Quest 3 PC Link — Video Stream Stall While Tracking Path Remains Active

Summary

Severity: High — prevents reliable extended PCVR use
Affected product: Meta Quest 3 / Wired PC Link
Connection: USB 3 wired Link
Current Link version: 208.0.0.57.535
Previously affected: 207.0.0.133.434
Quest OS: v2.7 — Build 5243367.3600.520
NVIDIA driver: 591.86 WHQL

One-line failure description

During an otherwise functional wired Link session, the Quest display stops receiving/updating PCVR video frames while headset tracking continues to be delivered to the PC. The PC application/game continues running.

This appears to be a partial Link failure rather than a complete USB disconnect or complete loss of the headset session.

Test Environment

1. Expected Behavior

During a wired Link session:

1. PCVR rendering continues on the PC.
2. Encoded video frames continue to be delivered to the Quest.
3. Headset tracking continues to be transmitted to the PC.
4. The Quest display continuously presents the current PCVR frame.

If a transport problem occurs, I would expect the Link session to either recover automatically or transition to a clearly disconnected state.

2. Actual Behavior

After an unpredictable period of normal operation, the Quest display suddenly stops updating.

At the moment of failure:

• the PC application continues running;
• Automobilista 2 continues running on the PC;
• the PC continues receiving headset movement/tracking;
• moving the headset changes the camera/view on the PC monitor;
• the Quest display remains frozen on an old frame;
• audio may continue;
• the headset is not behaving like a completely disconnected USB device.

In some failures, the frozen Quest image also develops severe visual artifacts, including colored/gray edge corruption and chroma-like distortion.

Critical observation

Tracking continues while the video stream appears to have stopped.

This distinction is important because it suggests that the failure may be occurring in the PC → Quest video path rather than at the physical USB connection as a whole.

3. Reproduction Pattern

The failure is intermittent with respect to time-to-failure.

Observed sessions have ranged from only a few minutes to approximately 20–60+ minutes before failure.

A typical reproduction sequence is:

1. Connect Quest 3 through wired USB Link.
2. Establish normal PCVR Link session.
3. Launch Automobilista 2.
4. Enter a VR session.
5. Run normally at 90 Hz.
6. After an unpredictable period, Quest display freezes.
7. Verify that the PC game is still running.
8. Move the headset.
9. Observe that headset tracking is still being received by the PC/game.
10. Observe that the Quest display remains frozen.

The failure therefore does not require a game crash to occur.

4. Post-Failure Recovery Behavior

There is another potentially useful diagnostic observation.

After the Link video failure, launching the Meta USB Cable Test can cause the Meta Horizon PC application itself to close/restart.

After the application restarts:

1. Meta Horizon becomes available again.
2. The USB cable test can then report a successful USB 3 connection.
3. Link can be established again.
4. The headset can return to the PCVR environment.

This suggests that the underlying USB connection may still be usable after the video-stream failure, while the Meta Link/OVR software stack may be left in an abnormal state.

5. USB / Cable Checks

The physical USB connection passes Meta's cable test.

Observed cable-test throughput:

~2.1–2.5 Gbps

The connection is recognized as USB 3.

USB selective suspend has also been disabled.

Multiple USB ports/controllers have been tested.

The important point is that the headset can establish a fully functional Link session and remain operational for a significant period before the failure occurs.

Therefore this is not simply a case of Link failing to detect the headset.

6. Version / Regression History

The problem appears to correlate with the Meta Horizon PC software/Quest software update period.

The system previously operated reliably with wired Link for extended PCVR sessions.

The problem became substantially more frequent around the following Meta PC client versions:

205 → 206 → 207

Version 207 was particularly problematic on this system.

The system has now automatically updated to:

Meta Horizon PC 208.0.0.57.535

The failure is still reproducible.

I am therefore interested in whether there is a known regression affecting the Link video pipeline across these releases.

7. NVIDIA Driver Testing

NVIDIA driver changes have also been tested, including clean driver installation/rollback procedures.

The current test configuration is:

NVIDIA 591.86 WHQL

With this driver, some sessions can last substantially longer, but the fundamental Link failure can still occur.

This makes it difficult to attribute the complete problem solely to the NVIDIA driver.

8. Meta Link Settings Tested

The following Link/ODT parameters have also been varied during troubleshooting:

• 90 Hz
• Link resolution
• H.264 / H.265
• encoder bitrate
• encoder width
• sliced encoding
• Link sharpening
• ASW settings
• latency/performance settings

Changing these parameters can alter performance or time-to-failure, but no configuration has provided reliable long-term stability.

9. Independent Comparison — ALVR

The same:

• Quest 3
• PC
• USB cable
• Windows installation
• NVIDIA GPU
• PCVR workload

has also been tested using ALVR/SteamVR.

ALVR has exhibited USB disconnect events during testing as well, so I am not claiming that the physical USB path is completely fault-free.

However, the characteristic Meta Link failure described above — specifically the Quest display freezing while tracking remains active and the PC application continues running — is much more strongly associated with the Meta Link software stack.

This makes it useful to investigate the Meta Link video/transport pipeline independently from the game.

10. Why This Does Not Look Like a Normal Game Crash

At the moment of failure:

PC side

• Game process remains alive.
• Rendering continues.
• Headset tracking continues to affect the game camera.

Quest side

• Display stops updating.
• Previously received image remains visible.
• Link session does not behave like a clean complete disconnect.

Therefore the failure boundary appears to be somewhere between:

PC rendering → encoding → Link transport → Quest video decoding/display

rather than:

game → complete application crash

11. Working Technical Hypothesis

This is only a hypothesis, not a claim about the implementation.

The observed behavior appears consistent with a failure in one of the following stages:

PC frame submission
→ Meta/OVR capture
→ hardware video encoding
→ Link transport
→ Quest decoding/display submission

while the reverse path:

Quest tracking
→ USB/Link transport
→ OVR runtime
→ PC application

remains operational.

The fact that tracking survives the failure is the main reason I believe this distinction may be useful to engineering.

12. Suggested Diagnostic Information

If Meta engineering can identify the affected component, the following would be particularly useful to determine whether this is a known issue:

• OVRServer/OVR runtime logs immediately before and after the failure
• Link transport errors
• encoder initialization/reinitialization errors
• encoder frame submission failures
• dropped/stalled video-frame counters
• USB transport errors
• Quest decoder errors
• compositor/frame-present errors
• GPU/NVENC-related errors
• watchdog/reset events
• state transitions of the Link session following the video stall

The failure timestamp can be provided for individual sessions if required.

13. Engineering Questions

Could the Meta PCVR/Link engineering team please confirm:

1. Is a failure mode where Quest tracking remains active but the PC → Quest video stream stops updating currently known?
2. Is this behavior currently under investigation in Link v207/v208 or recent Quest firmware?
3. Are there known regressions involving RTX 50-series GPUs / NVIDIA NVENC and recent Link releases?
4. Could the Link runtime enter a partially alive state in which the tracking channel remains operational while the video channel has stalled?
5. Are there specific logs or diagnostic counters that users can collect to distinguish a USB transport problem from a Link encoder/video-pipeline failure?
6. If a known regression exists, is a temporary rollback to a known-stable Meta Horizon PC version available or planned?

14. Reproduction / Debug Data Available

I can provide:

• complete PC configuration;
• exact Quest OS build;
• exact Meta Horizon PC version;
• NVIDIA driver version;
• ODT configuration;
• cable-test results;
• failure timestamps;
• Meta/OVR logs;
• additional controlled tests with different codecs/bitrates/resolutions;
• additional results from ALVR using the same hardware.

I would be happy to run specific diagnostic tests requested by Meta engineering if they can help isolate the failing component.

Final Note

The most useful observation from this report is:

When the Link failure occurs, the Quest display stops updating, but headset tracking continues to reach the PC and the PCVR application continues running.

This appears to be a potentially important distinction between a complete USB disconnect and a stalled PC → Quest video path.

Any confirmation from the Link engineering team that this specific failure mode is already known/investigated would be appreciated.