Skip to content

v0.3.90

Choose a tag to compare

View all tags
@github-actions github-actions released this 30 Apr 13:33
· 392 commits to main since this release
v0.3.90
2919151

Fixed

  • Render-node auto-detection on multi-GPU hosts. Pre-v0.3.90 the QSV / VAAPI ffmpeg commands hardcoded /dev/dri/renderD128; on a host with both an Intel iGPU and a discrete NVIDIA card (NUC9, etc.) the iGPU is often renderD129 and the hardcoded D128 would have libva try to load the iHD driver against the NVIDIA render node — guaranteed failure. encoder_caps now reads /sys/class/drm/<node>/device/uevent to identify each render node's kernel driver and picks the right one per encoder (i915 for QSV, i915 / amdgpu / radeon for VAAPI; never NVIDIA). Falls back to /dev/dri/renderD128 only if detection fails entirely (preserves single-GPU-host behaviour).
  • /api/stats/encoder-caps response now includes qsv_render_node and vaapi_render_node fields so the UI / debugging can show which /dev/dri/renderD* will be used for each encoder.

Docker images

Pinned to this release:

docker pull ghcr.io/i-ial9000/shrinkerr:0.3.90         # portable (amd64 + arm64, CPU)
docker pull ghcr.io/i-ial9000/shrinkerr:0.3.90-nvenc   # x86_64 + NVIDIA GPU (NVENC)
docker pull ghcr.io/i-ial9000/shrinkerr:0.3.90-edge    # as above but ffmpeg master
docker pull ghcr.io/i-ial9000/shrinkerr:0.3.90-edge-nvenc

Floating tags (:latest, :edge, :nvenc, :edge-nvenc) track the most recent release of each lineage.

Assets 2
Loading