introduced to some Core 2 Duo CPUs, and when IDA is active there is one core that runs sustainedly at 0.5x or 1x higher than the other(s). One can observe that by keeping a single-thread CPU-bound program such as SuperPI running, and look at HWiNFO summary at the meantime. Yet it may be right to say those CPUs in the C0 state always run at identical frequencies... I will check that when I get some idle time.
    – Fred Qian
    Commented Oct 3, 2020 at 8:32

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .