Does TranslucentTB affect performance?

What we measured

The first phase (settling in) started 10 seconds after launch, so it caught the end of TranslucentTB settling in (its memory was still growing from 2.5 to 9.3 MB during it). After that, CPU time only moves when something on screen changes, and even then it stays tiny: 78 ms out of 30,000 ms is about 0.26% of one CPU core.

RAM: private memory vs working set

Private memory is what belongs to TranslucentTB alone; it is the number to look at if you want to know what TranslucentTB costs. It was about 9 MB once running and 14 MB in the runs where the Welcome page had been shown, because that window's interface stays loaded in memory for the rest of the session.

The working set (51–65 MB) also counts Windows libraries that TranslucentTB has loaded but shares with other programs, such as the XAML framework behind its tray menu. Those pages exist once in memory no matter how many apps use them, so adding them up per app overstates the cost.

Inside Explorer, TranslucentTB loads a small helper (ExplorerTAP, 566 KB on disk) on the Windows 11 taskbar. Explorer's own private memory was 38.8 MB before TranslucentTB started and 35.5 MB after: no increase we could see, since Explorer's memory moves by a few MB on its own.

CPU: why it is near zero when idle

TranslucentTB does not poll. Its source registers Windows event hooks (SetWinEventHook, out of context) for the things that can change the taskbar look: windows being shown, hidden, minimized, moved, created, destroyed or brought to the front, Aero Peek, and so on. It also listens for Start, Search, Task View, battery saver and display changes. Between those events its threads sleep, which is why CPU time stays at 0 ms while the screen is still. There is no polling timer that wakes it up every few seconds (the only timer in the app delays reloading settings.json by 200 ms after you save it).

When an event arrives, TranslucentTB works out which state applies to each monitor's taskbar and, if the look changed, tells Windows once. That is a few milliseconds of work per window change, which matches the 78 ms we measured over 18 full open-maximize-restore-close cycles.

GPU and drawing the effects

TranslucentTB does not render anything itself. On the Windows 11 taskbar it swaps the taskbar's background brush; on Windows 10 it sets the taskbar's accent policy. Either way Windows' own compositor (DWM) draws the result, the same way it draws Windows' default translucent taskbar. We did not measure GPU load: our test machine has no graphics card. Clear and opaque need no blurring at all; blur and acrylic ask the compositor to blur the area behind a strip about 48 pixels tall.

Games and FPS

We have not benchmarked games, so we will not quote an FPS number. What the design tells you:

• A game running full screen covers the taskbar, so there is nothing to draw behind it.
• TranslucentTB uses CPU only when windows change. A game that stays in front generates almost no events.
• If you play in a window and use dynamic states, opening a launcher can switch the taskbar look. Add the launcher to ignored_windows (see the gaming look), or exit TranslucentTB from the tray while you play if you want to rule it out completely.

Laptops and battery

Because it sleeps between events, TranslucentTB itself is not a meaningful battery drain. If you want to reduce work for the compositor on battery, enable the Battery saver state with an opaque color: it applies automatically whenever Windows battery saver is on and takes priority over every other state (ready-made settings). We did not measure battery life.

If the taskbar lags or TranslucentTB is "not responding"

Lag is not normal for TranslucentTB. The usual causes are a conflict with another tool that also changes the taskbar (Windhawk mods, StartAllBack and others), or Explorer restarting repeatedly. Start with conflicts and Explorer crashes, and press Ctrl+Alt+Shift+F1 to reset TranslucentTB's state if the taskbar looks stuck.

How we measured

All numbers come from TranslucentTB 2026.2 (the official MSIX package) on a fresh GitHub Actions windows-2025 machine: Windows Server 2025 build 26100, which uses the Windows 11 24H2 taskbar, with 2 virtual CPUs and no graphics card. A PowerShell script read TotalProcessorTime, PrivateMemorySize64 and WorkingSet64 from the TranslucentTB process once per second for each 30-second phase. In the busy phase, a background PowerShell job in the same desktop session started Notepad, maximized it, restored it and closed it in a loop (18 cycles), with dynamic states enabled so every change required TranslucentTB to recompute the taskbar look. The full method, screenshots and raw files are in our test report.

Limits: one machine, a virtual one, one monitor, and short phases. A busy PC with many monitors and windows will see more events, but the per-event cost stays the same.

FAQ

Something wrong or out of date? Email hello@translucenttb.dev and we will fix it.