I wouldn’t necessarily trust any measurements of the display power, there’s no sensor there to measure it. I’d recommend playing with different brightness levels and measuring it yourself.
I also wanted to post an update on how I’m seeing a recent issue with KDE Plasma’s kwin platform that’s drawing a ton of extra power. I’m temporarily using lxqt right now, and my average power draw for text editing is at 6.7W.
I wanted to share a finding of mine. Having switched to LXQT with mutter as my window manager, my power draw has become incredibly consistent. Instead of bouncing between 7W and 10W like I was on Plasma, I am instead bouncing between 6.7W and 6.75W. I saw one spike to 6.8W once. That equates to 12.6hrs of battery life. It doesn’t have all the fancy gaming features I liked about Plasma 6.2+, but for my work use cases the consistent and long lasting battery life is amazing. I don’t know anymore if I’ll be going back to Plasma once I figure out how to resolve the power-draw bug my install currently has.
One annoying thing about this is the palm-rejection/disable touchpad while typing doesn’t work out of the box, and considering all I do is type it’s actually quite annoying. Thankfully, synclient touchpadoff=1 allows me to work around that for the moment.
If anyone is interested in battery above all else, I recommend giving LXQT + mutter a chance (other window managers seem to draw more power).
I’ve gotten down to 7 watts on Windows 11, airplane mode, lowest screen brightness, 60hz, etc. Note this is monitored use HWINFO64 so perhaps even lower without the continuous monitoring. I normally idle around 9-11 watts on normal Windows use, browser open, MS Teams, HWINFO64, 165hz, normal-high brightness, BT + Wifi, etc. Some food for thought.
Edit: I received my no-DGPU expansion bay, I’ll put that in and try again. I should have also mentioned my CPU is the 7840hs variant.
I think this may be the first Windows submission we’ve seen so far, and it’s a pretty good one too! I wouldn’t be surprised if airplane mode actually helps significantly on Windows as it essentially forces the various background process at work to not do anything.
Thanks for sharing.
That’s on par with our best linux scores. Thanks for sharing!
mind sharing your config on how did you achieve this of fw16?
currently optimized mine from 60 W (with all of fancy background services) down to 15 W for minimal builds
https://gitlab.com/sophronesis/nixos-config/-/blob/main/modules/low-power.nix
CPU: 8-core AMD Ryzen 7 7840HS w/ Radeon 780M Graphics (-MT MCP-)
Kernel: 6.18.20-1-lts x86_64
OS: EndeavourOS x86_64
Display: 2560x1600 @ 1.34x in 16", 60 Hz [Built-in]
DE: KDE Plasma 6.6.3
I tried to minimize my power draw today, just for this thread.
powertop --auto-tune
power profiles daemon set to energy-saver
Screen brightness 0% (no kernel patch)
Input modules lighting off
Camera & Microphone off
60hz refresh rate
Browser: Floorp
I got the following minimum numbers:
Is it the Wi-Fi that makes such a huge difference, or are there other differences between the two idle conditions?
The absolute lowest I’ve seen on my machine was 7.5W, but lately it’s been closer to 8.67W. I’m using Plasma 5.x.x though, so I guess it’s a major contributor to these numbers.
sending data back and forth wirelessly consumes a noticeable amount of power. In theory, if you’re not doing anything on the internet, then the wifi card shouldn’t be in use and turning it off won’t do much. For me, this is the case w/ my Linux install, but on Windows… well… that’s a different story.
@Numerfolt Thanks for sharing, those stats line up pretty well with my own usage!
Lately I’ve found that powertop autotune doesn’t impact my battery draw much anymore.
Turns out I forgot to set the display to 60Hz. 
Display at 60Hz, tunables adjusted, mic & camera off, Wi-Fi on:
6.01W @ 50% brightness
5.23W @ 0% brightness
Wi-Fi off:
4.9xW @ 0% brightness
It’s a fun exercise - I was surprised that physically turning off the mic and camera made a noticeable difference. I’m definitely going to check out LXQt.
Personally I’d be happy with <7W during normal usage, so e.g. web browsing or writing code, as that translates to 9h of battery life in the 5-80% range. Not sure how realistic that is.
I’ve found <7W really hard to attain in actual usage, but I’ve achieved ~7.5W before with occasional spikes to 10W (local text editing, no browser).
60hz definitely makes a difference, yeah 
. I have a script that automatically switches between 165 and 60 depending on ac/bat power.
I don’t think changing desktop env will help you much, I’ve tested using no desktop env and just having the terminal session and power draw does not go down 
Yeah . The intention of the exercise is to identify what the lowest possible power draw for a given machine is, that way anytime you’re drawing more power than that, it’s a conscious and intentional decision to trade battery life for some sort of functionality. This is opposed to just turning it on, using it, and getting whatever battery life you happen to get with whatever software and lifestyle configurations you’ve got going on.
This is one of the disappointing things about the Framework 16. I get 3-4 watt idle on the many 13’s (mostly 7840U/7640U variants w/ 120Hz displays) I have and I don’t understand why the 16 is so much higher given the same architecture, even without the GPU. My 16’s are actually on the shelf because of this and because I decided it was just too big/heavy and didn’t need the minuscule extra power it provides.
Are you getting less time on battery life though?
In 3. the browser is open as well
