To be honest, as far as I am concerned, no it is not something to be worried about.
I have occasionally also seen step changes, although smaller than yours. They have never ended up affecting the performance of IPFire in terms of the other performance graphs.
The change could be as simple as changes in the linux kernel changing how the freq is calculated or shown, ie the performance stays the same but the graph is different.
It could also be that there has been some other change in the kernel or iptables or some other package that causes a change in the requirements on the cpu and hence you see a change in the frequency performance.
In either case any change is never something that I have chased up and the performance of IPFire for me has not been adversely affected.
Are you seeing performance impacts, such as reduced throughput or other factors that make you concerned or is it related to seeing the steps in the cpu freq graph?
In other posts other users have tried to change the performance governor defined by IPFire to on-demand governor and found little difference in anything, including the cpu graph.
I have also seen comments that the kernel developers don’t do much with the cpu frequncy governors any more as with modern processors the difference are relatively small and some cput’s just ignore what is set and do their own thing anyway.
In conclusion, I rarely look at my cpu freq graph except when it comes up in the forum.
I occasionally look at Status > System to get an overview of load. Sometimes I notice the CPU freq changes after upgrades but I’ve just ignored it. This time I decided to investigate it - for no particular reason!
The CPU usage is low: averages 99% idle; I’ve not noticed any throughput slowdown.
I’ll continue to ignore it unless I learn otherwise.
In the past it has also gone from a lower level to the current higher level.
To be honest I just ignore the frequency graph. The Load Average of processes running on it has basically been the same over the whole time, so in terms of the processing power for traffic being managed, it has never been influenced by the frequency.
Running higher frequency automatically means consuming more power, because the cpu voltage needs to be higher. Therefore the systems heats up more and the livetime decreases as well.
I know I should not worry too much about the cpu frequency. But still due to power consumption and heat generation on a passively cooled system, I wanted to test something. For quite some time I have had a constantly high frequency. So I have saved the settings, downloaded 193 fresh and installed it on the (bare metal) system again. Overwriting the existing installation (193). After a basic set-up (just DHCP on RED and DHCP on green and add-on cpu-freq-util) I have restored the saved settings from my “old” installation, re-boot and voilà, the cpu freq is now very low. With exactly the same settings. I think that from time to time a fresh install and restore of the settings is good. Just like on Windows :-).
This is open source software and this is community of the developers of this router distro.
Kernel is tweaked, optimized, changed, in my opinion devs should know where the changes are made, how kernel works. And why the CPU “load changes” so dramatically between core releases.
It’s the result of their development.
. no problem to see…
