When Magma Freezes Over…

Intel has approached AMD about Mantle. Whyyyyyy?

Intel confirms that they have approached AMD about access to their Mantle API. The discussion, despite being clearly labeled as "an experiment" by an Intel spokesperson, was initiated by them — not AMD. According to AMD's Gaming Scientist, Richard Huddy, via PCWorld, AMD's response was, "Give us a month or two" and "we'll go into the 1.0 phase sometime this year" which only has about five months left in it. When the API reaches 1.0, anyone who wants to participate (including hardware vendors) will be granted access.

AMD inside Intel Inside???

I do wonder why Intel would care, though. Intel has the fastest per-thread processors, and their GPUs are not known to be workhorses that are held back by API call bottlenecks, either. Of course, that is not to say that I cannot see any reason, however…

Read on to see why, I think, Intel might be interested and what this means for the industry.

The First Driving Driver

The first reason, that I can think of, is that Mantle is expected to be significantly easier to write drivers for, compared to DirectX 11 and OpenGL. The API is described as not holding the game developer's hand, especially if it means dragging them, kicking and screaming, somewhere they don't want to go. I think I overextended my analogy a little, but basically it pushes responsibility on the developers to manage things like multi-gpu configurations and the specifics of memory access.

If Intel pushes all of this complexity over to the game developer, it will have less tweaking to do in order to reach equivalent performance. This is good for Intel, because this is not an area that I would consider them as an industry leader. There are two key areas where Mantle does not give a free lunch: hardware design and writing efficient compilers (to convert "shader" code into efficient machine language for their specific GPU).

I consider these (hardware design and compiler engineering) to be strengths of Intel. Of course.

And yes, I know that all of the major GPU vendors have acquired great compiler engineers. Some even go to the extent of manually replacing whole, compiled shaders with their own, equivalent ones to get that extra bit of speed. The point is that, I believe, Intel would much prefer rely on those strengths. I do not seem them wanting to make a robust, stable driver, no matter what a developer might throw at it, which is still efficient.

Beyond that, almost every man-hour spent on Intel's strengths (again, hardware and compilers) wil carry over to DirectX. They each use the same shading language and, according to my interview with Guennadi Riguer, the chief architect of Mantle at AMD, both drivers (Mantle and DirectX) use one, shared compiler. Of course, a fast and efficient, feature-complete GPU core will obviously help, regardless of the API.

Intel Atom-powered tablet, from Ryan's review.

A Second Reason and How It Affects the Industry

But then we get to the second, less obvious reason: the mobile market. If Intel gets in early, Mantle could help performance and battery life on cellphones and tablets, by balancing work between CPU cores, and let the processor spend more time asleep (saving power). Mantle's current, leading competitor is DirectX 12 (ignoring Metal because Apple makes their own processors and will do what they want). OpenGL does not, at least yet, have a suitable competitor, except for the highly-skilled developers who know its ins and outs, including vendor-specific extensions. Clearly, putting your hope in DirectX 12 is betting on Microsoft's mobile market share.

Mantle doesn't seem too crazy, now, at least in comparison.

On the same topic, it might, also, be easier on developers to create applications, especially games, because the API is much closer to DirectX than OpenGL (again, including the shading language which defines every material in any given scene). While it seems a little odd to port between desktop and mobile, at least the same shader library could be used, potentially unmodified if the mobile GPU is fast enough.

If Intel straight-up adopts Mantle, especially if they offer some compelling iGPUs alongside it, that puts a few more eyes on NVIDIA. Currently, it would be safe to assume that NVIDIA has little-to-no plans of supporting any API that is controlled and guided by a direct competitor. The company exerts as much control as they possibly can, in terms of how they deliver an experience to their users. Basically, they (habitually) want to be the ones to make their customers happy. On the other hand, if NVIDIA does adopt Mantle, that would leave us with a seemingly well-designed API, with significantly less driver tuning, with compatibility for DirectX-style materials, and with support for Windows, Linux, and other platforms… that is owned by AMD.

I should make another bowl of popcorn…