Measured, not claimed
What a notch app actually costs you
“Zero battery impact” is easy to write and hard to trust. Here are the numbers, how they were taken, and how to check Notchy yourself.
Light enough to forget it's running
Notchy 1.0.108 measures 0.9% of one core — and that was taken while the Mac was busy, so idle is lower. Native SwiftUI, no Electron, and anything you switch off stops costing you entirely.
Idle CPU across notch apps
Idle · Apple M2 · macOS 26.5 · May 2026Method: each app run alone, no user input, CPU summed across its processes and averaged over several 10-second idle windows. Paid apps measured in their trial build. These values sit near the measurement floor — the takeaway is that native notch apps (Notchy included) are all very light at idle, while the Electron-based one costs noticeably more. Reproduce it yourself: Scripts/measure_idle_cpu.sh samples both Notchy processes over fixed windows, discards the first (it carries a since-launch average), reports the median, and refuses to print a publishable number while a build or another busy process is running.
Shipping build, re-measured: Notchy 1.0.108 held a median of 0.9% of one core (Apple M2 · macOS 26.6, August 2026) — app plus widget extension, with the machine under active load rather than idle. It is quoted separately from the chart above precisely because it is not an idle figure, and comparing it to those idle numbers would be unfair to them.
Verify it in two minutes
- 1. Quit every other notch or menu-bar utility, so you are measuring one app rather than a stack of them.
- 2.Leave the notch closed, stop any media, and don't touch the trackpad — idle is the state that matters for battery.
- 3.In Activity Monitor, sort the CPU tab by % CPU and watch Notchy over a minute. Then check the Energy tab's 12-hour impact after a normal day.
- 4. Compare against another notch app the same way. Run each alone — two apps polling the same system state at once tells you nothing about either.
Where the cost actually comes from
Polling on the main thread. The expensive pattern is a timer that asks the window server or the accessibility API for something every couple of seconds, synchronously, on the thread that also draws the UI. It shows up as both CPU and as interface lag — Notchy shipped that bug, measured it, and moved the work off-main with a longer cadence.
Animating when nobody is looking. A per-frame gradient or a spring that never settles costs the GPU continuously. Notchy stages material changes and stops decorative motion when the panel is closed or the Mac is in Low Power Mode.
A web view in a menu-bar app. The single biggest difference in the table above is native versus Electron. There is no tuning that closes that gap.
Features you don't use. Turning something off in Notchy tears its timers down rather than hiding its UI, so an unused feature costs nothing at all.
Battery & CPU — FAQ
Do notch apps drain your battery?
A badly built one can. The cost is never the notch itself — it is what the app does while nothing is happening: polling system state on the main thread, animating a gradient every frame, or running a web view. Native apps that idle properly are effectively free; Electron-based ones are measurably not.
How do I check what Notchy is costing me right now?
Open Activity Monitor, switch to the Energy tab, and look at Notchy's 12-hour power impact with the notch closed and no media playing. That is the number that matters for all-day use — a momentary spike while an animation runs is not what drains a battery.
Why did some notch apps get accused of CPU spikes?
Usually a hot polling loop. Notchy hit exactly this class of bug twice and fixed both: an accessibility hit-test grid running every two seconds on the main thread, and synchronous window-server calls inside a poll. Moving them off the main thread and stretching the cadence took idle CPU from several percent to under one.
What does Notchy do differently in Low Power Mode?
Decorative animation is parked, timeline views drop their refresh cadence, and sampling only runs while a view is actually visible. On battery, the app deliberately does less rather than looking the same and costing the same.
Are these numbers a precise ranking?
No, and it would be dishonest to present them as one. Idle CPU for native notch apps sits near the measurement floor, where run-to-run noise is a large share of the value. The honest reading is: every native notch app is light, Electron ones are not, and Notchy is in the light group.
Does Notchy keep polling when I'm not using it?
Only for what you have switched on, and only where the data can't arrive as an event. Features you disable stop their timers entirely rather than idling — the toggles are real teardown, not just hidden UI.