People do need to realise, reduced CPU overhead is FAR from Mantle's only improvement, and DX has a huge long history of promised improvements that make little or no difference, DX11/.1/.2 have these. Bindless tables (iirc, bindless something or other anyway) was supposed to help hugely yet offers no real world performance improvement, etc, etc. DX/MS has tried to improve these things many times over and simply not delivered. Mantle is already delivering and effectively, MS has tried to make such changes for the past 4-5 years, with DX10 being their "let's build it from the ground up with none of the DX9 old style downsides"................ opps.
You have to factor in that one of the key problems dev's have with DX is that it's a blackbox, they send data in to it, they can't tell what's happening, they can't predict the outcome. It's a case or write some code that should work, send it into this black box of DX, pray it comes out the other side working. When it doesn't there is little to no feedback in why it's not working. Mantle from the ground up was designed to be far more open(almost completely), to integrate design and debugging tools that give dev's direct feedback to precisely where a problem occurred, rather than weeks of e-mailing back and forth with driver teams at both Nvidia and AMD trying to track down numerous unknown issues, they can get to the bottom of the issue themselves very quickly.
There is a reason they guys working with Mantle have said it's quite and easy and has taken few guys little time to port it to Mantle, because a huge amount of extra work is gone.
As for DX, I said when Mantle was announced, the very worst outcome would be MS would actually put some effort into DX, at best it would simply be great itself.
Mantle offers the low level access openGL will likely never do, the non platform specific tie in that DX will only ever offer, and proven end result. As yet, even if we presume MS can deliver what they've promised, what they haven't promised is key. They haven't promised to open DX to all platforms, while possible I can't see MS doing it AND they'll likely tie it to the next Windows version. They haven't promised or as yet hinted at improving the tools or drastically opening up DX to enable it to be more transparent to the devs.
So a dev can focus on Mantle, use the tools to help debug the game as they build it, requires less work fixing problems because they don't have to wait on Nvidia, AMD and MS to help with almost every problem, and have it work(at some stage) on every platform out there. Or they could focus on DX12, limited most likely to Windows 9, and most likely dx12 compatible cards as I can't see them improving code and updating the code in DX11 for instance to be more efficient. They'd still have to do other work again for SteamOS/Linux.
I'd be far happier if openGL was the one promising the world as at least it would work on any platform, but their history of delivering is questionable.
One of the issues is, if DX did become great for now, due to being kicked up the arse. As soon as Mantle say or openGL got knocked back down, then MS would feel no need to stay competitive. Again MS have a vested financial interest in consoles, game sales makes them hugely more cash on consoles. They've let DX stagnate with poor work for donkeys years. The only reason they want to improve it is because of competition, as soon as it's gone they'll stop improving it again.
guys making games, or gpu's, should be in charge of the API, not a company whose focus is console gaming sales, OS sales, and OS lock ins.
Performance drives sales, Nvidia/AMD sell gpu's, their focus has always been performance and always will be, it's why AMD has made a API and pushed performance forward. MS should not be in charge of the industry standard API as their agenda does not tie in directly with gamers. GPU makers agenda DOES tie in with gamers, as do game dev's. Someone, or some group, made of gpu makers/game dev's should be in charge of the industry standard, almost anyone but MS.
If MS win the "API war", then it will benefit everyone for the short term, and long term get us back to where we were right before Mantle, utter stagnation, and gpu's/api being held back for years.
