Whether it does or not, there is still hibernation. I know it is a pain to get working under Linux, but it does. (Ubuntu 20.04 on my laptop hibernates all the time. (encrypted))
I still don’t understand why Linux distros arbitrarily dropped hibernation support from the default configurations. Probably because Luks is a freaking mess, but I digress.
+1 definitely would like to know the status of s2 deep sleep.
I concur, 2disbetter. I have switched from Fedora to Gentoo to compile a custom initramfs just to get hibernate working with LUKS.
On Ubuntu 20.04 running 5.11
$ cat /sys/power/mem_sleep
[s2idle] deep
Tigerlake does not support S3 sleep.
However Anecdotally battery drain is around 40% over about a 20 hour period in S0ix when I run Ubuntu. Not great but not the end of the world if you plug in once a day.
Windows deals with this by hibernating as part of modern standby.
That’s about 2% per hour (1.1W), which is unfortunate to say the least (assuming that drain is linear…, 0.02/h * 55Wh = 1.1W).
However, digging into this a little, it looks highly dependent on configuration, bios, kernel, firmware, etc.
The X1 Nano (also Tigerlake but with an even smaller battery) has a thread on this (English Community-Lenovo Community). OP, is seeing 5% (2.4W) power drain, staff sees 0.8% (0.4W) power drain (the nano has a 48Wh battery).
There’s also a relevant article on Phoronix (Dell Getting Linux Power Management Optimized For Their Latest Systems + Upcoming Tiger Lake Desktop - Phoronix).
So, pulling random ideas out of a hat because I’m way out of my depth here… explicitly powering down USB/networking devices (possibly powering down the SSD?) before suspending may help? At the very least, it looks like there may be room for improvement in software.
Following up here, deep sleep is supported. Just not enabled by default:
How baffling and frustrating: it’s like Intel wants people to use ARM Macs, despite the Macs’ inadequate repairability. So we’re stuck between choosing what should be a simple and straightforward feature, in the form of functional sleep/hibernate that doesn’t impose a serious battery drain, and a laptop that can be repaired.
A Sophie’s choice, truly.
To be more accurate, you can force the activation of S2idle deep sleep on Linux, BUT the hardware does directly support it. This means while the laptop appears to be sleeping, and ‘is’ I’ve observed it using ridiculous amounts of power. Something like 15% in a 5 hour period.
As the Mandolorian say, hibernation is the way. At least for now.
s2idle and deep are different things. s2idle is the “light” sleep, deep is the power-saving one.
How did you determine that the hardware doesn’t support deep sleep? When I enable deep sleep (which the hardware at least claims to support), resumption takes significantly longer so something is happening.
I have it enabled on mine, but I have no way of knowing if it is in fact using deep sleep or not. Mine is not taking long to resume from sleep at all. Near instant.
Edit: I have verified using the below command that deep is reported. What I mean is that I have no way of knowing when I close the lid that deep sleep has actually been entered.
Fedora 35 is reporting “[s2idle] deep”.
@Kevin_Wilson In theory, echo deep | sudo tee /sys/power/mem_sleep will switch to “deep” sleep. But @2disbetter is right that the laptop is still drawing power (I haven’t tested how much of a difference it makes).
I’m not sure of the context so I just want to possibly clarify the meaning of that output. It means that s2idle is enabled, and deep is not. These two posts may be helpful:
Yes, that does work. It’s showing “s2idle [deep]” now. I haven’t checked the battery drain yet, just what the kernel reports.
hey there guys iḿ running the framework laptop on Ubuntu 21.04 and so far so good the only problem i’ve had is Deep Sleep: I have enabled and had to disable deep sleep about 3 times now, whenever i enable it and go to sleep the laptop just freezes and i can do nothing, i need to power off, reboot, and rollback the change, else it freezes again. any clues?
Kernel 5.12 is what works for me, including sleep.
5.14 doesn’t work at all - it freezes and never actually goes to sleep.
yup have kernel 5.14 i’ll downgrade and give it a try, but does it consume less batery on sleep?
I haven’t done any kind of battery test to be honest - I always have it plugged in.
These numbers are very unscientific but I use Solus with the 5.14 kernel and deep sleep works for me.
The battery drain when in sleep mode however, is not very good. I did a few tests when I was asleep where I’d charge fully and put the laptop to sleep in both s2idle and deep sleep. Deep sleep drained about 35% of the battery in 7 hours where s2idle usually either died or had only 10% battery or so remaining.
s2idle also came with the issue of the system no entering the higher idle states such as c8-c10 whenever it woke up where only a reboot seems to fix it. Deep sleep didn’t have that issue.
Using Fedora 35 (beta) with mem_sleep configured with deep enabled. Battery is at 100% as of writing this. I’m going to unplug and put into sleep mode for the night. I’ll log time of day of sleep and wake along with battery percentage. I’ll post results tomorrow morning.