One thing that has probably slowed down DX12 development is the fact it will support multiple vendors. While I'm not going to argue that supporting 4 vendors instead of Mantle's 1 will mean 4 times as much work I can imagine that if it is "close to the metal" then each vendor product will need quite a lot of work done on it. If not then Mantle would be close to supporting more vendors too (admittedly 'we' probably don't understand how close it is to this). Also AMD had all of AMD to do the AMD specific part of Mantle. I doubt the same amount of resource has been given to Microsoft to help with AMD's DX12 integration. I doubt Nvidia would have dedicated any more resource either. So MS has to cover 4 vendors with each one probably having less vendor specific help than Mantle had. So I think it's quite possible that despite DX12 lagging behind Mantle release date wise, it doesn't mean that it was started that long afterwards, if at all. Who knows, maybe it was MS talking to AMD about it and needing their input that gave AMD the idea for Mantle and AMD knew that since their API would only support 1 vendor that they could do it quicker... (I'm not saying that's how I think it happened, but there are probably a very small number of people that actually know).
Yes DX12 isn't actually out yet, but neither is Mantle. Mantle doesn't even go into beta until April apparently. It does have more game support at the minute though. Bit early to declare a 'winner' yet though I think. Doesn't even need to be a 'winner'.
I guess the next question is which API will support (and I mean actually support not just have the ability to if someone decided to do it) the big 3 (AMD, Nvidia and Intel) first? Assuming DX12 will by 'Holidays' 2015 (when is that? What 'seasons' do the Americans have in a year?), how much before or after will Mantle add Intel and Nvidia support?
I think that may influence developers, in time. Mantle might be more like what they've asked for, in some ways, but unless they asked for an API that only supports 1 vendor, DX12 may be more like what they want in others.
It's hard to say, maybe developers are happy to use multiple APIs in order to cover all vendors (I suspect most developers, or the companies that pay them, will want to be able to sell to as many customers as possible, not just GCN AMD owners). If they are then I can't see if being a problem for us, it'll likely just be another graphics option for us.
As for Mantle and DX12 being similar, well they are both graphics APIs trying to do the same basic job. Makes sense that there will be some similarities. Maybe there are some naming similarities because they're taking 'inspiration' from the same or similar sources? (DX11, OpenGL, hardware or driver names)
I suspect if you got a group of structural/mechanical (or whatever) engineers and asked them individually to design a method for getting a steady stream of cars over a river that more than one of them, without copying, may come up with the idea of a bridge. And there may well be some elements of that bridge that will be the same or similar. And that could be because they work in the same field and learnt similar stuff.
EDIT: Also, Humbug, unless you run at stock normally, why not do those tests with your normal OC?
