[SOLVED] My Ivy Bridge CPU isn't using the scaling governor set for Intel CPUs in rc.cpufreq.

Petri Kaukasoina · 03-18-2021, 04:28 PM

Newer kernels default to intel_cpufreq driver and schedutil governor, which should make a better job changing frequencies when needed. You get the old one (intel_pstate) back with 'echo active > /sys/devices/system/cpu/intel_pstate/status'. (And the new one with 'echo passive > /sys/devices/system/cpu/intel_pstate/status'.)

Code:

# cpufreq-info 
cpufrequtils 008: cpufreq-info (C) Dominik Brodowski 2004-2009
Report errors and bugs to cpufreq@vger.kernel.org, please.
analyzing CPU 0:
  driver: intel_cpufreq
  CPUs which run at the same hardware frequency: 0
  CPUs which need to have their frequency coordinated by software: 0
  maximum transition latency: 20.0 us.
  hardware limits: 1.60 GHz - 3.40 GHz
  available cpufreq governors: conservative, ondemand, userspace, powersave, performance, schedutil
  current policy: frequency should be within 1.60 GHz and 3.40 GHz.
                  The governor "schedutil" may decide which speed to use
                  within this range.
  current CPU frequency is 2.04 GHz.

Code:

# echo active > /sys/devices/system/cpu/intel_pstate/status
# cpufreq-info 
cpufrequtils 008: cpufreq-info (C) Dominik Brodowski 2004-2009
Report errors and bugs to cpufreq@vger.kernel.org, please.
analyzing CPU 0:
  driver: intel_pstate
  CPUs which run at the same hardware frequency: 0
  CPUs which need to have their frequency coordinated by software: 0
  maximum transition latency: 4294.55 ms.
  hardware limits: 1.60 GHz - 3.40 GHz
  available cpufreq governors: performance, powersave
  current policy: frequency should be within 1.60 GHz and 3.40 GHz.
                  The governor "powersave" may decide which speed to use
                  within this range.
  current CPU frequency is 1.60 GHz.

This is Sandy Bridge: Core i5-2400.

RadicalDreamer · 03-18-2021, 11:34 PM

Quote:

Originally Posted by Petri Kaukasoina

Newer kernels default to intel_cpufreq driver and schedutil governor, which should make a better job changing frequencies when needed. You get the old one (intel_pstate) back with 'echo active > /sys/devices/system/cpu/intel_pstate/status'. (And the new one with 'echo passive > /sys/devices/system/cpu/intel_pstate/status'.)

Thanks, I didn't know that intel_pstate was still around. I've been using intel_cpufreq just fine without knowing it. It is defaulting to that. Which driver and cpu frequency governer would you prefer to be set in rc.cpufreq for Intel CPUs as default?

Petri Kaukasoina · 03-19-2021, 04:03 AM

Quote:

Originally Posted by RadicalDreamer

Which driver and cpu frequency governer would you prefer to be set in rc.cpufreq for Intel CPUs as default?

I have 'chmod -x rc.cpufreq', and so I use the kernel defaults: intel_cpufreq and schedutil governor. It has a good performance and low energy consumption. (I use kernel 5.11.) I just tried a kernel compile, and intel_cpufreq+schedutil was 1% faster than intel_pstate+performance.

RadicalDreamer · 03-19-2021, 10:02 AM

intel_cpufreq and schedutil works great. I didn't even have to up my min freq percentage at all from default because schedutil is so responsive. Schedutil is very responsive and I think Final Fantasy XV works better with it. It is much better than powersave and performance. It doesn't require tweaking my min freq like powersave and it isn't insanely high like performance (3+ Ghz while doing nothing).

Since rc.cpufreq sets ondemand by default I suggest modifying the code in rc.cpufreq to this for applicable Intel CPUs:

Code:

# For CPUs using intel_cpufreq, always use the schedutil governor. This also
# provides power savings on Intel processors while avoiding the ramp-up lag
# present when using the powersave governor (which is the default if ondemand
# is requested on these machines):
if [ "$(cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_driver 2> /dev/null)" =
 "intel_cpufreq" ]; then
  SCALING_GOVERNOR="schedutil"

RadicalDreamer · 03-19-2021, 11:44 AM

I was going to set a relative's Skylake 6700K CPU to intel_cpufreq but it is using intel_pstate to my surprise. So maybe that is why no has complained. Why would a much newer processor be using intel_pstate instead? The driver was never changed by its user on either computer. Same 5.10.XX kernel being used. I'd still love to see intel_cpufreq added to rc.cpufreq though.

Petri Kaukasoina · 03-19-2021, 12:14 PM

linux-5.10.24/Documentation/admin-guide/pm/intel_pstate.rst

Code:

Active Mode
-----------

This is the default operation mode of ``intel_pstate`` for processors with
hardware-managed P-states (HWP) support.  If it works in this mode, the
``scaling_driver`` policy attribute in ``sysfs`` for all ``CPUFreq`` policies
contains the string "intel_pstate"...

Passive Mode
------------

This is the default operation mode of ``intel_pstate`` for processors without
hardware-managed P-states (HWP) support...
If the driver works in this mode, the ``scaling_driver`` policy attribute in
``sysfs`` for all ``CPUFreq`` policies contains the string "intel_cpufreq".
Then, the driver behaves like a regular ``CPUFreq`` scaling driver...

Google search tells "What is HWP? HWP is a technology introduced in Skylake which lets the CPU select its own stepping speed without the usage of the CPU Multiplier. Additionally it trottles/boosts itself much faster, which improoves overall CPU performance."

RadicalDreamer · 03-19-2021, 02:56 PM

Quote:

Originally Posted by Petri Kaukasoina

linux-5.10.24/Documentation/admin-guide/pm/intel_pstate.rst

Code:

Active Mode
-----------

This is the default operation mode of ``intel_pstate`` for processors with
hardware-managed P-states (HWP) support.  If it works in this mode, the
``scaling_driver`` policy attribute in ``sysfs`` for all ``CPUFreq`` policies
contains the string "intel_pstate"...

Passive Mode
------------

This is the default operation mode of ``intel_pstate`` for processors without
hardware-managed P-states (HWP) support...
If the driver works in this mode, the ``scaling_driver`` policy attribute in
``sysfs`` for all ``CPUFreq`` policies contains the string "intel_cpufreq".
Then, the driver behaves like a regular ``CPUFreq`` scaling driver...

Google search tells "What is HWP? HWP is a technology introduced in Skylake which lets the CPU select its own stepping speed without the usage of the CPU Multiplier. Additionally it trottles/boosts itself much faster, which improoves overall CPU performance."

Thanks for the information! I haven't spent much time with the computer with the Skylake processor recently but with powersave the CPU was too slow without upping the minimum freq. The computer has poor cooling and setting the governor to performance had it running almost to 4 GHz,raising the temperature to ~65 Celsius. I don't believe a word of that Intel marketing BS.

walecha · 03-19-2021, 09:38 PM

I'm using schedutil on haswell and ryzen. Schedutil should be default on install if using kernel 5.10+.