With DirectX 12 around the corner, you may be wondering what the purpose of AMD's Mantle is in the GPU API space. AMD hasn't been wondering about that, however, as the company has some very clear intentions for what it wants to do with its higher-performing, lower-level API.
When AMD's Mantle API hit, we were all curious about what kind of benefits it would bring. As it turned out, it did indeed notably improve performance in various games, however, it only worked on select AMD GPUs. It also required developers to actually code their games for the API, which presented additional work for them, increasing development costs. Despite that, a decent handful of developers did implement it, which lead to something very important: attention from the public, GPU vendors, and developers.
The result of this attention is that Microsoft and the Khronos group started to question whether it may be time for them to do something similar, and as you might have guessed, they did. Microsoft's DirectX 12 is expected to be very feature rich, and so is the Khronos group's upcoming Vulkan API, which comes to replace OpenGL. It may also be called GLNext.
With those two new APIs around the corner, AMD's Mantle appears to no longer serve any purpose because of its limited adoption compared to its competitors. AMD doesn't seem to mind this, however; the company has been working very closely with Microsoft and the Khronos group on the new APIs.
AMD announced that it will be opening up the Mantle 1.0 API, and is publishing a 450-page programming guide. Despite opening up the API, however, AMD is not encouraging developers to actually start developing for it. Instead, the company said that if you're looking to start developing for Mantle 1.0 -- don't; develop for DirectX 12 or GLNext instead. What AMD will do is continue to support partners in future Mantle-related projects, among which is Battlefield Hardline.
So, does this mean that AMD's Mantle is gone forever? No, AMD intends to hold on to it for the purpose of further developing the API, where it will work on new concepts to further improve it and bring new features to the table. At this time, however, we have no idea what these features will be.
As a refresher, what Mantle aimed to do was deliver a lower-level graphics API, which would give developers the possibility of accessing the graphics hardware more directly -- essentially, it removes some of the high-level coding abstraction, reducing overhead and therefore increasing performance.
On March 5, AMD will be holding a press conference at GDC where the company will announce the details of what it intends to do with regard to opening up the Mantle API. And hopefully, AMD will also reveal what it intends to do in the future with its API.
Follow Niels Broekhuijsen @NBroekhuijsen. Follow us @tomshardware, on Facebook and on Google+.
Thank you for your hard work and pushing the pendulum AMD!
My chime in was that, even if Mantle is rendering "obsolete" by DirectX12, it has served its purpose. AMD will keep Mantle around for a bit, just to make sure that DirectX12 doesn't stagnate like DirectX11 did.
Even if DirectX12 "wins" - AMD still wins, because software is catering more to their (current) hardware. Also, now someone else is doing the heavy work. Mantle will probably integrate any features that DirectX12 develops, and vice-versa. Tech Darwinism at its greatest.
Except directx12 is microsoft windows 10 only. So, if you want only one, then it would be vulkan, which will work on every platform.
AMD has had a hand in this, they deserve some good PR for committing to this, specially in their "underprivileged" financial position, compared to Intel and nVidia. They gave the industry a very good nudge and aren't going to sit on it now that it's done it's job.
AMD is coming out a much more open and consumer friendly brand than nVidia lately. I hope they manage to grab some foundries for the next gen, with the promises of VR optimization and Oculus partnership, new memory systems...
Still, i am bit sad if AMD indeed shuts mantle down. It would have been perfect api for legacy windows platforms and for those who say opengl/glnext or whatever is better option... well how many opengl only games we have? Hopefully this will change because of steam machine, but based on past trend of opengl i wouldn't count on it.