It is a problem if you don’t deactivate PL2 and the CPU always runs at its temperature limit. Also the power consumption is high. My “download” laptop with an N6005 and Manjaro also works with schedutil.
My mini-PC has a PL2 limited to 25W by default.
Its power consumption at the power outlet is a maximum of 12W with IPFire.
The temperature is stable between 45 and 50°C without Fan (it’s hot in Marseille 
).
I left the governor on performance by default
So what…. if you live in a climated house @ 20°C it doesn’t matter where you live in. In my place without air conditioning @ 30°C ambient temperature you will not have a N100 @ 6W TDP @ 45°C.
no air conditioning at home
Did the update but can’t see any changes in the WUI.
That is because there are no in you face changes in the WUI.
With regards to the customservices file, the new aspect is the addition of the DNS over TLS protocol in the list but this updated default list will only be provided if you do a fresh install.
For an update of an existing install tour existing customservices file is left alone otherwise it would overwrite any additions you have made.
The other new thing was the ability to see which firewall rules are using specific group entries. There was never any problem with this addition to the groups page. It worked as expected in the Testing and final releases. This information is only shown when you place your mouse pointer over a specific location as described in post 2 of this thread that was marked as the solution to the original question.
Now I got it. That is handy. Thx.
What you’re saying vs. the actual behavior is not congruent.
Here’s another graph from the reboot with nothing changed on the box. CPU is back down. The only thing that ever changes the CPU frequency range is when I upgrade and then reboot, the CPU becomes range bound at 2x what it was before. I did a reboot today just to see what would happen and the CPU returned back to what it was before I did the core update.
I wonder anyway why there are so many ups and down with no governor set to do so. However the kernel governor can only “ask” the system hardware to do something with the clock. Maybe this is related to the system firmware. You may test it with a different governor. Even if you have only powersafe (I wonder about that) you should see a different behaviour. If not, the firmware messes up something.
1 Like
I thought p-states were used and this limits anything you can do with governors.
Yes. I just grabbed another Mini-PC I have with an N100, some Topton Mini-PC and this one shows up only performance and powersafe as well.
I will have a look on my N305 Proxmox. Maybe I get an invalid argument output when trying to set it to schedutil as well. I’m not sure if I just added the command to cron without checking.
The N100 Processor base frequency is 0.8GHz. That matched to mine and I don’t have that peak range of 1GHz. But PL2 overclocking is disabled.
1 Like
What’s the point of buying an N100 PC and limiting it to 800 MHz?
I see that it doesn’t really lower the temperature, even with a fan.
How much less power does the socket consume?
And in terms of performance, with IPS enabled, URL Proxy, and even QoS, what does a speed test with a 1 Gbps WAN connection yield?
serious lack of moderation here 
https://community.ipfire.org/t/core-update-196-released/14398/34
Why don’t you do someting?
Because manufactorer overclocking sucks → see Intel 14th gen, AMD 7000X3D series, ASUS ROG Ally etc.
2 Likes
The throughput usually is not even going to be the main impact here. At 800 MHz you will however see a huge drop. Bigger will be latency spikes because the CPU cannot do the work it needs to do. Forwarding a packet on the network should happen as soon as possible. If queues fill up, your system might take longer to process the packet than what it takes to cross the Atlantic Ocean.
Yes I have huge CPU load here. I use everything except VPN. 97 FW rules, IPS on RED, lots of URL filters. ClamAV etc. And since tonight I download at full Gbit speed on WAN.
I checked to governors on my Celeron 6305 laptop and N305 Mini-PC and I’m wrong. They all just have performance and powersafe. However my laptop with the Celeron 6305 on Manjaro changes the freqencies between 0,8 and 1,8 Ghz, even if it’s on powersafe 
.
Looks like I’m out of date. I used the situation and installed PVE 9.0 on the N305 and checked the frequencies with
watch 'cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_cur_freq'
with performance governor → frequencies change between 0,8 and 1,8 GHZ. So it’s the same like my Laptop. That’s interesting. Something is different here to the other CPU families.
With my private 24/7 HV with a 12900KS this is totally different.
Google AI says:
Yes, Alder Lake-N processors, even when set to power-saving modes, can exhibit frequency changes due to the dynamic nature of Intel’s Turbo Boost Technology and power management features. These features allow the processor to dynamically adjust its clock speed based on workload demands and thermal conditions, even in low-power states.
The AI seems to be wrong. Not Turbo Boost is responsible for that behaviour, but Intel Speed Shift!
OK for my laptop I can not disable anything, but on the N305 machine I was sure I disabled it…
Check: I just disabled PL2, but Speed Shift is still active. That’s even better, so dynamic frequencies and power consumption is firmware-controlled and works even without governors. In that case with 1,8 GHz instead of just 0,8 Ghz of the N100 and mmore cores it’s much better.
1 Like
Core Update 197 has gone into testing today which will affect CPU frequency.
Help with testing would be appreciated.
1 Like
i have an i3-1215U based system on my desk right now.
not heavily tuned bios settings and it draws ~18W from
the wall with just one i226@2.5gbit online under 6.14.8-2-pve
with the difference when idle is ~1W between performance and powersave
with load the two can jump to the same ~40W but the
performance governor feels somehow snappier 
For my 12900KS it’s a huge difference. With performance, even with PL2 disabled, the system takes about 80W (CPU 125W limited to 65W). The small room, where all the tech stuff is in, gets quite warm. With schedutil the Systems power consumption goes down to about 20W.
When I changed it to schedutil I watched the utilization. It rised just a little.
1 Like
these numbers are quiet interesting 
is the system the pve host or bare-metal ipf 