Now that there have been FW13 AMDs out in the wild for a few weeks now, I am interested to hear about real-life experiences of battery life on Windows 11 - particularly for regular office type work such as browsing, MS Office, Zoom/Teams etc., which for me represents best the type of work where I will be away from a power outlet.
I know there is a lot of benchmark data available, but I’m not really planning on watching Youtube videos for 10hrs with minimum screen brightness, or running Cinebench 24/7
Running powercfg/batteryreport gives a nice summary of estimated total battery life per day for those who are interested.
Would very much appreciate input from new FW 13 AMD owners!
I do not use Windows 11, I prefer Arch Linux. Though, I hope this will be a reasonable comparative. I tend to mostly watch YouTube videos on my Framework for now. I also do some web browsing. I use it at a sensible screen brightness (around 25-30%) with speakers on 40 to 80% depending on content and if I am trying to listen while cooking or something. I can say it lasts 6 hours easily and probably more. I have a Ryzen 7 7840U which, iirc, grants me a slightly larger battery as well so YMMV.
@Shijikori Have you configured your Linux with any power management? I remember seeing a recent video review about 13 battery life (productivity based) and it was close to 10 hours. Seeing as how your workload is minimal, you should be able to get considerably more on that laptop; otherwise it’s bad news for future Linux users of Framework 16 like me.
Linux battery life is currently borked IMHO. Needs some major optimization / tuning to come close to Windows right now. Am currently dual booting Windows 11 / Ubuntu, but was previously running F39 beta.
Screen shot shows nearly ~4 hours of screen on time, 10 hours of sleep time over a 24 hour period and only used 45% battery total. Mixed productivity usage.
Edit: Also attaching my estimated battery report
@Asad1 I do not use any power management besides what KDE does out of the box. I prefer to avoid using TLP as, in my experience, it likes to simply kill bluetooth and/or wifi the moment you go on battery, requiring a reboot to regain the functionality (that experience was not on this laptop, I simply have a negative opinion of power management available on Linux). The scaling governor defaults to powersave and I have not changed it so the CPU is targeting power saving, although I am aware this is not the most aggressive power saving. I should note that I have seen my chip draw 20W or more on battery due to not having power saving features on, besides whatever the firmware 3.02 does ootb.
That is only my personal report based on my personal usage though. Again, YMMV. I would assume one of the biggest battery drain is the speaker volume and those times when the CPU draws a lot of power watching YouTube with Discord in the background.
Also, I did not charge it full and than sat down and used it until dry, there’s a decent bit of charge lost to s2idle. Far from the most scientific test but given I did say I can get 6 hours out of it pretty consistently, including long periods in s2idle passively using a little bit of battery, you should have a good amount of battery life.
I have just posted a reply in [GUIDE] Linux battery life tuning.
So far, my battery life is worse with my Ryzen 5 7640U than with my i5-1135G7.
Especially in FHD video playback, I’m looking at 4-5h instead of 5-6h.
At the moment, I hope it’s some sort of driver thing on Linux and that it will get better.
Mostly, because it does not correlate with the overall observations of significantly better battery life.
@chrisp_chicken do you use VAAPI for video playback? This can help save power a lot.
I did apply the config parameters for Firefox as specified here: [GUIDE] Linux battery life tuning
media.ffmpeg.vaapi.enabled set to
I am still observing about 11W in Full HD video playback, though.
I see between 10 and 15W on FHD playback as well on YouTube, I would say that fits. So, other than a smaller battery, it would come down to the power saving features of your distro and how they are setup.
@Shijikori In that case, I perhaps set my expectations too high. I was anticipating longer battery life than with my 11th gen Intel mainboard, which supposedly isn’t very energy efficient.
Thanks for the insight!
No your expectations aren’t high. There are driver issues IMHO. My wife’s 2-3 year old laptop with a Ryzen 5900HS uses less power during YouTube playback on Linux.
Windows 1080P YouTube playback for me is around 7-8 Watts, Linux is 11-13 Watts even with VA-API working correctly. For offline 4k30 h.264 mp4 videos Windows is 6-7 watts and Linux is 10-12 using va-api. This is literally a 50%+ increase. That increase could half your video playback battery life.
I think we need to collect some logs and submit them to AMD, because the battery performance on Linux should be much better than what it is. This is not as intended by a 4nm cutting edge chip IMHO. It should not perform worse than their 7nm chip produced years ago.
I do agree but I don’t see this as terrible battery life coming from a brand new platform. I already get 2 hours more than my ThinkPad E595 with a Ryzen 5 3500U and worlds better performance.
I’m experiencing the same poor battery performance as you. I think it’s driver issues, since a lot of people and reviews are reporting much better efficiency than any of the previous intel gens - and 11th gen is the worst.
Here is the battery on fedora 39, with no tweak, battery saver on at 30%.
My task is mainly light coding, internet browsing, slack, and signal, which should be slightly more power consuming than typical office work.
I have taken two video conference on zoom from 65% to 40% and another one from 20% to dead.
My battery life on the AMD running windows 10 also didn’t seem better than my 1135g7 framework. The newer (I think) beta drivers from framework didn’t seem to help but drivers directly from AMD’s website based on my admittedly somewhat limited initial testing have.
The laptop seems somewhat cooler now, and sleep power usage according to sleep report seemed to decrease from around 2.5W+ (at one point being over 9W in sleep) to perhaps half that with the previous major offenders no longer seeming to show up. Still not ideal (and perhaps twice as high as I’m getting for sleep on the 1135 (at least on the times the audio drivers aren’t keeping the 1135 active)), but an improvement I think. But it’s unclear at this point what else might be using so much power.
The GNOME Power Statistics application shows an average of ~6 hours to bring the battery from 90% to around 15% charge. Nearly all web browsing, some remote desktop connections, and only a small amount of video playback and video calls.
I do notice that the Framework chassis becomes warm to the touch during these basic activities, whereas my Surface Laptop 3 with a 10th gen Intel i7 will stay cool until I push it with extended media playback, or if I try and send/receive a chat message in that ogre of a program they call Teams.
So it feels like the Ryzen 7640 in my Framework is having to work a lot harder to do basic tasks as it stands now. I’m hopeful AMD can make improvements here as time goes on. I feel like on paper, we should be seeing much better power performance.
Got to keep in mind that you don’t get both the performance and power consumption advantages with a node shrink. It’s one or the other. Depends what AMD chose to do but it would be quite a shot in their own foot to node shrink and do worse power efficiency and I don’t mean better performance for same power consumption but just worse efficiency overall. The platform is new and definitely needs some firmware and driver improvements to give it’s full benefits but what we have now, to me, looks to be pretty good already.
Same. I tried Powertop when new to Linux and saw that all it was doing was writing settings to sysfs, so I just configured udev rules to do that for me without any extra tools needed. That and a nice undervolt got Linux to have better battery life than Windows on my current non-Framework laptop.