Here is the D3D12nBodyGravity sample that AMD forked from Microsoft to add concurrent execution of a compute queue along to the default/graphics queue.
As you can see, there is no compute queue at all running the sample on the R9 280 (a GCN1 GPU, second screenshot) while with the R9 380 (a GCN3 GPU, first screen-shot) everything runs fine. You can also notice issues on the flip queue, which result in application stuttering.
This issue is present from some months, with the first WDDM 2.1 driver IIRC.
Please note that this is NOT related to the "quick response queue" at all. This is related to the ability to running in parallel compute and default queues (that so famous part of the multi-engine D3D12 feature known as "asynchronous compute" or "async. compute" or whatever people like to call it now).
Please also note that I post this "tech" stuff here because there isn't a Direct3D developer forum at all in the entire AMD Community portal.
Suprised there isn't official AMD reply yet. It's quite big topic in some forums already. Atleast some kind of reasoning would be nice to hear.
I hope it's just the problem, and fix it. I do not want to lose this function on their video cards 280x and hd7970
Edit (December 14th, 2016):
Please note that I’m not an official AMD spokesperson on this subject, and I should have made clear when posting, that this is only my personal understanding of the situation, which might be wrong or at least contain some invalid assumptions. It appears the official driver team has already issued a statement that they are aware of the issue and currently looking into it.