Now that I have a few desktop environments installed, I decided to do a test of power consumption; this laptop's battery is not at its best moments and in the time i have to wait until I get a new one, I would like to use the desktop that can manage the battery best as possible when I do a fresh install on a SD card (HDD is close to failing and therefore not reliable). And the test I did got me... confused
Using the "powertop" terminal program I checked the power consumption in W (I think it stands for Watts) and, well, these are the results (highest power consumption reached for each one on the time of opening the documents folder and creating text files where I wrote this info):
GNOME: 9.66 W
GNOME (Wayland): 9.46 W
Cinnamon: 10.0 W
Cinnamon (Software rendering): 10.8 W
Cinnamon (Linux Mint Debian edition 5): 9.83 W
KDE Plasma (Manjaro, kernel 6.5): 12.0 W
xfce: 10.0 W
With the exception of LMDE and Manjaro, cpupower was running in the background and limiting the cpu usage to 1 ghz for each of the 4 cores (the speed it would have used without that would have been 2 ghz) and using the "powersave" profile
What got me confused about this was... isn't gnome known for being the heaviest one and xfce for being the lightest one? If so, I find it weird that gnome with and without wayland and with a few extensions activated got the best results out of all of them while a very basic almost fresh install of xfce showed higher power consumption
I can certainly see how this is confusing.
But the difference here is that you are equating power usage against resource usage (memory and r/w, i/o).
They are different things. And if I may be blunt - The test you describe is not a very reliable metric. Power consumption can vary a lot from moment to moment. And at idle, a lot of it has to do with with what is automatically running in the background. Many things will initialize, run and terminate periodically, even when you have left the computer idling.
Gnome is a heavy desktop - in that it uses resources more heavily than some other distros. But Gnome is actually quite good with Battery Life on notebook computers, in spite of its heavier CPU and RAM requirements.
Light Desktops like LMDE, LXQT, XFCE and LXDE will show less CPU, RAM and other usage - but the baseline power consumption on a computer will still be very similar across all Desktop Environments due to the majority of the power consumption being in the running of the computer and system - not really the desktop.
Your test seems to catch the periodic power usage depending on the moment, more than a difference in desktops.
Reduce screen brightness... this is the number one consumer of power.
Use the integrated graphics over dedicated cards. Even at minimal resolution it will use more power as it is am extra peripheral... didn't come with the board but was added to it.
Limit the number of background applications and startup applications. The more things happen in the background or are waiting for you to interact, the more resources in use, more power in use.
There are very few laptop motherboards that will use less than 5 watts alone. Desktops can be lower, but you will not have top of the line processor, large amounts of memory and possibly only integrated graphics.
Measuring anything but wattage would be a waist since the boards require constant voltage and amperage in order to work properly. If the power supply is doing its job, there will be little to no fluctuation, in either, at all.
That's just... interesting. I've always thought that more resource usage = more energy consumption = battery lasts less
I'm guessing it goes way deeper than that
well... that's a fair guess. I just thought the desktop's impact on power usage would be bigger, since on a post I've seen a while ago someone was comparing power usage on different distros and got over 10w in some (mostly ones that use gnome as the desktop) and staying below 5w on others (cinnamon and xfce), but now that you mention it, i dont think that person specified if all the tested distros were under the same conditions in terms of background and start-up apps
I always use it with the brightness at almost the minimum, I don't think I can get it any lower while staying usable
as far as i know, i think my laptop only has integrated graphics card, not dedicated
disabled almost everything except the basic system processes and just in case always run a script to close gnome software right after booting up, just in case
yeah that's fair. I wrongly assumed that a big part of the power usage would be coming from the desktop environment (for simple tasks, of course, games are usually heavy enough to make the difference between desktops almost unnoticeable), but now i see where i was wrong
thanks to both of you for the tips and the explanations!
Are you able to link to this post? I find those claims to be... incredible...
Your conclusions are not wrong.
More resource usage can equate to more power consumption.
The Gnome Developers have employed some neat tricks to make Heavy Gnome light on its feet. A bit like a Linebacker doing ballroom dancing.
While Gnome will draw more power at some times, it will shut down other tasks at times, and delay other tasks. This helps to offset power drain.
And yet, you had completely different results with your test, showing Cinnamon as higher watt usage.
In Science, there are several things a person needs:
A control. Without a control against which to compare the data, data is meaningless.
The ability to independently reproduce results. If a process is understood, it should be consistent.
The ability to predict an outcome. If your model is accurate and well understood, it should be able to define the next outcome by following the principles of the physics governing the process.
You must Show Your Work. You need to demonstrate that your data sets are reliable and how you got the data. Just making an unsupported claim means little. Allow me to demonstrate:
"I am testing Zorin OS wattage usage and it just was running at idle pulling fifty million megawatts. Personally, I think that this is unacceptable for a dirigible. I ask the Zorin Developers to reconsider using candle wax in their Desktop Environment."
Data without support is as trustworthy as a used car salesman. Without supplied support, the statements are suspect and are as likely to be gibberish as they are accurate.
Complex systems require complex tests. The wattage can vary, a lot, from moment to moment and that varies depending on which processes start running and which stop running. It can have variation due to hardware differences. It can have variation due to personally installed software and personal configurations and customizations. What is in the startup and what isn't.