Hello @Caitex212, we're using one issue report per unofficially supported title. Usually I'd transfer this to #4021 or #938, but your choice to use an LLM has greatly degraded the signal to noise ratio of your feedback. Feel free to add your experience to the preexisting compatibility report(s) as it makes sense, but please skip the AI padding.
proton experimentalx1 2026-09MANGOHUDx1 2026-09PROTON_ENABLE_WAYLANDx1 2026-09PROTON_NO_ESYNC=1x1 2026-09PROTON_NO_FSYNC=1x1 2026-09PROTON_NO_NTSYNC=1x1 2026-09PROTON_USE_WINED3D=1x1 2026-09STEAM_COMPAT_TRANSCODED_MEDIA_PATHx1 2026-09xaudio2_7.dllx1 2026-09
Compatibility Report
System Information
I confirm:
steam-1222700.log
Symptoms
A Way Out: Periodic, regular-interval audio crackling, present even sitting idle in the main menu (no gameplay, no multiplayer connection required to reproduce). No PipeWire xruns are reported (
pw-topclean) while it happens. Cutscene audio progressively drifts out of sync with the video — audio and video start perfectly synced at the beginning of a cutscene, and the offset grows worse the longer the cutscene runs.Squad: Same as A Way Out
I suspect these two may share a root cause given they both appeared/are present on this GPU and both point away from a standard audio-config explanation (see below).
Reproduction
A Way Out: Launch the game, reach any in-engine cutscene, and observe audio drifting behind video over the course of the scene. Or just observe the crackle in the menu.
Squad: Launch the game and sit at the main menu; crackling is audible immediately, at a steady repeating interval.
Troubleshooting already performed (A Way Out)
MANGOHUD_CONFIG=fps_limit=60(real, hard cap, not just gamescope's-r, which does not throttle the client render loop) — drift persists identically.GST_PLUGIN_FEATURE_RANKto force software decode — no change.pw-top,pw-metadata) — no resampling/rate mismatch.dmesg | grep -i tsc).XAudio2_7.dllis loaded native (not FAudio) — rules out FAudio reimplementation bugs.Haze1/Content/Movies/splash_awo.mp4); in-game "cutscenes" are rendered live by the engine (UE-based), not pre-rendered video, soSTEAM_COMPAT_TRANSCODED_MEDIA_PATH/fozmediav1cache is a red herring for the actual gameplay cutscene drift (though the splash file itself does showFailed to find H264 bitstream filter/ missing AACcodec_dataerrors in the log — possibly a separate, minor issue, sincesteamapps/shadercache/1222700/never gets created on this system despite normal launches and "Verify integrity of game files").PROTON_NO_NTSYNC=1 PROTON_NO_FSYNC=1 PROTON_NO_ESYNC=1, and ntsync alone disabled) — no change to the drift in any configuration. Default/active backend is ntsync (ntsync: up and running.in log).PROTON_USE_WINED3D=1to bypass DXVK entirely — no change.perf record -gprofile during active drift: 94.6% of sampled CPU time attributes toRenderThread 1, almost entirely inside the kernel'spi_state_update_owner(priority-inheritance futex) reached via repeated syscalls — but this did not correlate with any actual fix when sync backends were swapped, so it may be incidental rather than causal. Happy to attach the rawperf.dataif useful.render.explicit_sync/direct_scanouttoggled, withPROTON_ENABLE_WAYLANDforced to 0 (XWayland) and left default, and with VRR disabled.Troubleshooting already performed (Squad)
pw-topERR column clean) during crackling.schedutilandperformance— no change.taskset -c 0-9 %command%) — no change.