[Bug 217931] amd-pstate lacks crucial features: CPU frequency and boost control

[bugzilla-daemon@kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=217931

--- Comment #60 from Artem S. Tashkinov (aros@gmx.com) ---
Under Linux 6.9.4 the amd-pstate driver doesn't adhere to the maximum set
frequency for my Ryzen 7 5800X:

# cpupower frequency-info
analyzing CPU 10:
  driver: amd-pstate-epp
  CPUs which run at the same hardware frequency: 10
  CPUs which need to have their frequency coordinated by software: 10
  maximum transition latency:  Cannot determine or is not supported.
  hardware limits: 550 MHz - 4.93 GHz
  available cpufreq governors: performance powersave
  current policy: frequency should be within 550 MHz and 1000 MHz.
                  The governor "powersave" may decide which speed to use
                  within this range.
  current CPU frequency: Unable to call hardware
  current CPU frequency: 3.54 GHz (asserted by call to kernel)
  boost state support:
    Supported: yes
    Active: yes
    AMD PSTATE Highest Performance: 166. Maximum Frequency: 4.93 GHz.
    AMD PSTATE Nominal Performance: 128. Nominal Frequency: 3.80 GHz.
    AMD PSTATE Lowest Non-linear Performance: 59. Lowest Non-linear Frequency:
1.75 GHz.
    AMD PSTATE Lowest Performance: 19. Lowest Frequency: 550 MHz.

# lscpu -ae
CPU SOCKET CORE L1d:L1i:L2:L3 ONLINE    MAXMHZ   MINMHZ       MHZ
  0      0    0 0:0:0:0          yes 4929.0000 550.0000 3124.8250
  1      0    1 1:1:1:0          yes 4929.0000 550.0000 3585.6809
  2      0    2 2:2:2:0          yes 4929.0000 550.0000  550.0000
  3      0    3 3:3:3:0          yes 4929.0000 550.0000 3554.8350
  4      0    4 4:4:4:0          yes 4929.0000 550.0000 3434.6189
  5      0    5 5:5:5:0          yes 4929.0000 550.0000 3589.3491
  6      0    6 6:6:6:0          yes 4929.0000 550.0000  550.0000
  7      0    7 7:7:7:0          yes 4929.0000 550.0000  550.0000
  8      0    0 0:0:0:0          yes 4929.0000 550.0000 3377.0410
  9      0    1 1:1:1:0          yes 4929.0000 550.0000 3545.8970
 10      0    2 2:2:2:0          yes 4929.0000 550.0000 3487.3730
 11      0    3 3:3:3:0          yes 4929.0000 550.0000 3185.1221
 12      0    4 4:4:4:0          yes 4929.0000 550.0000  550.0000
 13      0    5 5:5:5:0          yes 4929.0000 550.0000 4315.6221
 14      0    6 6:6:6:0          yes 4929.0000 550.0000 3148.2390
 15      0    7 7:7:7:0          yes 4929.0000 550.0000 3555.2141

---------------------------------------------------------------------------------------

And it's even more confusing since we now have:

1. * /sys/devices/system/cpu/cpufreq/policyX/scaling_available_governors
  performance powersave

2. How is it related to 

/sys/devices/system/cpu/cpufreq/policy0/energy_performance_preference which can
be of 
default performance balance_performance balance_power power?

3. And what's the "default" state?

---------------------------------------------------------------------------------------

It's all so simple with acpi-cpufreq :-(

1. I can enable/disable boost (not possible under amd-pstate on 6.9.4)
2. I can easily set maximum CPU core clock frequency (not possible for Ryzen 7
5800X under amd-pstate on 6.9.4)
3. I can set governors (performance powersave ondemand conservative)

With amd-pstate it's a huge maddening mess.

I'm going back to acpi-cpufreq until 6.10 at least for my desktop.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are the assignee for the bug.