Sady my Acer bios has no such option.
Yeah I don't have C6 listed, I can enable CState and have PState = 0, which is what I have. Never found the C6 in question, it's probably called by a subtle name. loll
I do run 30C @ 2.20Ghz on idle though, testing various kernels right now to see which one seems to use the most of the powersaving features without losing Ghz when in use.
What CPU do you run on this Acer ? also a 5500u ? it can be called different as i mentioned before. Sadly this "c-state " has in Bios many names. It can be called also very simple energy save mode enable or something like this.
Ya have 5500u in my Acer laptop.
And there doesn't seem to be any setting regarding energy/ CPU in the bios.
Anyways the amd-pstate-epp driver is doing its trick by giving 0.4GHz min CPU frequency.
When you reach 400 Nhz idle this feature is enabled by default. What is normaly enabled in notebooks but sometimes not on Desktop-PCs.
The xanmod kernel, which has the amd-pstate drive(same as the linux-amd tested in the above posts), gives a minimum CPU frequency of 0.4GHz(when using on-demand frequency governor profile), and tends to be on the lower end when idle (similar to linux-bore with amd_pstate_epp).
The xanmod kernel also provides more number of frequency governor profiles for corectrl, which effectively changes the CPU frequencies as per the profile, as compared to
linux-amd, which only had performance and CPU-utilization profiles.
Today when I tested in the morning, I got some interesting results(probably because the system cooled down overnight):
These are the highest multi-core scores, though the single-core scores are not so great.
which kernel made which benchmark ?
All of them are with the xanmod kernel
I checked how long the battery lasts, between 80%-50% with the xanmod kernel :
I did this test with a browser running (for day-to-day browsing) and Spotify music running in the background.
Thanks Austin for sharing your testing.
So as result your battery lives longer with linux-xanmod in compare to Linux-amd and Linux-bore ?
Yeah maybe you test next time with all runing what you are usualy use to run.
But there are partly extreme resource Hungry apps as for example when you make a zoom-meeting.
But anyway it was interesting what results you get.
Well first let me thank you for the tip. Just updated the Bios with the Gandalf-OS, worked like a charm.
(Though I find it quite disturbing that my HP-Laptop can update the Bios without deleting the Linux-UEFI-entry but my MSI mainboard cannot ^^)
So, before I try it out, I read that
As a reminder, AMD P-State is only supported with AMD Zen 2 processors and newer.
And SGS seems to think the same. Did anybody try it on a Zen1-processor yet and can confirm one way or the other? If not I'll do the work, but you know... lazy and all
linux-bore is "more appropriate" for KDE than
linux-amd when it comes to poewr saving?
I am testing AMD and a bit of BORE, compared to my old time custom config TKG-BMG.
As filo said in general usage it's very hard to see any differences between kernels. But when it comes to super power saving modes when idling, you can see differences when some kernels just don't care about those.
LINUX-AMD currently does not support NVME sensors. I have flagged that and the guy will add it in next release, that's cool. AUR (en) - linux-amd (archlinux.org)
Both AMD and BORE are small packages in size compared to my custom TKG kernel. I like that too.
BORE supports INTEL, but with linux-amd if you take a look at its config.x86_64 you can definately tell it's focused at AMD (which makes sense with the name, doh! loll), as many of the INTEL configs are not set, set to N or deleted. For instance
But still I'm interested to understand more your point of view that BORE would be better in power saving for KDE?
Ups, if I think again I mixed something?
Only the AMD Kernel use the 2200-5000 MHz range and on my old inxi values I could see that zen also goes down to 400 MHz on the laptops. I think I have, for now, found my kernel.
I admire the people who investigate this deeper and carry out test series, I as a layman can only read off the power supply, what flows through there in each case in watts.
I must edit my post, thanks.
I think the Zen1 has not this feature of "c-state" but zen1 has also some mode what can take the cpu into sleep or deeper sleep. called "c6" what might be similar to "c-state". but correct me if i am wrong.
Today I did some tests with my FlightGear flight sim, and I found linux-xanmod, and linux(arch) gave the highest fps.
Linux-bore also gave similar fps, but the fps were inconsistent.
I think, for now, I will settle with linux-xanmod.
It would be nice if someone could run the Phoronix CPU tests suite with linux-zen, xanmod, bore, and linux(arch). I think it will give us some better insight.
i will do this on my Desktop and notebook when kernel 6.0 is out. Relased it is already according to kernel.org so should be soon in updates.
I found something new, which I would like to share:
Till now, we have established,
linux-zen & linux(arch) use acpi-cpufreq
linux-amd and linux-xanmod use amd-pstate
linux-bore uses the new amd_pstate_epp
All of these drivers work differently.
But all of these utilize the CPU according to the frequency scaling profile set, which you can view by:
❯ cd /sys/devices/system/cpu/ devices/system/cpu🔒 ❯ ls cpu0 cpu2 cpu6 cpufreq kernel_max online smt cpu1 cpu3 cpu7 cpuidle microcode possible uevent cpu10 cpu4 cpu8 hotplug modalias power vulnerabilities cpu11 cpu5 cpu9 isolated offline present devices/system/cpu🔒 ❯ cd cpu0/cpufreq cpu/cpu0/cpufreq🔒 ❯ cat scaling_governor ondemand
[cpu0, cpu1, cpu2 ... are your CPU cores]
You could change the profiles through the terminal, but I don't know exactly.
But there is a GUI(Corectrl) for that!
For some reason:
and linux-amd's default frequency governor was performance which caused:
linux-xanmod's default frequency governor too was performance, and it too behaves like linux-md, but only when I change the profile to ondemand(incorectrl) did it start behaving like linux-bore i.e cpu-frequencies tending to 0.4GHz.
And I believe because xanmod's default frequency_governor is performance which gives consistent high frequencies, the fps was also consistent as compared to linux-bore ( whose frequencies varied on a wide range from 0.4GHz-4Ghz even with 'performance' frequency_governor profile, due its amd_pstate_epp driver )