Next DirectX may pack Mantle-like low-level features

Scorpus

Posts: 2,162   +239
Staff member

There's been a lot of talk recently surrounding Mantle, AMD's low-level API that gives developers better access to graphics card hardware, with less CPU overhead and greater coding efficiency. Tests show that Mantle is quite effective at improving performance on entry-level and mid-range systems, especially where CPU power is lacking, compared to DirectX.

But the guys at Microsoft don't want to be left behind when it comes to DirectX. Looking at sessions for the Game Developer's Conference (GDC), which is scheduled for mid March, reveals Microsoft's own Anuj Gosalia will be detailing what's in store for the next DirectX. Here's an excerpt from the session's description.

For nearly 20 years, DirectX has been the platform used by game developers to create the fastest, most visually impressive games on the planet. However, you asked us to do more. You asked us to bring you even closer to the metal and to do so on an unparalleled assortment of hardware. You also asked us for better tools so that you can squeeze every last drop of performance out of your PC, tablet, phone and console.

Bringing DirectX "even closer to the metal" sounds a lot like what AMD has been trying to achieve with Mantle, although the updated DirectX API likely won't be restricted to just one vendor's hardware. Another session on Direct3D, hosted by Development Lead for Windows Graphics Max McMcullen, also discusses low-level improvements.

Come learn how future changes to Direct3D will enable next generation games to run faster than ever before! In this session we will discuss future improvements in Direct3D that will allow developers an unprecedented level of hardware control and reduced CPU rendering overhead across a broad ecosystem of hardware.

OpenGL won't be left out either, with a session at GDC titled "Approaching Zero Driver Overhead in OpenGL". Hosted by presenters from AMD, Intel and Nvidia, the session will detail "high-level concepts available in today's OpenGL implementations that radically reduce driver overhead--by up to 10x or more."

It's possible some of the components of OpenGL that reduce driver overhead are already available, but we'll have to wait until GDC to learn more about what's in store for the two most popular graphics APIs.

Permalink to story.

 
Awesome. I was hoping that both DirectX and OpenGL would take this approach natively without having to rely on third-party middleware.
 
If apple owned mantle Microsoft would be getting its directx shut down and games made with it shut down, and a lawsuit in it
 
Just gave me a thought.

Would future DX versions (or this DX12, DX 11.2, whatever they be calling it) be supported in the Xbox One? Because if performance is as good as they say it is, that could make for some very unhappy Sony fanboys (although to be fair, all fanboys are very sad)
 
I rarely say this but, thank you AMD for pushing things forward in this sector! Considering AMD have a reputation of Driver issues, they were the last company I would expect to push something like Mantle across but they did it, they proved it helps and now the competition has to catch up, I like this a lot and would probably explain why Nvidia have been quiet, because they knew this would happen.
 
Even if they say it will be better. it's still a long time away from launch. we're lucky if we get half of what's written there in 2015.
 
If Microsoft was owned by Apple it would sue AMD for copyright infringement and for not paying Microsoft patent money for direct x

Your posts are always so comical.

Thanks! Even tho I dont write them in a comic thanks for telling me that I should put them in a comic :)
 
Last edited:
Even if they say it will be better. it's still a long time away from launch. we're lucky if we get half of what's written there in 2015.
Microsoft and D3D have always moved with glacial pace, so no surprise there.
The OpenGL extensions are further advanced, and will probably be easier to integrate. (BTW: Anyone wanting a sneak preview of the extensions should go >>here<<)

The other half of the equation, is that Mantle is having a painful birthing process as well. If every game with Mantle support continues to launch without Mantle then AMD are burning a first-to-market advantage such as it is.
Since AMD seem intent on keeping Nvidia and Intel away from Mantle for most of this year (at least), I'm pretty sure that even at a leisurely pace, D3D and OpenGL will overhaul Mantle sooner than later - given that Nvidia, Intel, and AMD are involved.
Assuming that the "closer to metal" DX and OGL get both the TWIMTBP and Gaming Evolved thumbs up, I can't see either Nvidia or Intel wanting Mantle at all. Given that AMD's own GCN 1.0, VLIW4, and VLIW5 architectures aren't even Mantle compatible five months after AMD unveiled the API, it remains more a PR/marketing feature (especially given the scarcity/price of GCN 2.0/1.1 cards in some markets) for the most part.

Kudos to AMD for re-kickstarting the low overhead gaming API - just a pity that they didn't go hell for leather in software/beta test hires to ensure a less protracted and problematic introduction.
Very interested to see how this works out. I hope DX12 also brings Ray Tracing :D.
MS might incrementally add options (along with taking a few away by making DX12 a Windows 9 exclusive), but real-time ray tracing isn't coming to gaming any time soon. Even a less compute intensive SVOGI-type iteration would need to be judiciously applied to allow it to be accessible to consumer graphics cards.
 
Last edited:
AMD doens't have the money, time or the resources to maintain acceptable performance and improvements to Mantle over time, and it was only a matter of time before MS would fire back, and they have.

Waiting for Mantle on a per title basis is not going to work to gain mainstream acceptance. Thief just released, and reviewers are already tearing it a new one, and a LOT of people read and trust those reviews. Add the fact that the Mantle implementation on that titile is late, and now you have early adopters not experiencing it, and by the time it is available, players may have already beat the game and moved on to other titles.

Mantle is good in theory, but even AMD says it benefits low end CPU's with mid to higher end GPU's, and that just isn't good enough.
 
Is it possible AMD simply wanted to give MS a push in this direction. That is AMD knew Mantle would not survive the winter, so to speak. But yet wanted the features available across the board, without doing the hard work themselves. The more I think about this the more I think AMD pushed a button and ultimately got what they were aiming for. If AMD can get MS to do this, it will make their APU's look even better. And in the end they can drop Mantle support, at MS cost of R&D with DirectX. If I were to guess, I'd say AMD has been conniving little B__ards this whole time. I'll give AMD a (y) for their success, either way.
 
Is it possible AMD simply wanted to give MS a push in this direction. That is AMD knew Mantle would not survive the winter, so to speak. But yet wanted the features available across the board, without doing the hard work themselves.
Unlikely IMO. AMD could have got much the same response by announcing Mantle, and deluging the press under a mountain of PowerPoint slides (an AMD forté). Add some feelgood partnership buzz from EA DICE, and it would have accomplished much the same result without expending any serious effort...or at least less than what has been expended.
The more I think about this the more I think AMD pushed a button and ultimately got what they were aiming for.
The cause and effect might be the other way round. Remember Carmack talking up Nvidia's OpenGL extensions (see my previous post for link) just after Mantle launched? Those weren't achieved overnight. In fact Nvidia seemed to have been hard at work aiming for a closer to metal approach since the Kepler architecture began achieving fruition.
On the D3D front things are murkier. Might be a "me to" reaction, and also might be a "why is our hardware vendor trying to undermine us" reaction. Might also be a reaction to the extensions already being worked into OpenGL by both AMD and Nvidia
And in the end they can drop Mantle support
Not anytime soon. Ditching a new feature short term usually brings about a collective :SMH: reaction from the industry...and AMD can ill afford that. EA DICE also have a vested interest.
 
although the updated DirectX API likely won't be restricted to just one vendor's hardware

Mantle is an API just like DX, it is not vendor specific, AMD already stated that "Mantle" is NOT proprietary.

Mantle does not talk directly to the hardware as the Term "low-level" is being over stated.

"It's not low-level in the sense that we are exposed to individual architectural decisions," Baker said. "For example, Mantle still abstracts the details of the shader cores themselves, so that we don't even know if we are running on a vector machine or a scalar machine. What isn't abstracted is the basic way a GPU operates. The GPU is another processor, just like any other, that reads and writes memory. One thing that has happened is that GPUs are now pretty general in terms of functionality. They can read memory anywhere. They can write memory anywhere."

http://www.tomshardware.com/news/oxide-mantle-gpu-directx-star-swarm,26029.html
 
Mantle is an API just like DX, it is not vendor specific, AMD already stated that "Mantle" is NOT proprietary.
Mantle does not talk directly to the hardware as the Term "low-level" is being over stated.
What you mean is that Mantle is hardware agnostic - which is likely true.
Now, if the tech isn't proprietary, then it is open source. If it is open source then developers other than EA DICE and AMD should have access to the code. Do they? Of course not.
AMD's own project manager is already on record as dismissing this, and made it abundantly clear that Mantle is AMD tech and any use of it by a third party will require a licence.
PCGH asked Raja Koduri about Mantle and its "openess". He replied that AMD doesn't see Mantle as an open standard like OpenCL or OpenGL. He also tried to brush aside comparisons with Glide but then stated: If a competitor were to approach AMD to make their own backend and drivers for Mantle, AMD would not dismiss them right away.
AMD's own Mantle FAQ also has a conspicuous avoidance of the phrase "open source", just a repeated mention of a nebulous "later in 2014" timeline for an SDK for anyone wanting to buy in.
 
Back