Arch Linux

patchouli

Member

Registered: 2024-09-18

Posts: 19

[SOLVED] Poor CPU performance on Arch Linux

I have an Intel Core i9-13900k. Brand new unit.

Up to date system. Only 16 threads (or 18, for some reason) are utilized in various CPU benchmarks. Cinebench via WINE, stress, stress-ng,  and passmark-performance-test.
In PassMark, my CPU scores 45k.
I booted up into a Fedora live ISO, there my CPU is fully utilized out of the box. The PassMark test scores at 61k there. Significantly higher.
The CPU is 16P cores + 8E cores, totaling at 32 threads.

I figured this is some configuration error because I ran benchmarks on Arch before my CPU was fried (the new unit is a replacement) and the whole CPU was utilized correctly.
I'm unsure how to troubleshoot this; but I'd like to utilize my entire system.

Some information:

$ sudo cpupower frequency-info
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:  Cannot determine or is not supported.
  hardware limits: 800 MHz - 5.50 GHz
  available cpufreq governors: performance powersave
  current policy: frequency should be within 800 MHz and 5.50 GHz.
                  The governor "powersave" may decide which speed to use
                  within this range.
  current CPU frequency: Unable to call hardware
  current CPU frequency: 1.51 GHz (asserted by call to kernel)
  boost state support:
    Supported: yes
    Active: yes

Setting the policy to performance doesn't help.

I've tried linux, linux-zen, linux-cachyos.. no luck.

Here's my command line:

root=UUID=b4b53932-d0bc-41d8-a278-b7f3fa6fbf3c rw nvidia.NVreg_UsePageAttributeTable=1 nvidia.NVreg_DynamicPowerManagement=0x02 nvidia.NVreg_PreserveVideoMemoryAllocations=1 nvidia.NVreg_TemporaryFilePath=/var/tmp nvidia.NVreg_EnableGpuFirmware=0 intel_iommu=on iommu=pt i915.enable_guc=3 i915.max_vfs=7 module_blacklist=xe libahci.ignore_sss=1 quiet splash loglevel=3 systemd.show_status=auto rd.udev.log_level=3 add_efi_memmap default_hugepagesz=1G initrd=intel-ucode.img initrd=initramfs-linux.img

Any help on troubleshooting this would be appreciated, thank you.

Last edited by patchouli (2025-01-13 22:37:27)

Head_on_a_Stick

Member

From: The Wirral

Registered: 2014-02-20

Posts: 8,738

Website

Re: [SOLVED] Poor CPU performance on Arch Linux

https://wiki.archlinux.org/title/Microcode?

Those 13th & 14th gen Intels had voltage issues that probably need correcting. Intel issued an update on 2024-11-12.

---

Para todos todo, para nosotros nada

patchouli

Member

Registered: 2024-09-18

Posts: 19

Re: [SOLVED] Poor CPU performance on Arch Linux

https://wiki.archlinux.org/title/Microcode?

Those 13th & 14th gen Intels had voltage issues that probably need correcting. Intel issued an update on 2024-11-12.

I run the latest microcode, can be seen in my command line as well. I update the BIOS too. The tests performed on the live ISO from Fedora don’t have it applied either. I read on GitHub that the voltage changes need to be applied on boot, and that the microcode img won’t affect it.

Last edited by patchouli (2025-01-13 18:37:44)

patchouli

Member

Registered: 2024-09-18

Posts: 19

Re: [SOLVED] Poor CPU performance on Arch Linux

Got troubleshooting help on reddit. I had scx running configured to use scx_bpfland with -c 0:

  -c, --nvcsw-max-thresh <NVCSW_MAX_THRESH>
          Maximum threshold of voluntary context switches per second. This is used to classify interactive tasks (0 = disable interactive
          tasks classification)

          [default: 10]

Got rid of that and I'm good now.

Last edited by patchouli (2025-01-13 22:38:27)