[Solved] Performance governor not active at boot.

AAVVIronAlex · 2024-06-30 09:48:29

Hello guys, I have recently found out that I started to have issues with power governors again.

This is what I get when I request to see the cpupower governor:

earthmapspictures@H-E-R-A-N:~$ cpupower frequency-info 
analyzing CPU 23:
  driver: intel_pstate
  CPUs which run at the same hardware frequency: 23
  CPUs which need to have their frequency coordinated by software: 23
  maximum transition latency:  Cannot determine or is not supported.
  hardware limits: 1.20 GHz - 4.60 GHz
  available cpufreq governors: performance powersave
  current policy: frequency should be within 1.20 GHz and 4.60 GHz.
                  The governor "powersave" may decide which speed to use
                  within this range.
  current CPU frequency: Unable to call hardware
  current CPU frequency: 1.20 GHz (asserted by call to kernel)
  boost state support:
    Supported: yes
    Active: yes

I then checked if I have the governor set in the file:

# Define CPUs governor
# valid governors: ondemand, performance, powersave, conservative, userspace.
GOVERNOR='performance'

My CPU (for some reason) does not have the ondemand governor:

earthmapspictures@H-E-R-A-N:~$ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors
performance powersave

Examining the problem further, lead me to see that the issue does not arise from the cpupower service:

earthmapspictures@H-E-R-A-N:~$ sudo systemctl status cpupower.service 
● cpupower.service - Apply cpupower configuration
     Loaded: loaded (/usr/lib/systemd/system/cpupower.service; enabled; preset:>
     Active: active (exited) since Sun 2024-06-30 13:31:39 +04; 9min ago
 Invocation: 61392ddd089648838a77204ed4cd5b48
    Process: 1012 ExecStart=/usr/lib/systemd/scripts/cpupower (code=exited, sta>
   Main PID: 1012 (code=exited, status=0/SUCCESS)

Jun 30 13:31:39 H-E-R-A-N systemd[1]: Starting Apply cpupower configuration...
Jun 30 13:31:39 H-E-R-A-N systemd[1]: Finished Apply cpupower configuration.

A few days ago I installed the KDE tool called powerdevil, that was set on balanced, but as there is no balanced power mode, it was apparently setting my system to powersave. I tried to toggle performance on the slider, but nothing really changed. So I resorted to removing the powerdevil package with pacman -Rdd, which will ignore dependencies. I restarted, and nothing changed.

I am thinking of setting the manual command (which does work and makes me think that there is something that KDE is doing that is making my CPU change the power governor) as an autorun command as a "last resort".

Any recommendations?

Thanks for reading, if you have any questions please tell me.

Last edited by AAVVIronAlex (2024-07-01 20:13:52)

Lone_Wolf · 2024-06-30 10:54:03

driver: intel_pstate

Intel's pstate makes heavy use of governors internal to the processor that require other methods to set, see https://wiki.archlinux.org/title/CPU_fr … y_boosting

What is the output of

$ cat /sys/devices/system/cpu/cpufreq/policy*/energy_performance_available_preferences

?

Also See https://wiki.archlinux.org/title/Power_ … e)_support

AAVVIronAlex · 2024-06-30 11:17:42

Lone_Wolf wrote:

Intel's pstate makes heavy use of governors internal to the processor that require other methods to set, see https://wiki.archlinux.org/title/CPU_fr … y_boosting

sudo echo 0 > /sys/devices/system/cpu/intel_pstate/no_turbo

This command (toggling turbo mode on (I do not think it was off to begin with)) returns a privilege issue, although I specified sudo.

earthmapspictures@H-E-R-A-N:~/.config/MangoHud$ cat /sys/devices/system/cpu/cpufreq/policy*/energy_performance_available_preferences
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power 
default performance balance_performance balance_power power

As for the Intel HWP, I am fairly new to that, can you elaborate on what I should do? The option there seems to only lead to me having more power saving, which is the opposite of what I want to achieve.

Last edited by AAVVIronAlex (2024-06-30 11:18:34)

justm3 · 2024-06-30 12:48:30

# Define CPUs governor
# valid governors: ondemand, performance, powersave, conservative, userspace.
GOVERNOR='performance'

De-capitalize GOVERNOR to governor

AAVVIronAlex · 2024-06-30 13:05:49

justm3 wrote:

De-capitalize GOVERNOR to governor

There are two reasons why I would not think that would solve my issues:

1. Because some threads on the internet specified that it is not case sensitive.

2. Because that was initially what I had written there.

I specifically want it to be performance mode every time I boot into my system. Even forced performance sits at lower CPU clocks, which is fine for me.

As said before, my last resort is autorunning that command everytime with system startup.

AAVVIronAlex · 2024-06-30 19:39:52

Lone_Wolf wrote:

Also See https://wiki.archlinux.org/title/Power_ … e)_support

Should I try using this method I found on a reddit thread? Creating a systemd service?

[Unit]
Description=Performance CPU frequency governor

[Service]
Type=oneshot
RemainAfterExit=true
ExecStart=/bin/sh -c '\
 echo performance | tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor'
ExecStop=/bin/sh -c '\
 echo powersave | tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor'

[Install]
WantedBy=multi-user.target

I also removed this package: power-profiles-daemon

Here is a link to the website:
https://www.reddit.com/r/Fedora/comment … t/garzthx/

Last edited by AAVVIronAlex (2024-06-30 19:56:10)

AAVVIronAlex · 2024-06-30 21:29:22

SOLUTION:

As I initially reckoned, the KDE power management service had something to do with the governor changing. In this case I was wrong when I deleted the powerdevil package, as that was the KDE shell "extension" for power management settings and etc. I had to remove the power-profiles-daemon which was recommended to me by the powerdevil notification on the KDE Task Manager. Deleting the manager and setting the GOVERNOR='performance' to governor='performance' fixed the issue, this also proves the Reddit thread wrong, I will make sure to point it out there too, just so it is no longer misleading to people who come across it.

At this point I cannot believe it works, but it does. Seems like whatever the power-profiles-daemon was doing was happening after systemd was loading the cpupower.service and the cpupower being set to performance. That was likely happening when KDE was turning on.

Anyway, thanks for the help.

Lone_Wolf · 2024-07-01 12:15:34

Great , once you are sure it's solved , please prepend [Solved] to the thread title (edit first post).

AAVVIronAlex · 2024-07-01 13:54:58

Lone_Wolf wrote:

Great , once you are sure it's solved , please prepend [Solved] to the thread title (edit first post).

There is one problem, I have to delete parts of the title.

Lone_Wolf · 2024-07-01 14:02:58

Yep, there's a character limit .

Maybe use something like [Solved] performance governor not active at boot ?

Arch Linux

#1 2024-06-30 09:48:29

[Solved] Performance governor not active at boot.

#2 2024-06-30 10:54:03

Re: [Solved] Performance governor not active at boot.

#3 2024-06-30 11:17:42

Re: [Solved] Performance governor not active at boot.

#4 2024-06-30 12:48:30

Re: [Solved] Performance governor not active at boot.

#5 2024-06-30 13:05:49

Re: [Solved] Performance governor not active at boot.

#6 2024-06-30 19:39:52

Re: [Solved] Performance governor not active at boot.

#7 2024-06-30 21:29:22

Re: [Solved] Performance governor not active at boot.

#8 2024-07-01 12:15:34

Re: [Solved] Performance governor not active at boot.

#9 2024-07-01 13:54:58

Re: [Solved] Performance governor not active at boot.

#10 2024-07-01 14:02:58

Re: [Solved] Performance governor not active at boot.

Board footer