[已解决] Intel lunar lake Ultra7 258V 使用 balance 能源模式卡顿

技术交流与探讨 内核与硬件
1

我正在使用 Thinkpad X1 Carbon gen 13, 我的CPU是 Intel Lunar lake Ultra7 258V. 我的内核版本是 6.14.1-arch1-1, 桌面环境是 KDE. 我安装了 power-profiles-daemon 进行调频. 我发现如果我把 Power profile 设置成 balance 后不久, 我的系统就会卡顿. 我发现 CPU 频率一直保持在很低的范围 (大概 400MHz 左右).

$ cpupower frequency-info
-------------------------------------------------------------------------
analyzing CPU 3:
  driver: intel_pstate
  CPUs which run at the same hardware frequency: 3
  CPUs which need to have their frequency coordinated by software: 3
  energy performance preference: balance_power
  hardware limits: 400 MHz - 4.80 GHz
  available cpufreq governors: performance powersave
  current policy: frequency should be within 400 MHz and 4.80 GHz.
                  The governor "powersave" may decide which speed to use
                  within this range.
  current CPU frequency: 400 MHz (asserted by call to kernel)
  boost state support:
    Supported: yes
    Active: yes

然而在 power save 或者 performance 模式下我的电脑正常工作.

更新一下, 使用 thermald 可能不太被推荐 (见这个Github Issue), 这个问题更新 BIOS 到最新版本就解决了.

2

那就用 powersave 啊。「balance」是用的哪个 driver 呢？

3

确实我几乎一直在用 powersave, 最近才发现有这个问题. balanced 是我在 KDE 下设置的能源模式, 我想他应该指的是

  energy performance preference: balance_power

这一行, driver 应该一直是 intel_pstate.

我想这应该对应于 https://wiki.archlinux.org/title/CPU_frequency_scaling#Intel_performance_and_energy_bias_hint 在 power save 模式下有

$ cat /sys/devices/system/cpu/cpu0/power/energy_perf_bias
15

在 balanced 模式下有

$ cat /sys/devices/system/cpu/cpu0/power/energy_perf_bias
8
4

我想起来以前装 TLP 的时候感觉很卡顿, 就换成了 power-profiles-daemon, 现在看来不是 TLP 的问题 ,

5

好奇怪这好像又不是 Intel performance and energy bias hint, 我参照 https://wiki.archlinux.org/title/CPU_frequency_scaling#Intel_performance_and_energy_bias_hint 的三个方法设置 EPB 值, 发现虽然 /sys/devices/system/cpu/cpu*/power/energy_perf_bias 的值变化了, 但是卡顿问题没有出现, 而且 cpupower frequency-info 中 energy performance preference 也没有从 power 变成 balance_power. 懵, 不知道 power-profiles-daemon 调的到底是什么.

另外如果我把 power-profiles-daemon 卸载且把服务停止, 那么卡顿问题就会出现, 看起来好像也不是 power-profiles-daemon 的 bug?

6

https://wiki.archlinux.org/title/CPU_frequency_scaling#Setting_the_EPP

7

参考 https://unix.stackexchange.com/a/664093, 我把 Windows 能源模式调整为了最佳性能 (电池和电源供电都设置了). 目前看来问题好像已经被解决了, 太神秘了, 我不知道发生了什么 :o

对不起看起来不是这样, 坚持了一小会后 CPU 又开始限制在很低的频率了 :frowning:

8

试试 intel_pstate=passive 这个 kernel parameter？感觉可能是有啥 bug，lunar lake 还是太新了

9

我尝试了这个, 但是问题还在 (sad) :frowning:

$ cpupower frequency-info
----------------------------------------------
analyzing CPU 6:
  driver: intel_cpufreq
  CPUs which run at the same hardware frequency: 6
  CPUs which need to have their frequency coordinated by software: 6
  energy performance preference: balance_performance
  hardware limits: 400 MHz - 3.70 GHz
  available cpufreq governors: conservative ondemand userspace powersave performance schedutil
  current policy: frequency should be within 400 MHz and 3.70 GHz.
                  The governor "schedutil" may decide which speed to use
                  within this range.
  current CPU frequency: 400 MHz (asserted by call to kernel)
  boost state support:
    Supported: yes
    Active: yes

调整到 balanced 后没多久 (大概半分钟?) CPU 频率就被限制到 400MHz 左右. 不管是电源供电还是电池供电都是如此.

我怀疑是 Thinkpad 的问题 (例如 https://wiki.archlinux.org/title/Lenovo_ThinkPad_T480#CPU_stuck_at_minimum_frequency), 但是我现在没有时间调试这个, 毕竟 powersave 已经够用了 :expressionless:

10

我很久没用过 intel 的 cpu 的机器了，看起来是

  energy performance preference: balance_performance

这一行会影响 cpu 的频率？这个就是 power-profiles-daemon 和 tlp 控制的部分（我记得 tlp 似乎有配置这个相关的选项）？不过我这边 amd 的 cpu 上的 cpupower frequency-info 并没有这一行。。

11

太神秘了, 我刚刚发现只要我进行如下操作:

  1. 启动 Windows 并在电源设置里更改能源模式 (如平衡 -> 最佳能效).
  2. 重启进入 Archlinux.

那么在Archlinux下第一次设置 balanced 模式不会出现卡顿(通过KDE的托盘面板). 但是如果切换了模式 (如 balanced -> performance -> balanced), 那么过了大概半分钟就会出现同样的问题 :frowning:

analyzing CPU 1:
  driver: intel_pstate
  CPUs which run at the same hardware frequency: 1
  CPUs which need to have their frequency coordinated by software: 1
  energy performance preference: balance_performance
  hardware limits: 400 MHz - 4.70 GHz
  available cpufreq governors: performance powersave
  current policy: frequency should be within 400 MHz and 4.70 GHz.
                  The governor "powersave" may decide which speed to use
                  within this range.
  current CPU frequency: 1.07 GHz (asserted by call to kernel)
  boost state support:
    Supported: yes
    Active: yes

(可以看到 energy performance preference: balance_performance 而且 CPU 频率正常)

我完全晕过去了qwq

12

贴一个 balanced 下 turbostat 的输出不知道有没有帮助 (

# turbostat
--------------------
turbostat version 2025.02.02 - Len Brown <lenb@kernel.org>
Kernel command line: BOOT_IMAGE=/@/boot/vmlinuz-linux root=UUID=**** rw rootflags=subvol=@ loglevel=3 quiet
CPUID(0): GenuineIntel 0x23 CPUID levels
CPUID(1): family:model:stepping 0x6:bd:1 (6:189:1) microcode 0x116
CPUID(0x80000000): max_extended_levels: 0x80000008
CPUID(1): SSE3 MONITOR SMX EIST TM2 TSC MSR ACPI-TM HT TM
CPUID(6): APERF, TURBO, DTS, PTM, HWP, HWPnotify, HWPwindow, HWPepp, HWPpkg, EPB
cpu0: MSR_IA32_MISC_ENABLE: 0x00850089 (TCC EIST MWAIT PREFETCH TURBO)
CPUID(7): No-SGX Hybrid
CPUID(0x15): eax_crystal: 2 ebx_tsc: 172 ecx_crystal_hz: 38400000
TSC: 3302 MHz (38400000 Hz * 172 / 2 / 1000000)
CPUID(0x16): base_mhz: 3300 max_mhz: 3700 bus_mhz: 100
cpu0: MSR_PLATFORM_INFO: 0x804043df0812100
4 * 100.0 = 400.0 MHz max efficiency frequency
33 * 100.0 = 3300.0 MHz base frequency
cpu0: MSR_TURBO_RATIO_LIMIT: 0x2a2a2a2a2a2d3030
42 * 100.0 = 4200.0 MHz max turbo 8 active cores
42 * 100.0 = 4200.0 MHz max turbo 7 active cores
42 * 100.0 = 4200.0 MHz max turbo 6 active cores
42 * 100.0 = 4200.0 MHz max turbo 5 active cores
42 * 100.0 = 4200.0 MHz max turbo 4 active cores
45 * 100.0 = 4500.0 MHz max turbo 3 active cores
48 * 100.0 = 4800.0 MHz max turbo 2 active cores
48 * 100.0 = 4800.0 MHz max turbo 1 active cores
cpu0: MSR_SECONDARY_TURBO_RATIO_LIMIT: 0x2525252525252525
37 * 100.0 = 3700.0 MHz max turbo 8 active cores
37 * 100.0 = 3700.0 MHz max turbo 7 active cores
37 * 100.0 = 3700.0 MHz max turbo 6 active cores
37 * 100.0 = 3700.0 MHz max turbo 5 active cores
37 * 100.0 = 3700.0 MHz max turbo 4 active cores
37 * 100.0 = 3700.0 MHz max turbo 3 active cores
37 * 100.0 = 3700.0 MHz max turbo 2 active cores
37 * 100.0 = 3700.0 MHz max turbo 1 active cores
cpu0: MSR_CONFIG_TDP_NOMINAL: 0x00000016 (base_ratio=22)
cpu0: MSR_CONFIG_TDP_LEVEL_1: 0x00100040 (PKG_MIN_PWR_LVL1=0 PKG_MAX_PWR_LVL1=0 LVL1_RATIO=16 PKG_TDP_LVL1=64)
cpu0: MSR_CONFIG_TDP_LEVEL_2: 0x002100f0 (PKG_MIN_PWR_LVL2=0 PKG_MAX_PWR_LVL2=0 LVL2_RATIO=33 PKG_TDP_LVL2=240)
cpu0: MSR_CONFIG_TDP_CONTROL: 0x00000000 ( lock=0)
cpu0: MSR_TURBO_ACTIVATION_RATIO: 0x00000015 (MAX_NON_TURBO_RATIO=21 lock=0)
cpu0: cpufreq driver: intel_pstate
cpu0: cpufreq governor: powersave
cpufreq intel_pstate no_turbo: 0
cpu0: MSR_MISC_PWR_MGMT: 0x000030c2 (ENable-EIST_Coordination ENable-EPB DISable-OOB)
cpu0: MSR_PM_ENABLE: 0x00000001 (HWP)
cpu0: MSR_HWP_CAPABILITIES: 0x01121a38 (high 56 guar 26 eff 18 low 1)
cpu0: MSR_HWP_REQUEST: 0x80003805 (min 5 max 56 des 0 epp 0x80 window 0x0 pkg 0x0)
cpu0: MSR_HWP_REQUEST_PKG: 0x8000ff01 (min 1 max 255 des 0 epp 0x80 window 0x0)
cpu0: MSR_HWP_INTERRUPT: 0x00000005 (EN_Guaranteed_Perf_Change, Dis_Excursion_Min)
cpu0: MSR_HWP_STATUS: 0x00000000 (No-Guaranteed_Perf_Change, No-Excursion_Min)
cpu0: EPB: 6 (balanced)
cpu0: MSR_IA32_POWER_CTL: 0x00e4005b (C1E auto-promotion: ENabled)
cpu0: MSR_PKG_CST_CONFIG_CONTROL: 0x74008008 (UNdemote-C1, demote-C1, locked, pkg-cstate-limit=8 (unlimited))
/dev/cpu_dma_latency: 2000000000 usec (default)
current_driver: intel_idle
current_governor: menu
current_governor_ro: menu
cpu0: POLL: CPUIDLE CORE POLL IDLE
cpu0: C1ACPI: ACPI FFH MWAIT 0x0
cpu0: C2ACPI: ACPI FFH MWAIT 0x21
cpu0: C3ACPI: ACPI FFH MWAIT 0x60
cpu0: MSR_PKGC6_IRTL: 0x00000000 (NOTvalid, 0 ns)
cpu0: MSR_PKGC10_IRTL: 0x00000000 (NOTvalid, 0 ns)
Uncore Frequency package0 die0: 400 - 4600 MHz (400 - 4600 MHz) 400 MHz
RAPL: 15420 sec. Joule Counter Range, at 17 Watts
cpu0: MSR_RAPL_POWER_UNIT: 0x000a0e03 (0.125000 Watts, 0.000061 Joules, 0.000977 sec.)
cpu0: MSR_PKG_POWER_INFO: 0x00000088 (17 W TDP, RAPL 0 - 0 W, 0.000000 sec.)
cpu0: MSR_PKG_POWER_LIMIT: 0x812800dd8128 (UNlocked)
cpu0: PKG Limit #1: ENabled (37.000 Watts, 28.000000 sec, clamp ENabled)
cpu0: PKG Limit #2: ENabled (37.000 Watts, 0.000977* sec, clamp DISabled)
cpu0: MSR_VR_CURRENT_CONFIG: 0x000002f8
cpu0: PKG Limit #4: 95.000000 Watts (UNlocked)
cpu0: MSR_DRAM_POWER_LIMIT: 0x00000000 (UNlocked)
cpu0: DRAM Limit: DISabled (0.000 Watts, 0.000977 sec, clamp DISabled)
cpu0: MSR_PP0_POLICY: 0
cpu0: MSR_PP0_POWER_LIMIT: 0x00000000 (UNlocked)
cpu0: Cores Limit: DISabled (0.000 Watts, 0.000977 sec, clamp DISabled)
cpu0: MSR_PP1_POLICY: 0
cpu0: MSR_PP1_POWER_LIMIT: 0x00000000 (UNlocked)
cpu0: GFX Limit: DISabled (0.000 Watts, 0.000977 sec, clamp DISabled)
cpu0: MSR_IA32_TEMPERATURE_TARGET: 0x05640000 (95 C) (100 default - 5 offset)
cpu0: MSR_IA32_PACKAGE_THERM_STATUS: 0x88390c00 (43 C)
cpu0: MSR_IA32_PACKAGE_THERM_INTERRUPT: 0x02000003 (100 C, 100 C)
cpu0: MSR_MISC_FEATURE_CONTROL: 0x00000000 (L2-Prefetch L2-Prefetch-pair L1-Prefetch L1-IP-Prefetch)
13

我刚刚安装了 thermald 并启动他, 得到

○ thermald.service - Thermal Daemon Service
     Loaded: loaded (/usr/lib/systemd/system/thermald.service; disabled; preset: disabled)
     Active: inactive (dead)

4月 14 18:58:04 PsiThinkpadArch systemd[1]: Starting Thermal Daemon Service...
4月 14 18:58:04 PsiThinkpadArch thermald[26878]: 35 CPUID levels; family:model:stepping 0x6:bd:1 (6:189:1)
4月 14 18:58:04 PsiThinkpadArch thermald[26878]: [/sys/devices/platform/thinkpad_acpi/dytc_lapmode] present: Thermald can't run on this platform
4月 14 18:58:04 PsiThinkpadArch thermald[26878]: Unsupported cpu model or platform
4月 14 18:58:04 PsiThinkpadArch thermald[26878]: Try option --ignore-cpuid-check to disable this compatibility test
4月 14 18:58:04 PsiThinkpadArch systemd[1]: thermald.service: Deactivated successfully.
4月 14 18:58:04 PsiThinkpadArch systemd[1]: Started Thermal Daemon Service.

所以我修改了 thermald.service

# systemctl edit --full thermald.service
-----------------------------
[Unit]
Description=Thermal Daemon Service
ConditionVirtualization=no

[Service]
Type=dbus
SuccessExitStatus=2
BusName=org.freedesktop.thermald
ExecStart=/usr/bin/thermald --systemd --dbus-enable --adaptive --ignore-cpuid-check
Restart=on-failure

[Install]
WantedBy=multi-user.target
Alias=dbus-org.freedesktop.thermald.service

并且启动了他. 现在好像没问题了, 三个模式(powersave, balanced, performance)都可以正常工作, 而且 balanced 下 CPU 频率不会被限制在很低的水平. 就是不知道 thermald 是否真的和这台 Thinkpad 兼容 :expressionless:

14

可以用 override 的方式修改 service 文件，之后更新就不会覆盖你的修改了，见比如 How do I override or configure systemd services?
原来 systemctl edit 会生成一个 override 文件（用 --full 不会倒是），我都是自己创建的。。

thermald 听说能改善某些 intel cpu 的性能（不过我看到这个相关的都还是 tiger lake 的时代的说法，并且在我当时用的笔记本上没有明显的作用）

15

谢谢提醒, 我也只是试试 thermald (没想到真的有用), 维护的时候我会注意. Tiger Lake 已经很久了吗, 时间过得好快 lol.

16

在互联网冲浪中我发现了这个 https://askubuntu.com/questions/1520160/cpu-throttling-problems-on-my-thinkpad-x1-carbon-i7-10510u. 看起来是一个常见的问题 (对 Thinkpad 来说). 是否应该移动到笔记本模块?

我还在 thermald_daemon 的 github issues 中发现 Lenovo 应该有个自己的固件来调频, 而且 Lenovo 不支持 thermald (正如 这个问题的回答 所说). 所以这个问题应该是一个 Thinkpad 的问题 (或者联想的问题?), 我应该报告这个问题吗? :expressionless: