Closed eslerm closed 5 months ago
eslerm
commented
5 months ago Running the M2 with macOS 14.0 for 8.5 hours in sleep mode reported no discharge using pmset -g batt. So, macOS is running close to 0% discharge per hour during sleep and Linux is ~2%.
(All other OS' were fresh installs of latest releases, with wifi enabled, and packages updated.)
eslerm
commented
5 months ago Running a fresh install of Ubuntu Asahi and Sway on the same M2, I disabled the display with wlr-randr and let the computer idle for 9 hours. This drew 1.86% charge per hour.
It feels like the suspend state is not being entered during sleep. In that case, less power will be used during "sleep" if wifi is disabled.
From earlier tests, there were no logs during sleep, but some logs are posted to journalctl after wake related to going into sleep: https://pastebin.com/9d4A9N58
For whatever reason, 1 hour tests showed no battery draw earlier. Please attempt 3+ hour tests to verify this issue.
eslerm
commented
5 months ago @marcan stated that this is not a regression.
To verify I installed an old version of Asahi (Ubuntu Asahi Kinetic built on 10-May-2023) and ran it over night. This drew 1.77% charge per hour, or ~20% after ~12 hours.
Closing this issue, as I've treated this as a regression, and opening a new feature request.
Ubuntu Asahi has received multiple reports of large battery use during sleep. Overnight battery loss is ~20% charge.
To test this, I put a 2022 M2 Air into sleep mode under different conditions. Battery loss per hour was roughly the same running Ubuntu Asahi Gnome, Ubuntu Asahi Sway, and Fedora Asahi Remix KDE Plasma (1.93%, 1.99%, 1.80% battery loss per hour respectively).
Comparing this to an amd64 machine is unfair (especially with different battery capacity), but my AMD T14s g2 lost 0.40% battery per hour.
These four tests were ran for 12.25, 6.25, 9, and 48.25 hours respectively. A fairly rough experiment :)
Is 2% battery loss per hour during sleep expected on Asahi?