When AMD (AMD +1.28%) first announced its Mantle graphics API, the promise of radically improved performance over Microsoft's (MSFT 0.55%) DirectX -- the de facto standard in the PC gaming industry -- seemed almost too good to be true. With the first games to support Mantle now available, the API provides large performance gains for certain configurations, but more modest gains otherwise. While Mantle is impressive, it's likely no game changer for AMD.
Source: AMD
The biggest problem with Mantle is that game developers need to add support for the API, a task which adds both cost and time to the development process. Given that Mantle is only supported on AMD GPUs -- not those from rival NVIDIA (NVDA 0.24%) -- the return on investment for game developers is questionable at best.
Microsoft isn't taking Mantle lying down, with the dominance of DirectX critical to the success of the Windows platform. At the upcoming Game Developers Conference, Microsoft is set to host two presentations that promise to unveil plans to make future versions of DirectX more efficient by reducing overhead, the same feat that AMD has accomplished with Mantle. In addition, a presentation about OpenGL, the other major graphics API, promises to show how the existing implementation can reduce overhead by a factor of 10 or more. With DirectX already widely used by game developers, and OpenGL apparently able to match Mantle's performance gains today, Mantle doesn't seem to have much of a chance of gaining wide acceptance.
A two-horse race
Both DirectX and OpenGL have been around for a long time, and both are very mature graphics APIs. Almost all triple-A PC games use DirectX, and the Xbox One uses a modified version of DirectX that allows for more direct hardware access. OpenGL, on the other hand, is cross-platform, and the embedded version has become the standard on mobile devices. Some high-profile PC games use OpenGL, like the phenomenally successful Minecraft, but DirectX has largely remained the standard in the world of PC gaming.
Where does Mantle fit in? If the performance gains are indeed short-lived, then it doesn't fit in at all. DirectX will take some time to catch up, but OpenGL, at least according to the GDC presentation, is already capable of matching Mantle's performance improvements. And with OpenGL supported by all major graphics hardware vendors and functional on all major platforms, supporting Mantle doesn't make much sense.
What this means for AMD and NVIDIA
Mantle was supposed to provide AMD with a big advantage over NVIDIA's hardware, and while the performance gains are real, they will likely be short-lived. AMD certainly deserves credit for pushing the graphics industry forward, lighting a fire under Microsoft to improve DirectX, but Mantle is extremely unlikely to supplant either DirectX or OpenGL.
This leaves AMD in the same position that it was in before Mantle, spreading itself too thin across multiple businesses and falling behind in all of them. NVIDIA's recently announced GPUs based on the new Maxwell architecture provide massive gains in power efficiency compared to AMD's products, and at the high end of the market, NVIDIA's products offer the best performance per dollar, according to Tom's Hardware.
While Mantle has certainly compelled Microsoft to improve DirectX, I doubt the API will be around five years from now. OpenGL already provides a cross-platform alternative to DirectX while being supported by all graphics vendors, not just AMD, and any performance advantage that Mantle has will be short-lived. If the goal of Mantle was to push the graphics industry forward, then AMD succeeded. But Mantle doesn't give AMD an advantage against NVIDIA, and the latter will likely keep its market-leading position.
