I don’t mind helping fund things I want. I would’ve used the other platforms if they met my needs. I wouldn’t mind giving some more money if it delivered the features I wanted. Money is a great motivator.
I need to replace an ancient Synology NAS, and I want an HP Gen10 MicroServer… the AMD GPU is slow, but the integrated Radeon 7 would pick up all the slack if HW transcoding worked! I’m disappointed we’re halfway into 2020 and this still isn’t enabled for AMD yet.
(The MicroServer Gen10 Plus are Intel, but even the models with iGPUs have firmware limits that prevent QuickSync or HW transcoding from working. And they only have one PCIe slot, I don’t want to waste on a GPU. So that’s out.)
It does seem kinda crazy that Plex is allowing themselves to miss the AMD boat here. It’s full-steam and well out of the harbor already, and seems like it should be quite low LOE given that it’s well supported in the underlying tools/libraries, assuming the characterization of the original reason is accurate. (And I don’t mean this as an “AMD > Intel” argument. I have no loyalty. I’m just saying it’s undeniably too strong in the market to ignore.)
Any word here? I just built a brand new Synology replacement with a Ryzen Pro 4750G APU and was surprised to learn that Plex does not support hardware transcoding with AMD APUs…
Zen2 APUs are in the wild with impressive performance… I’m surprised that there is still no movement on this. AMD GPU support has been getting a lot of love in the Linux Kernel releases as well.
With the upcoming release of RDNA2 and AMD upstreaming tons of radeon driver code to the kernel, this should be addressed sooner rather than later. The support for ffmpeg is perfectly fine these days and unless plex mangled the ffmpeg libraries and binaries they use beyond being able to merge from upstream, there is no reason why this support cannot be enabled.
Using ffmpeg on CLI, hw transcoding works perfectly on my Ryzen 3 3200G. Hope this can be implemented quickly, certainly 4k transcoding is a bit much for this CPU to handle, while it’s built-in the GPU handles it just fine…