I would agree with this, I use Nvidia cards for professional work on Linux and I’ve never had a problem. Yeah there’s some upfront work configuring the drivers, but I’ve never had it take more than an hour to setup.

I’m not really convinced that a GPU backend is needed. Was there ever a comparison of the different CLIP model variants? Or a graph optimized / quantized ONNX version?

I think the proposed solution makes a lot of sense for the task at hand if it were integrated on the pic-rs end, but it would be worth investigating further improvements if it were on the lemmy server end.