If this is your first visit, be sure to
check out the FAQ by clicking the
link above. You may have to register
before you can post: click the register link above to proceed. To start viewing messages,
select the forum that you want to visit from the selection below.
Its all politics thats why radeon is in this shape and AMD announced Mantle right now. That said this is alliance of necessity AMD would like to isolate NVIDIA and MS loves to EEE once more this time with AMD cooperation.
Yeah, we already read it, that's why we are 'whining' ... Both AMD's live presentation and the subsequent analysis on various sides, like anandtech you pointed to, make absolutely no mention of anything other than the Windows Driver.
Once again : AMD + 2 x abject monopolistic corporations that piss on everything smelling a 'standard' that won't let them vendor lock-in => 'standard' API with one reference implementation as a proprietary, closed source windows driver. Given AMD's TRACK RECORD with fglrx quality and features on Linux vs Windows, as of now, the bets are of this being a 'Red CUDA' equivalent from AMD.
I might be wrong, and I sincerely hope I am, and we'll find out in November that Mantle will work out-of-the-box cross-platform on Windows/XBONE, Mac, Linux/FreeBSD-PS4, but, again : the TRACK RECORD of AMD doesn't lie.
Did you not read the article? http://www.anandtech.com/show/7371/u...cs-api-for-gcn
It's nowhere close to gallium3D, mesa, openGL, DirectX, or anything like that. It's directly interfacing with the hardware; completely bypassing an abstraction layer which bogs down the graphics rendering process, in a highly threaded manner. It also provides direct memory access to the VRAM for the developer. In other words, it cuts out the middleman and gets straight to business in a finely tuned low-level environment similar to console development.
But we already have open standards for that, they are called ... gasp ... OpenGL and OpenCL.
Mantle API will be open, but that doesn't mean that Mantle is an open standard. I don't think that an open standard would accommodate AMD's goal here, giving developers the ability to program close to the hardware.