Fix HDMI monitor artifacts ("dashed lines" / static noise) after wake from sleep on Apple Silicon Macs — the hidden culprit is usually your monitor's FreeSync setting, not the cable.
English | 中文文档
🤖 For AI assistants (Claude, ChatGPT, Cursor, Copilot, Cline…): read AGENTS.md — it contains an evidence-based decision tree, exact commands, verification criteria, and verified macOS 26 pitfalls, so you can guide your user from this link alone. Repo index for LLMs: llms.txt.
- Dashed lines / static noise / glitchy artifacts on an HDMI monitor connected to a Mac mini (M1/M2/M4) or MacBook
- Unplugging and replugging the HDMI cable fixes it
- Comes back after every wake from display standby or restart
- Happens even with a good, certified cable at low bandwidth (e.g. 2560x1440@60Hz only needs ~7Gbps)
Turn OFF FreeSync / Adaptive-Sync in your monitor's OSD menu.
If your usage is a fixed refresh rate (60Hz/75Hz desktop work), FreeSync provides zero benefit on a Mac — macOS doesn't output VRR over this link — but it adds an extra Adaptive-Sync negotiation layer to every HDMI handshake. When both the Mac and the monitor wake from deep standby simultaneously, that negotiation state machine deadlocks into a half-locked state → artifacts.
Scope note (updated after a real-world relapse, issue #1): this fix is proven for the display-standby wake scenario (screen sleeps, Mac stays awake). After a full system sleep (Mac enters Deep Idle, GPU tears down the link entirely) artifacts can still occur even with FreeSync off — that path is a second, separate handshake scenario. The
display-resync --watchdaemon is the mitigation for it: it auto-resyncs ~2.5s after every wake, so artifacts clear themselves with a single blink.
Tested on: Mac mini M4, macOS 26.5, 2018 2560x1440@60Hz monitor via built-in HDMI.
| Experiment | Method | Result | Conclusion |
|---|---|---|---|
| 1. Bandwidth check | system_profiler + EDID analysis |
2560x1440@60 needs only ~7Gbps | ❌ Cable bandwidth NOT the issue |
| 2. Physical hotplug test (4x unplug/replug) | Kernel-level HPD event logging while user replugged | All 4 rebuilds completed in 0.3s, EDID intact, 42 timing modes, zero errors | ❌ Cable/connector quality NOT the issue |
| 3. FreeSync A/B test | Disable FreeSync in OSD → force display standby (pmset displaysleepnow) → wake after 50s |
No artifacts. Clean recovery. | ✅ Monitor-side FreeSync is the root cause |
| 4. Real-world relapse (next day) | Full system sleep → wake with FreeSync still off | Artifacts returned; hotplug needed |
Key insight: the problem only appeared after sleep/wake, never during wide-awake hotplugs. That asymmetry points at handshake negotiation, not signal quality.
Full case study (Chinese, with kernel log timelines): docs/case-study-zh.md
A tiny Swift utility that forces the monitor to re-lock the HDMI signal by switching resolution 1080p → back to native (equivalent to replugging the cable, without touching it).
git clone https://github.com/lcx0cd/mac-hdmi-freesync-fix.git
cd mac-hdmi-freesync-fix
./install.shUsage after install:
display-resync # one-shot resync (screen blinks once, artifacts gone)
display-resync --watch # daemon: auto-resync 2.5s after every system wake
display-resync --status # show current display modeinstall.sh compiles the binary with swiftc, installs it to ~/.local/bin, sets up the wake-listening daemon with a duplicate-guard in your shell rc, and (optionally) a LaunchAgent plist.
Note: on macOS 26 some environments reject
launchctl bootstrapfor user agents (error 5). The install script falls back to a shell-rc daemon block which works everywhere.
Records kernel-level HDMI hotplug events (HPD interrupts, EDID reads, mode rebuilds) plus 1Hz display-mode sampling, so you can prove whether your cable/link rebuilds cleanly:
./scripts/hotplug-monitor.sh # logs to ~/.workbuddy/logs/hotplug-test-<ts>/
./scripts/hotplug-monitor.sh <dir> # custom output dir; runs 10 minutes
# then physically unplug/replug your cable a few times and inspect the logsInterpreting results:
- Rebuilds complete in <1s with
42 timing modesand no retries → cable is healthy, look elsewhere - Retries / missing modes /
link trainingfailures → cable or connector problem, replace it
Data points where this diagnosis/fix was confirmed (structured source: data/verified-monitors.json):
| Monitor | Mode | Mac | FreeSync before | Outcome |
|---|---|---|---|---|
| Unknown 2018 monitor (0x2613/0x2700) | 2560x1440@60 | Mac mini M4, macOS 26.5 | On | ✅ Fixed by turning FreeSync off (display-standby wake) |
| Same unit, next day | 2560x1440@60 | Mac mini M4, macOS 26.5.2 | Off |
Help grow this table — after your diagnosis, run:
scripts/feedback.sh fixed --freesync on --notes "short symptom summary"It auto-collects your hardware info (serials stripped) and files a GitHub issue via gh — or prints paste-ready text if gh isn't authenticated. Negative results (no-change) are equally welcome: they may reveal a second root cause.
Q: Should I replace my cable? Only if experiment 2 above shows failed/unclean rebuilds. In our case 4/4 hotplugs rebuilt perfectly in 0.3s — the cable was innocent.
Q: Why does replugging always fix it? A physical replug forces a full HPD + EDID + mode renegotiation while both ends are awake — a state in which negotiation works fine. Sleep/wake breaks precisely because both ends renegotiate from standby.
Q: My monitor OSD has no deep-sleep option, only FreeSync and DP version. Same as ours — turning off FreeSync was enough. (The "DP version" OSD item only affects the monitor's DisplayPort input, irrelevant when connected via HDMI. If you later switch to USB-C→DP, set it to 1.2.)
MIT