Testing Six Different Linux Distributions On The Intel Core i9 13900K "Raptor Lake"

[sb56637]

Thanks, is there a gnome/wayland option?

Yep. I disable Wayland by default for several reasons, but you can revert the change by removing WaylandEnable=false in /etc/gdm3/daemon.conf .

[agd5f]

I currently use the long-term-support 5.15.75 kernel. I will probably update to 6.1, when released, if that will be the next LTS.

The AMD P-State driver has been added to 5.17, so it is newer.

I assume that the new AMD P-State driver might behave like the intel_pstate driver, which bypasses the standard CPUFreq governors, so it might also have a variable clock frequency even when the powersave governor is chosen.

However, I assume that the vast majority of the computers with AMD CPUs and Linux have not been updated yet to kernels newer than 5.17, so when "powersave" is chosen on them, the standard governor is used and not the one specific for AMD CPUs.

The governors are still in play whether you use the old ACPI pstate cpufreq driver or the newer AMD CPPC pstate driver. The difference is that the older ACPI interface expose 3 clock levels and the newer CPPC interface defines an abstract performance continuum which mean you have more or less as many clock levels as there are bits in the abstract interface. So the old ACPI interface may define 1Ghz, 2Ghz, 3Ghz as the discrete clock levels. With CPPC, you have a continuum of clocks between 1Ghz and 3Ghz. The old ACPI interface will naturally ramp up quicker because there are only 3 levels to pick from whereas with CPPC, you have a wide range of available clocks. The governors will operate over any underlying cpufreq driver. The cpufreq driver just provides the means to change the CPU clocks. The governor sets the policy for changing those clocks.

[krzyzowiec]

Clear Linux looks interesting, but it seems there is no support for either Snaps or LXD and they have no interest in adding either of those. It's quite unfortunate, since Ubuntu is rather slow.

Not according to these benchmarks it's not. In fact, the performance looks great!

[Sin2x]

As expected, Debian beats Ubuntu which only rolls further downhill with each subsequent release.

[Anux]

I have said very clearly that I speak only about desktop Zen, because I have not tested any laptop Zen.

The behavior is like I have said at least on Zen 2 Ryzen 7 3700X and on Zen 3 Ryzen 9 5900X, and on an ASRock Mini-ITX MB and an ASUS ATX MB.

I have not understood what exactly you disagree with. On a laptop CPU it is expected that the idle clock frequency is much lower than on a desktop CPU, i.e. 0.4 GHz vs. 2.2 GHz.

Yes it is totally possible that the min frequency is higher on the desktop parts (see my G APU min is 1600 vs 400/800 on mobile) but it should still change frequency and not stay at the same.

Someone had the problem in april and it was clearly accepted as a bug: https://bugzilla.kernel.org/show_bug.cgi?id=215800
Try a newer kernel, check the frequency after a short standby and check for bios updates.