HaveAGitGat / Tdarr

Tdarr - Distributed transcode automation using FFmpeg/HandBrake + Audio/Video library analytics + video health checking (Windows, macOS, Linux & Docker)
2.72k stars 89 forks source link

Add Intel Arc hardware acceleration support. #982

Open Finn2708 opened 1 month ago

Finn2708 commented 1 month ago

This PR addresses #902.

It updates Dockerfile.base to jammy and adds the latest media runtime (and probably some unrequired dependencies). Unfortunately, I can't test Dockerfile.final, but in Dockerfile.base I can transcode with HandBreakCLI using my Arc A380.

Let me know if I can assist with testing the final build.

hollanbm commented 1 month ago

This PR addresses #902.

It updates Dockerfile.base to jammy and adds the latest media runtime (and probably some unrequired dependencies). Unfortunately, I can't test Dockerfile.final, but in Dockerfile.base I can transcode with HandBreakCLI using my Arc A380.

Let me know if I can assist with testing the final build.

While I'm generally a fan of all of the changes you made to the Dockerfile, the giant RUN commands make it harder to fully leverage the Docker layers cache.

I would instead open a second PR, with your refactors. It's hard to tell what is required for ARC, versus what isn't. Which could hold up the PR longer than is needed

HaveAGitGat commented 2 weeks ago

@Finn2708 thanks I'll try incorporating these changes.

@hollanbm as far as I'm aware it's better to use fewer RUN commands as each one creates a new layer which can dramatically increase image size depending on what's happening. So yes in development it's faster to use multiple RUN command so the layer cache can be used but not ideal for end image.

HaveAGitGat commented 2 weeks ago

But yes adding the RUN commands and changing the indenting does make it difficult to see exactly what has changed, I'll need to go through line by line.