Skip to main content
Seraph!
Participant
August 2, 2026

Premiere Pro 26.3 freezes/white-screens on repeated timeline playback thread wait chain traced

  • August 2, 2026
  • 0 replies
  • 2 views

Premiere Pro 25.6 freezes/white-screens on repeated playback of the same timeline section — all threads blocked on single internal thread

 

Premiere Pro repeatedly freezes (Windows "Not Responding" state, playback audio/video continues briefly in background) when rewatching the same section of a timeline more than 2-3 times. This happens across different projects, including a brand new project with fresh footage — not tied to one sequence or clip.

System specs:

  • Premiere Pro version: 26.3
  • CPU: Intel Core Ultra 9 285K
  • GPU: RTX 4070 Ti Super, Nvidia Studio Driver (latest)
  • RAM: 32GB (31GB installed), Premiere allocated 25GB
  • OS: Windows 11
  • Storage: Boot NVMe (Gen 4), ~50% free, media cache on same drive
  • Renderer: Mercury Playback Engine GPU Acceleration (CUDA)

What I've already ruled out:

  • GPU/renderer settings (confirmed CUDA active, GPU usage stays near-idle even during freeze)
  • H264/HEVC hardware decode routing (tested Intel+Nvidia both enabled, then Nvidia-only  freeze duration shortened slightly but issue persists)
  • RAM (never maxes out, ~20/25GB during heavy use)
  • Disk (Gen 4 SSD, 50% free, normal activity during freeze, tried excluding Premiere from Windows Defender real-time scanning)
  • GPU driver stability (checked Event Viewer for TDR/display driver timeout events — none found)
  • Windows Magnifier (was briefly toggled on, turned off, no change)
  • Reproduced on a brand-new project with fresh footage, so it's not project-specific corruption

Diagnostic finding: Using Task Manager's "Analyze Wait Chain" on Premiere Pro's process during a freeze, every blocked thread (checked 4 separate threads) was waiting on the exact same single internal thread (Thread ID 33404 in my session). CPU usage actually drops during the freeze rather than spiking, indicating Premiere is blocked waiting on this internal thread rather than doing heavy compute — this looks like an internal lock/thread contention issue rather than a hardware bottleneck.