Measuring Linux Input Latency: X11 vs Wayland, VRR, DXVK
A custom click-to-photon device tests X11 vs Wayland, VRR, and DXVK low-latency on Linux, challenging widely repeated gaming latency myths.
A developer built a custom click-to-photon latency measurement device to test long-standing claims about Linux gaming performance: an Adafruit QT Py RP2040 acting as a USB HID mouse, paired with a photodiode sampling screen brightness changes roughly every 24 microseconds.
Using this open-source rig, the author compared X11 versus native Wayland, VRR on/off, and the dxvk-low-latency fork on/off, all running Diabotical (DirectX 11) through Proton on CachyOS. Hardware included an RTX 4070 Super and a 500 Hz QD-OLED monitor, with 100-click test runs per scenario.
The results challenge the popular advice that "Wayland has bad input lag, use X11 instead": the eight main test cases landed within just 0.72 ms of each other (medians 4.21-4.93 ms). XWayland, however, added 3.13 ms over native Wayland, and in uncapped frame-rate scenarios the dxvk-low-latency fork reduced latency by 0.84 ms.
For engineers and competitive gamers, this matters because it replaces anecdotal tuning advice with a repeatable, end-to-end measurement methodology - showing that display server choice barely matters once flip mode/direct scanout is properly configured.