[TRACKING] Linux battery life tuning

I can get it with “inxi -Da | grep rev” on my system.

I had to run this as root using sudo.

@nrp
SK Hynix P31

inxi - Da output:
Drives:

model: SHGP31-500GM-2

serial: CJ09N90041050A*** (user redacted)
rev: 41060C20

I assume rev is the firmware revision, but I don’t know for sure.

I did not take awesome metrics with and without nvme.acpi, but I believe having this on cuts s2idle standby drain by 60-80%. I’ll try to get to doing some more specific metrics.

On Slackware Linux here with powertop (i5 11th gen, 32GB RAM). I have 2 USB-C and 2 USB-A currently installed, and 4.5 hrs to 7.5 hrs (I am yet to test the full 10 hours 20 minutes that it shows me).

However I only see a marginal/negligble increase in battery life on stand by with “nvme.noacpi=1”.

model: WDS100T1X0E-00AFY0
rev: 614900WD

I’m using “Vitals” Gnome extension, resuming from s2idle I can briefly see the drain rate during that time (above the dropdown box in the panel) but I calculate it from the total “Energy (now)” before sleep and after.

2 Likes

SSD: SN850 1TB
SSD firmware version: 614900WD
Framework laptop BIOS version: 3.07
OS: Ubuntu 22.04 LTS
Kernel version: 5.15.0-35-generic
Kernel command line: quiet splash nvme.noacpi=1
Sleep / suspend method: systemctl suspend
Sleep / suspend duration: 8 hours
Battery charge lost / drained: 33%

1 Like

@A_Fan - that drain seems high. What modules do you have installed, and what sleep mode (s2idle / deep) do you have set?

For reference, I’m running manjaro, sk hynix gold p31 ssd, kernel 5.17.9-1 ,4x usb-c, and I am losing under 10% overnight.

Cards removed to limit the variables.

Sleep mode, whatever is OOTB with the Ubuntu 22.04 LTS. No change there.

Mind you, I seem to see two different drain rates on my system…even with no hardware changes, nor software configuration changes. The only thing that changed is how many times the system has been placed into suspend since it first powered up.

On the first suspend, it was draining maybe 1-1.5% an hour… However, on subsequent suspend, it was draining like 4% an hour.

I would guess that it is s2idle. If you already know the following, skip it. If not, you can check what you currently have in place by running cat /sys/power/mem_sleep. If you want to set it to deep, add “mem_sleep_default=deep” to the grub config and update grub.

Thanks for pointing this out. I’ll take another look after my sleep test.

I feel that someone might want to write a script that collects and shows what/which battery related optimization is active / inactive. Something like what the Get-SpeculationControlSettings cmdlet (on Windows) would do…to summarize a system’s spectre fixes.

1 Like

SSD: WD_BLACK SN750
Firmware version: 111130WD

Isn’t that what the acpi_osi='!Windows 2020' kernel command line option helps for? Mind you: some quoting is required to protect the ! and some tools rely on quotes in the grub configuration as well. For instance, Fedora uses /etc/default/grub where GRUB_CMDLINE_LINUX_DEFAULT has its value delimited by "...". A tool like “grub-customizer” messes up the quotes. I think the '...' delimiters are valid too, but I’m happy to be corrected.

1 Like

To clarify, with that firmware, are you seeing differences between having nvme.noacpi=1 enabled or not?

1 Like

Yes, huge difference. It lasts much longer on standby.

I’ve tried auto-cpufreq for the past 3 weeks and I seem to be getting better battery life, about 1 - 2 hours more. I don’t have time for now to do a thorough test, but I am pretty sure that using only auto-cpufreq is better than using tlp. I hope someone can confirm this through testing.

1 Like

Does it improve the battery life, if I remove components that I will not use?
Since you can easily remove or add components this should be considered.

Removing camera, Audio/microphone, fingerprint reader, Wifi module or mousepad could help some users to save energy.

Does everything help, or only some things?
Can some bugs appear if I have a component removed? Can i make some simple changes to prevent these bugs?

Thank you already for replies!

removing the wifi card potentially could save power, as will removing unused active expansion card modules (active = not usb-c). I believe the camera and microphone switches cut power to the respective modules.

1 Like

Some you can just disable in the BIOS, I imagine the laptop also wouldn’t look very good with the fingerprint reader removed. Speakers are definetly removeable though. I don’t see how removing these would save a lot of power though, as they can just be hardware disabled. Although it definetly makes the laptop lighter which is a plus! (You would not believe how light the laptop feels without the battery).

2 Likes

I wouldn’t go that far to save battery life, I imagine the savings would be minimal unlike just removing the swappable modules on the laptop.

1 Like

Hi, I’m experimenting right now on the best setup for my framework laptop in terms of battery life and memory. I’m currently averaging 4.5W on idle and 900 MiB of RAM on a fresh boot, I would like to ask how much are you guys averaging in idle? Please also post your specs so I can compare with mine, thanks!

My specs: i5, 64GB, 3 USB-A Expansion cards, and a USB-C expansion card.
Running on Arch 5.18.3 with LeftWM and Polybar. (Planning to hop on GNOME, as I’ve already tried KDE. Also will try to install Void again soon but I just cant take it with the servers being so far from my country, thus the very bad download speeds that I get on Void.)

1 Like

2.5W with wifi connected (bluetooth disabled), nothing open, screen brightness at minimum and KB light off, light usage is between 6-7W.

2xUSB A (each adds about 300mA drain) and 2x USB C, 2x 8GB ram, i5 1135G7, Samsung 980 pro 1TB

Fedora 36 with gnome - uses a fair amount of ram maybe 1.7GB on fresh boot but I have not kept a close eye, more interested in energy usage.

I use TLP to reduce power consumption.

1 Like