Fantastic!

Well this is nice and I wish DX12 was that much worse than Vulkan, but in reality either that benchmark or the drivers are totally broken. In ideal circumstances I'd expect both to produce a roughly equivalent result. Don't take me wrong but there's simply no technical reason why DX12 would have that much more overhead. The two APIs are simply way too similar and share most of their basic design.

To prove my point, you can easily find more balanced results on the web.