Assuming this isn’t against the rules, I am reposting this from my Reddit.
I have my 12th gen (i5-1240p, running Windows 11, with a used 250 gb 960 evo ssd), and I’ve answered some questions about it in a previous post. Last night, I tested power draw in sleep mode. The results were a 5% drain over > 5 hrs (between 10 and 3 am). After that, if I am not mistaken, it may have automatically entered some hibernation, as the battery did not appreciably decrease. This result seems good, but I don’t know what the expected drain would be.
Sleep standby is only a small part of overall battery consumption.
I see on the forums that there is a full guide for Windows battery optimization made for the 11th gen. Is there any similar resource made/planned for the new model, or even general tips? I am a novice at this type of optimization, though I have used Throttlestop before. Knowing what apps can help, and what needs to run on the background (i.e. Intel Graphics Command Center), how to dynamically tweak BIOS, expansion card management, etc. would be very helpful. An official guide would be amazing.
Edit: Added sleep battery draw chart
Assuming your port config of 2 USBC, 1 USBA, 1 HDMI, this is improved from 11th gen where we’re seeing 2-3%/hr. However this is still far behind non-Framework ultrabooks which drain 0.2-0.5%/h (10x improvement)!
I am indeed. As I stated above, I am a complete novice. Why don’t the expansion cards shut off during sleep mode? If I’m not mistaken, it is fairly easy to sleep different ports/components on a computer in power settings. Why doesn’t this happen by default? Should I stop it?
My chart shows the battery practically stopped draining after 3 am, staying on 95% throughout. I don’t fully know how deep sleep works. I assumed that it would have just switched to hibernation, which should cause 0 power drain, but with that in consideration, it may be much more comparable to modern ultrabooks.
Another question(s): I know sleep mode sips power because it keeps the volatile memory (RAM) powered. Does the framework using SODIMMs instead of LPDDR4 mean battery drain in sleep will always be worse than comprable ultrabooks?
Hibernation, by comparison, transfers the RAM contents to non-volatile storage, shuts off completely, and restores the content on startup. I know there is a slight delay when starting, but this isn’t much of a problem with modern SSDs. Is the main reason not to use it almost exclusively then power draw on startup? Is there a point where it is more efficient than sleep?
I always thought that was the biggest reason for the massive gap between the battery life of the Framework and other modern Windows ultraportable. Honestly, if that isn’t the issue, it’s a bit frustrating it can be so bad.
Is the only real barrier the expansion cards then? Does the size of the developer and scope of the optimizations for the CPU, iGPU, screen, and other components have a role too?
Even the expansion cards don’t justify the poor life. Would it be too difficult to automatically shut off the recessed usb-Cs connected to certain cards during everyday usage or even with manual hotkeys? Ideally framework would include a configuration utility in the driver package or at least for download. I don’t know how easy that would be for a third-party app, but seeing as the user can already manually edit them in control panel, it doesn’t seem impossible.
The HDMI card could only receive any power when charging or when the project button is hit, for example. Or each bay could be controlled manually and be key assigned.
Is there any actual HARDWARE barrier to better battery life?
Good questions, few answers. You’re correct, even with only 1 USB-C card connected I’m still seeing 5x higher drain than other laptops.
During standby, the display shouldn’t be drawing power and the CPU should be in an low-power state. I’m also not sure about on-the-fly powering up and down the mainboard ports. Root-causing the source of these drains is difficult:
I am kind of speaking in general, not just during standby. It seems like this has been in-the-works for a while. Whatever progress has been made has if I’m not mistaken been minimal enough the average user hasn’t had times improved significantly since launch.
And while root-causing is difficult, especially for the end-user, the engineers and developers, if sufficient time is budgeted, should be able to make significant progress unless it truly is a hardware flaw of modular computers or the current framework design.
As the only real negative compared to other Ultrabooks, it seems like a pretty big barrier to higher sales. The review scores, especially for 12th gen, seem much lower than they would be with a competitive battery life, and I wouldn’t be surprised if it leads to significant number of users holding back or not truly considering it in the first place.
If the problem is fixable, there should be a massive sales incentive. It is the only thing stopping me recommending it to…all the students I know.
There was a conscious tradeoff made by choosing to use a higher-resolution display instead of the standard 1080p. This is probably the largest contributor to battery drain, and it has similar battery life to other similar-size laptops with similar high-res screens (e.g. ThinkPad X1 Titanium or HP Spectre 14) which I think is acceptable.
Other factors like expansion cards and socketed memory also contribute to drain, but IMO Framework, being champions of reparability/upgradability, rightfully prioritized these features over battery life. Intel 12th certainly didn’t help things either, being less efficient than 11th. If Framework were to ever release an AMD mainboard, I would expect battery life to improve.
wrt engineering effort to improve–Framework is very small and they probably only have 1-2 people with the ability to do this complex power analysis and troubleshooting. We’ve already seen efforts on this end, with firmware fixes for drain during shutdown and (as you experienced first-hand) improved standby drain vs 11th. IMO they’re doing great with the resources they have.
Personally for my use-case the battery life is acceptable but I agree that it is definitely a con compared to the competition.
That’s completely fair. Maybe a lower-res screen or more efficient could be offered in the same chassis, but I am pretty satisfied with the current panel.
Assuming I don’t completely misunderstand how this works though, I still think the problems with the expansion cards could be resolved since usb can be switched off manually in the OS.
For my use case, I won’t be upgrading the RAM before I either I buy a new motherboard in several years or Framework goes caput and I buy a new laptop.
Considering this, it is a bit frustrating power-hungry SODIMMs are used on top of a replaceable motherboard, but I accept the trade off if the draw is fairly minimal, and the point of the modularity is that even if I don’t need it, someone else does.
Hopefully, programs like throttlestop and CPU power draw adjustments will be better optimized/documented for 12th gen as the generation continues. Until then, I’ll get a power bank and stop sweating the small stuff.
Can you see what the battery report from Windows 11 says? It’ll give fairly accurate hours / minutes spent in standby / sleep, and the energy used during that period.
Let’s assume it’s 5% for 5 hours (and not 5.95% for 5 hours)…due to the lack exact data at the moment. Then that’s around 2.75Wh over 5 hours…that’s 0.55W. Way better than the 1.45W I’ve been seeing with the 11th Gen (1 x USB A, 1 x HDMI, 2 x USB C).
@Kieran_Levin
If the 12th gen mainboard does in fact have improved sleep energy efficiency, can it be ported back to the 11th gen mainboard?
I might be reading this wrong, but I think it took 4 hours 8 seconds to drain 5% (2,495 mWh) and suspend fully.
Also, it seems to say my battery capacity has dropped nearly 2%/1000Mwh? Surely I’m reading it wrong; that level of deterioration isn’t normal. If it continues linearly, the battery wouldn’t hold a charge after less than a year.
So that’s just under 0.65W. This is really good for a [2 x USB C, 1 x USB A and 1 x HDMI] arrangement. It’s more than 50% improvement than an 11th gen board.
Question then (to Framework): What would it take to have this improvement made to the 11th gen mainboard?
