Questions about AMD's New "Mantle" API

Qb2k5

Qb2k5

New member
I watched the stream here: http://www.livestream.com/amdlivest...625b&utm_source=lslibrary&utm_medium=ui-thumb

At 2:38:00 you can see the Frostbyte programmer mention that they are working together with AMD on a new graphics API called "Mantle", which hasn't been done on the graphics front since the mid-late 90s.

Is this its own API? This is not like CUDA where the renderer is DX/OGL and the Physics engine is CUDA? This is like DX, OGL, and Mantle in their own categories?

If this is true I hope it takes off. Id hate to see another fragmentation in the graphics front but there are only three major gpu manufacturers out vs like having 20 from the 90s.

I guess this is AMD's version of glide.:D

And the audio being done on the gpu is AMDs answer to the powerful dedicated audio in the PS4 and Xbox One. I wonder what Nvidia and Creative has in store? Could future CPUs from Intel and AMD have dedicated cores or co-processor for audio?
 
 
AMD didn't win ****. They may have got the PS4 and Xbone but it is yet to be seen if that will help them on the PC side, and designing a proprietary graphics API for open platforms is hardly unification. Devs will still have to program to Direct3D unless they are alright with losing two thirds of there potential market.

I am actually rather surprised at this, given AMD's former lauding of open standards. Funny how things change.
 
Is this truaudio stuff supposed to work in conjunction with Mantle? I just don't see any other way for it to pick up positional cues.
 
spike700 said:
AMD didn't win ****. They may have got the PS4 and Xbone but it is yet to be seen if that will help them on the PC side, and designing a proprietary graphics API for open platforms is hardly unification. Devs will still have to program to Direct3D unless they are alright with losing two thirds of there potential market.

I am actually rather surprised at this, given AMD's former lauding of open standards. Funny how things change.
Click to expand...

How do you know Mantle is proprietary?
 
so this mantle, it's something open source?
could nvidia write their own code to access the api to work with their chipset architecture, or is this hardware locked to amd only?
 
spike700 said:
Getting the PS4 and and Xbone is hardly winning. They still play second fiddle in almost every other market and there presence in some is laughable.
Click to expand...

Multi-billion dollar contract isn't winning? I think we can discount your silicon valley analysis, thanks though.
 
Skynet said:
Wow.
Click to expand...

koolaid-good.png
 
caveman-jim said:
products for tablet, ultraportable, notebook, desktop, server and cloud gaming. What fragmentation? AMD won.
Click to expand...

You know full well that my comments are a response to this statement and in the same context, but whatever AMD won, there is no fragmentation, we all know AMD absolutely dominates all these markets :rolleyes:

Multi-billion dollar contract isn't winning? I think we can discount your silicon valley analysis, thanks though.
Click to expand...

It is just that, a contract. One that was sorely needed by a company that has spent the better part of the past decade hemorrhaging money. While it may be a win for AMD in its own right, it is most certainly not the win you implied.

How do you know Mantle is proprietary?
Click to expand...

Because it only works on AMD hardware? Unless the API is agnostic enough that NV or other companies can write their own mantle driver to interface with other hardware and mantle is indeed open source. I am sure you already know the answer to this, but if it does turn out to be the case it would make your "AMD won" statement kind of superfluous.
 
H001iGAN said:
so this mantle, it's something open source?
could nvidia write their own code to access the api to work with their chipset architecture, or is this hardware locked to amd only?
Click to expand...

Q1 Yes
Q2 Yes that what happened to Tomb Raider with TressFX which saw Radeon 7970 GHz Edition blew GTX 680 out of the water by around 50%. When Tomb Raider gone gold and Nvidia received Tomb Raider gold code to looked into TressFX code then they wrote their own code in Geforce driver and Tomb Raider patch. End resulted with GTX 680 matched 7970 GHz performance. Same thing will go with Battlefield 4 with Mantle if AMD claimed that with Radeon 290X will blow Titan out of water by up to 50% turned true then Nvidia will wait for Battlefield 4 gold code to look into Mantle and write their own code that will see Titan matched Radeon 290X performance.

AMD claimed TressFX and Mantle are revolutionary so that not true, that are AMD marketing gimmicks. AMD TressFX=Nvidia PhysX=AGEIA APEX, AMD Mantle=Nvidia CUDA/Cg=3Dfx Glide.
 
This is a really interesting development. Just thinking about the potential it has to influence the momentum of projects like SteamOS, which could do well with an API capable of replacing the functionality of DirectX, that isn't OpenGL.
 
The future...

The future...

The way I interpret Mantle it will enable reduced effort on the part of developers to port the console versions of their games' rendering engines to the PC. It will negate the need to rewrite many of these low-level optimizations into an alternate path for DX 11.1 that may be expensive to develop and ultimately still inferior. In doing so AMD's strategy may have 2 developer impacts:

1) Make AMD graphics hardware & CPU overhead excel on apps using Mantle and do so inexpensively by leveraging console optimizations.
2) Reduce the return on investment (ROI) of developing a fully optimized DX11.1 render path as the market of users desiring this path shrinks by 30+% and in doing so increase the benefit gap offered by Mantle. This impact doesn't occur with early Frostbite 3 engine games such as BF4 because Mantle is too new but this would be the natural longer term consequence of PC games with support for Mantle.

The consequences of these developments are good for AMD and possibly very bad for Nvidia and Intel. Since it is extremely unlikely either company would quickly adopt Mantle as a matter of hubris (and it may not even be technically feasible at this time) then two possible scenarios could develop in the next 2-3 years:

3) Nvidia/Intel pay money and donate software engineers to game engine developers to increase the resources they spend on their DX11.1 paths. If they cannot stay competitive in absolute performance they may look to fund proprietary image quality enhancements or other differentiators such as PhysX support to redirect interest away from AMD.
4) Nvidia/Intel implement their own proprietary low-level graphics APIs and allocate their money and software engineering resources to enable select game engines to run with their API. The ROI for this strategy is highest for enabling their APIs for different game engines than those that support Mantle. The end result would be yet more graphics APIs and fracturing of the market for PC games along the lines of games that run best on AMD/Nvidia/Intel with less overall consistency across games.

Which of these outcomes happens depends on how great of a differentiator Mantle will be. If it delivers < 10% performance improvement over DX11 on performance/enthusiast systems then it may not put enough competitive pressure on Nvidia & Intel to do anything as drastic as 4). However, if the performance delta exceeds 10%, and especially if it exceeds 20%, these companies will be under a great deal of pressure to go with their own proprietary solutions else they will see their market share erode and Mantle be deployed to ever more key titles.

Some alternate strategies from Nvidia/Intel could emerge:

5) Lobbying Microsoft to standardize a Mantle-like API.
6) Lobbying game developers to switch back to OpenGL with low-level optimization techniques such as bindless resources giving many of the same advantages as Mantle.

But ultimately I think it will go down as follows:

7) Nvidia introduces their own proprietary API and battles AMD for 2+ years, ultimately failing due to Mantle having first-mover entrenched status and natural competitive edge due to console alignment. After this dark phase in the industry the companies come to an agreement to make Mantle an industry standard in the same way that SGI adapted IRIS GL into OpenGL and Apple adapted Grand Central Dispatch into OpenCL.

I'll try to remember to look back on this post in 2016 and see if I predicted any of it right :)
 
Nvidia and Intel have already done everything they can so far to **** AMD over. Nvidia's TWIMTBP breaking features and crippling performance on ATI/AMD GPU's, Intel's Broken compiler crippling performance on Non-Intel CPU's, both of them Paying BILLLIONS to OEM's and Game Developer's to remain Intel or Nvidia only, Nvidia Focus Group members slandering AMD on forums, etc...

At Nvidia's urging, the Tech Report publishes yet ANOTHER article going "Look, look, look! Look at how shitty AMD's Crossfire performance is!!!" the DAY BEFORE this new product line and Mantle feature are announced. AMD even asked Scott to hold off and he refused. What a dick. There's no way that guy isn't taking money from them.

It's about time AMD fights back. I hope Mantle takes off in a huge way. I hope game developers optimize the **** out of games for AMD CPU's and GPU's.

It's about ****ing time.
 
Dr. Zhivago said:
Nvidia and Intel have already done everything they can so far to **** AMD over. Nvidia's TWIMTBP breaking features and crippling performance on ATI/AMD GPU's, Intel's Broken compiler crippling performance on Non-Intel CPU's, both of them Paying BILLLIONS to OEM's and Game Developer's to remain Intel or Nvidia only, Nvidia Focus Group members slandering AMD on forums, etc...

At Nvidia's urging, the Tech Report publishes yet ANOTHER article going "Look, look, look! Look at how shitty AMD's Crossfire performance is!!!" the DAY BEFORE this new product line and Mantle feature are announced. AMD even asked Scott to hold off and he refused. What a dick. There's no way that guy isn't taking money from them.

It's about time AMD fights back. I hope Mantle takes off in a huge way. I hope game developers optimize the **** out of games for AMD CPU's and GPU's.

It's about ****ing time.
Click to expand...

Well said! :yes: I stopped reading over Tech Report because of all of this. :bleh:
 
Suppose BF4 on Radeon absolutely rapes Nvidia in benches. Will certain review sites just refuse to bench the Mantle patched version? There are quite a few sites that are in bed with Nvidia right now, I think we could see this happen.
 
You must log in or register to reply here.
Back
Top