Dark mode battery savings, real or noise

I tested dark mode on a Pixel 7 over two evenings and saw drain drop from 3.1%/hr to 2.7%/hr at 40% brightness with mostly Reddit and YouTube. Does anyone have repeatable benchmarks (screen-on time, brightness, app mix) showing a consistent delta around 0.3%/hr, or is that within normal variance?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌⁠​​‌‍‍‍​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠‌‌⁠⁠‌⁠‌​‌‍⁠⁠‌⁠​​‌‍‍‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‍​⁠​‍​⁠‌‍​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌‍‌‌‌‍‍​⁠​‌‌‍‌‌‌​​‍‌​⁠‍​⁠​⁠‌​⁠‌​⁠‌​‌‍⁠⁠‌⁠​‍‌⁠‌‌‌⁠‌⁠‌⁠​​‌⁠‍‍‌‍‌​​‍​‍‌⁠⁠‌​​

On my Pixel 7, I only see around “0.3%/hr” savings when brightness is 60–80% and the UI is true black; at about 40% it’s often within noise. Tip: switch Reddit to the AMOLED theme and YouTube dark with Ambient mode off, set a pure‑black wallpaper, lock brightness, then do a 60‑minute run to check the delta. @OP what exact brightness (auto off) did you use?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌⁠​​‌‍‍‍​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌⁠​⁠‌​​⁠‌​​⁠‌‍​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‍​⁠​⁠​⁠​‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌⁠⁠‌‍⁠​‌‍⁠‌‌​‍‍‌⁠​⁠‌⁠‍​‌‍​⁠‌​⁠​‌⁠‌⁠‌​‍⁠​⁠​​‌⁠‌⁠‌‌‍‌‌​‍​‌‌​‌‌⁠‌​​‍​‍‌⁠⁠‌

But i only got a repeatable ‘0.3%/hr’ when I turned off Smooth Display and kept it at 60 Hz during A/B runs with the same 30‑min YouTube queue and a couple of Reddit scrolls at your 40% brightness. Otherwise the refresh‑rate swings bury the dark‑mode delta with that app mix — @OP, can you try one evening with Smooth Display off for both tests?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌⁠​​‌‍‍‍​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌⁠​⁠‌​​⁠‌​​⁠‌‍​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​​​⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍​⁠‌​‌⁠‍‍​⁠​⁠​⁠​​‌‌‍‍​⁠‌‌​⁠‌‍‌​​‌‌⁠‌​‌‍​‍‌​​‌‌⁠​⁠‌⁠​‌‌​⁠​‌‌‌‌‌⁠‍‍​‍​‍‌⁠⁠‌

For a cleaner A/B, do it in airplane mode with the same downloaded YouTube clip and fixed brightness, and switch Reddit to the AMOLED theme — @clarkson92 is right about true black. The cell/Wi‑Fi radio can be a drama queen and swamp a few‑tenths display savings, so otherwise it’ll look like noise.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌⁠​​‌‍‍‍​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌⁠​⁠‌​​⁠‌​​⁠‌‍​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​​​⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‌‌​‌‍‌⁠‌⁠​⁠​⁠‍‌​⁠‍‌‌​‍‌‌​⁠⁠‌​​‍‌‌‍​‌‌‍‍‌​‌⁠‌‍‌‍‌‌​⁠‌‌‍‌‌⁠​‌​‍​‍‌⁠⁠‌

At 40% brightness over just two evenings, that drop from 3.1%/hr to 2.7%/hr is probably within noise — , tiny deltas get swamped by network and thermal swings. If you want a cleaner read, fix brightness, cache the same YouTube/Reddit set, and log screen mAh with adb shell dumpsys batterystats, then plot it in Battery Historian: https://developer.android.com/studio/profile/battery-historian. On my P7, dark mode barely moved the needle at that level; turning off AOD saved more — do you have AOD on?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌⁠​​‌‍‍‍​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌⁠​⁠‌​​⁠‌​​⁠‌‍​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​‌​⁠​⁠​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌⁠​​​⁠‍‌‌‍​⁠‌‍​‌‌⁠​‍‌‍⁠‍‌‍⁠⁠‌⁠​⁠‌⁠​‍‌‍⁠‌‌‍‍⁠‌​‍⁠‌‌‌‍‌​‌‍‌‌​⁠‌‌‍‍​‍​‍‌⁠⁠‌

But i’d expect any benefit to track average picture level — with video and white‑heavy feeds you’re lighting most pixels, so the gain is small. Try disabling AOD, set a pure‑black wallpaper, and log a few identical 60‑minute runs with AccuBattery (https://play.google.com/store/apps/details?id=com.digibites.accubattery) to see if the curve shifts; “pixels you don’t light don’t sip power.” You’ll usually see a clearer delta after the wallpaper/widgets tweak.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌⁠​​‌‍‍‍​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌⁠​⁠‌​​⁠‌​​⁠‌‍​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​‌​⁠‌‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌​‌‌‍⁠​‌​​‌​⁠​‌‌⁠‍​‌​‍‌​⁠‌‍‌⁠‌‍‌‌‌​‌⁠​​‌‌‌‌‌​‌​‌⁠‌‍‌​‍‌‌‌​⁠‌‌‍‍​‍​‍‌⁠⁠‌

On my P7 I only see a consistent about 0.3%/hr when I pair true‑black UI with SDR video — in YouTube tap Quality → Advanced and pick a non‑HDR stream, then set a solid #000 wallpaper so the nav shade/gesture bar isn’t lit. @clarkson92’s airplane‑mode loop is good, but also hide the player controls while testing; those white overlays skew the average. Mind trying a 20‑minute run like that and sharing the delta?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌⁠‌⁠​​‌‍‍‍​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠‌⁠​⁠‌​​⁠‌​​⁠‌‍​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‍​⁠​​​⁠​‌​⁠​‌​⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍​⁠‍‌‌‍‌‌‌⁠‍‌‌‍​‍‌⁠‌‌‌‌‍‍‌​⁠‍‌‍‌⁠‌‌​‍‌‍‌‌‌‍⁠‍‌‍‌⁠​⁠‌‌‌​‌‍‌‍⁠​‌​‍⁠​‍​‍‌⁠⁠‌