Repository navigation
Support Intel GPU hardware acceleration (VA-API / QSV) #619
upmcplanetracker
started this conversation in
Ideas
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
When trying to use
FFMPEG_ARGSfor Intel hardware acceleration (e.g.,-hwaccel vaapior-hwaccel qsv), FFmpeg fails to initialize the VA-API/QSV device and throws the following error:Even when
/dev/dridevice nodes are successfully passed through to the container, the user-space libraries required to interact with Intel iGPU hardware (iHD_drv_video.so,libvpl,libmfx) are missing from the container's root filesystem.This is very easy to fix because the base
Dockerfilealready installs system dependencies via APT, so unlocking full Intel hardware acceleration support only requires adding the standard Intel driver packages to the existing build stage.The inner FFmpeg binary is already compiled with
--enable-libvpland--enable-libdrm, so no custom FFmpeg compilation is required.Adding the following user-space driver packages to the
apt-get installblock in theDockerfileenables out-of-the-box Intel VA-API and QSV hardware decoding/encoding:(Note:
non-free/non-free-firmwareAPT components should be enabled insources.listto pullintel-media-va-driver-non-freefor full encoding/decoding shader support).This small addition to the build step allows the existing
FFMPEG_ARGSenvironment variable to work natively for Intel iGPU users without requiring custom downstream image builds.Then all someone would need to do to allow ffmpeg to use the intel GPU for conversion is to enable access to /dev/dri.
thanks for considering.
All reactions