Hello!
I am looking at getting a 16 DIY and I am concerned about battery life and heat because I’ve read a bunch of reviews that said it only lasts like 4 hours? but others that say it will last like 8? I know this depends a lot on what you are doing and your program, so anyone have any info on running Windows on it, with the basic expansion bay and mid-line cards and such (I am just barely fluent enough in computer jargon for this apologies). I do a lot of stuff in the field and away from outlets so I need all the battery life I can get
I also have heard a lot of issues with heat and the fans not turning on until after it gets hot…most of these reviews were old is this something that has been resolved? And I am under the impression it’s less of an issue without the expansion bay?
Thank you so much, I imagine many of those issues have been answered on these forums but I am kinda overwhelmed so all help is appreciated!
I find the battery life pretty good. The amd gpu drivers are getting better battery life wise, with enabling panel self refresh recently.
I use Linux so cannot really comment on the battery life on Windows.
The FW16 manufacturing is moving from Liquid Metal to PTM7958. If you are purchasing a new FW16, I would wait until you can make sure your one will be using PTM7958.
Hopefully FW16 have or will (maybe someone else can answer this bit) publish a date when all new FW16s will be using PTM7958, and that there are no old LM ones being shipped from stock. FW are coming out with a kit to upgrade from LM to PTM, but my expectation is that it might be a tricky process, so best to wait so you don’t have to do it.
The PTM should sort out any fan problems.
There are some other problems with the FW16, but you don’t ask about them, so I won’t bring them up. They have been discussed at length on this community forum.
Just came across this thread. I have one of the first FW16 (batch 2) and battery life is not good. I run latest Ubuntu on it, and best I can get when doing decent work (coding, texts, etc) is 5h. If something power hungry runs (like local AI) then you can watch the battery level go down.
Other than that I’m happy with my machine. If there’s every a solution for the battery (exchanging for a better one or some amazing external solution - my anker powerbank/battery gives me only about +3h)
I use Linux instead of Windows, but my personal experience with battery life is that it lasts 6-15hrs during light to moderate load. I can’t personally comment on the Windows battery life, but I’ve seen reports in another thread that suggest Windows could have better battery life ( Minimum Power Draw Competition - Framework Laptop 16 - Framework Community).
In Windows I get about 5 hours with moderate use, nothing too demanding with OneNote, Excel, Discord, Internet browsing and YouTube during breaks.
But I have limited the charge to 80% and plug it in at 20% again
I have the top spec-ed FW16 with the GPU. I get about 4 to 5 hours on Windows 11 just web browsing. During a full day, I’ll have to charge it once or leave it plugged in.
Conversely, my wife has a M4 Macbook Pro and even when doing heavy rendering of genomics data, her battery will last at least a full day of work, if not two.
The Framework is a good portable workstation, but leaves a lot to be desired if performance and battery life are your main concerns. A lot of people ask me if they should get a Framework 16, and most of the time, my reply is “just get a Mac”.
I don’t know how you guys get such amazing battery life. I am running Ubuntu 24.10 and it drops to under 2 hours (at 100%) the moment I join a Google Meet call.
To be more exact, 15-16W consumption with almost no workload (writing this post), 13W idle and ~30W if a Google Meet tab is open.
I idle around 6W, so there’s definitely something pulling unnecessary power on your install. I’d expect a video call to draw a notable amount of power though, so that doesn’t surprise me. Fedora out-of-the-box seems to have pretty good idle power draw. If you’re using Plasma, there was some sort of bug in Plasma that had skyrocketed the power consumption to around 13W, but that seems to have been fixed in a recent version.
At least now I know I need to investigate and aim towards 6w, so thank you for that!
I suppose it’s normal that Meet consumes a bit of power, but it was excessive though, it got so hot I had to change the fan profile…
It’s non-linearly better. Here’s a graph I made back before the FW 16 came out and people were speculating about the battery life. You can ignore the orange “w/ GPU” plot, thankfully it turns out that the w/ GPU and w/o GPU battery draw are practically the same… unlike the device I had prior to this. (The orange line is just the blue line +8W).
As you can see though, the gains from reducing power draw grow non-linearly the further you can reduce power consumption, particularly once you get under ~10W. If you’re pulling 6W and manage to drop to 5W, that’s an extra 3-4hrs. However, if you drop from 15W to 14W, it is likely only a few minutes of difference. The more you optimise, the more you’re rewarded, and the more addicting the progress becomes. In my experience anyways, I don’t know anyone outside of this forum who actually cares about such things.
If the Framework 16 could be reduced to 1W consumption, it should last 85hrs. Granted, that’s not going to happen, not in any useful capacity, powering the screen’s control board alone consumes at least 1W, but that datapoint does do a good job emphasizing what we’re giving up by not optimizing.
I am well aware which is why I am still a bit salty about the hw decoder using more power than on windows XD
My point was that the difference between the 13 and the 16 is still quite huge any I am not entirely sure why. on the 13 you can get min idle numbers ~2.5W (min brightness, wifi connected, bt connected, everything else off) and somewhat more usual ones around 3 so 6 being a good value still sounds like a lot coming from that reference frame.
Video playback still uses more power than I’d like but that got a lot better since the thing came out (ubuntu probably didn’t get most of those fixes because of the old kernel though).
That’s insane. That would imply a maximum battery life of ~22hrs which, despite the significantly reduced capacity, would be greater than the FW 16.
My first thought was that it must be the CPU, because afterall the 13’s use the low-power skus, but I don’t actually think that’s the case. I’ve seen the CPU on my 16 drop to almost no power draw, leaving the majority of the consumption up to various other components. I wish I knew what caused those differences.
Not particularly useful capacity but yeah the math checks out for the most part (theoretically 61Wh / 2.5w is 24.4h acording to my calculator but close enough).
It’s literally the same chip with a slightly different config. I can push 85W though the U just fine untill the cooler saturates XD.
The difference is likely at least partially in the other parts of the laptop like the screen, keyboard and fans and stuff (half a w here and there adds up quick, I don’t know how good the power management on the keyboard and led matrix and stuff is). The fans on the 13 at full blast pull 2W on their own. The display is also bigger so more power for the same brightness and min brightness may not be the same (actually I should probably retest the min idle numbers with the new patches that allow the screen to dim lower).
Would not be surprised if intel and amd platforms beat it there cause it’s on a much older node and may not have as advanced power management features. The perf per w definitely looses against the beast that is the 7x80u/hs, those zen4 cores are just monstrous.
Edit: I also have a consumer lenovo yoga and a t480s with pretty much the same chip and the min idle on the yoga is way higher than on the thinkpad (~2.5W vs ~5.5W). So the engeneering around the chip matters a lot.
I think a lot of it comes down to the display. Even with the screen “off” (but not unused), the display itself still consumes ~3W if I remember my measurements correctly. That consumption alone is more than your 2.5W total system draw. If I deduce that fixed 3W draw from my ~6W pull, that’d put me in 3W territory. Someone recently reported a minimum of 4.5W draw via windows on the Framework 16, which I have yet to dig into replicating on Linux.
Talking myself through it, the screen being the biggest difference makes sense both intuitively and with the data I have available.
