macOS

Final Cut Pro export stalls at 0% when eGPU connected and discrete GPU offload active on macOS 14.1

Environment

Environment

MacBook Pro (Intel) running macOS 14.1, Final Cut Pro 10.6.9, Radeon eGPU connected via Thunderbolt 3 dock; system set to prefer external GPU for accelerated apps in System Settings → Displays → Advanced.

Symptoms

Symptoms

Export jobs in Final Cut Pro (any project >1 track with effects) hang at 'Preparing' or show progress stuck at 0% for an extended period; the UI remains responsive but no encoding progress occurs. The activity monitor shows high GPU driver processes and no disk or CPU write activity to the destination. Disconnecting the eGPU and restarting Final Cut Pro allows exports to proceed normally using internal GPU/CPU.

Observations

Observations

Console logs during the hang show repeated messages from the GPU driver and IOAccelFamily indicating 'render pipeline stall' and occasional 'GPU via Thunderbolt: device offline' entries. Activity Monitor shows 'com.apple.render' worker threads pegged but the Compressor/Export processes are idle waiting on GPU completion. Connecting the eGPU changes the 'Metal device' presented to the OS; Final Cut Pro reports the external GPU as the active device in its preferences when connected. Rebooting with the eGPU attached recreates the issue; updating to the latest Final Cut Pro patch provided no change. eGPU firmware updated to vendor latest.

Already attempted

Already attempted

Tried exporting to different destinations (internal SSD, external USB‑C drive); no change. Disabled 'Prefer eGPU' in System Settings — exports succeed (using internal GPU). Reset NVRAM and SMC-equivalent for Intel Macs. Booted in Safe Mode (disables eGPU) — export works. Reinstalled Final Cut Pro. Collected sysdiagnose and exported logs but no obvious user-space crash reported.

Question

Question

What diagnostic steps should I take to determine whether the hang is caused by a driver eGPU offload regression, a Final Cut Pro GPU scheduling issue, or a Thunderbolt dock firmware problem, and what safe mitigations allow continued editing/exporting while retaining eGPU performance for other workloads?

COMMUNITY RESPONSES

Verified community troubleshooting

0 replies

No answers yet..

Showing 0 of 0 community responses
AI DIAGNOSTICS

Incident analysis

Updated 1w ago• Trouble ID

Most likely causes

Metal or GPU driver regression on macOS 14.1 that mismanages offloaded render jobs on external GPUs. 75%
Throttled/unreliable Thunderbolt connection or dock firmware causing the eGPU device to go offline during job initialization. 60%
Final Cut Pro not handling a specific external GPU vendor's feature set, leading to hangs when presented as the primary device. 30%

Summary

The problem reproduces only when the eGPU is active and Final Cut Pro attempts to use the external GPU for encoding/rendering. Console logs indicate stalls in the Metal/IOAccel stack and device offline messages, which points toward a driver or Thunderbolt connectivity issue that leaves FCP waiting on GPU work that never completes. Because disabling eGPU forces FCP to use the integrated GPU and exports succeed, the application pipeline is intact. Suspect causes are eGPU driver/Metal regression, Thunderbolt dock firmware resetting the device under load, or Final Cut Pro not handling device resets gracefully.

Suggested Fixes

Diagnostic step 1
Reproduce and collect: with eGPU connected, reproduce hang and capture a sysdiagnose (use Activity Monitor → Sample Process or command-line sysdiagnose) and Console logs filtered for 'GPU', 'Metal', 'IOAccel' and time ranges of the hang for submitting to Apple or the eGPU vendor.
Diagnostic step 2
Test with the eGPU directly connected to the Mac (bypassing the dock) to rule out dock firmware or cable issues. Use a known-good Thunderbolt cable.
Diagnostic step 3
Try a different eGPU enclosure or GPU if available to determine whether the issue is vendor/model-specific.

Continue troubleshooting

Matched from the title, description and tags