Mantle and DX11 the same in BFH

Mantle could allow the game to push more than in DX11, but it also costs more to implement such things.

It simply makes most sense for the studios to use Mantle (and DX12, Vulkan) to target the lower-end, rather than upping the bar for the higher-end PCs

 

I have another unsurprising, yet rarely discussed finding regarding Frotbite's Mantle implementations.

Although the AMD document introducing Mantle briefly discusses 'Asymetric CrossFire' and 'Other Asymetric Multi-apater' configurations with 1:2, 1:3, "and beyond" performance ratios (http://www.amd.com/Documents/Mantle-Programming-Guide-and-API-Reference.pdf, Page 109, PDF-page 111), my quick benchmark shows that it isn't really flexible.

I have a 290X as my main VGA but I happened to have another 290 (not X) at hand. I reset the 290X to it's default 1000/1250 clocks and I down-clocked the 290 to 550/1250 (10% less SP means I need ~1100MHz to simulate a reference 290X with the reference non-X and thus 550MHz to simulate a half-290X, discounting the memory bandwidth which shouldn't be a notable bottleneck anyway).

The result was what I expected but not what I hoped for. I ended up with the fps numbers produced by a single 290X (with ~100% GPU utilization on the down-clock 290 and ~50% utilization on the 290X with the stock clocks). So, CrossFire with Mantle failed to figure out and use a 2:1 GPU1:GPU2 ratio.

Is this scenario supposed to be any better or the same with DX12?

 

Mantle isnt updated anymore for bf series.
dx11 is the get go sadly.