Developers Using AMD Mantle API

One of the presumed advantages of owning an AMD HD79xx or R9 2xx care is Mantle. A new AMD developed graphics API that allows way better graphics on PCs.

Three more developers are on the "bandwagon". One says:



Mantle is only for PCs. The new AMD powered consoles will not use Mantle directly. Rather, the console OS already provides the advantages of Mantle.

Mantle Being Adopted

Not bad, since Mantle was recently released. Maybe it actually will become the "next big thing" in PC graphics over the next couple years. AMD developed tessellation had to wait years for MS and Nvidia to jump on board. Maybe not so for Mantle. Another reason to consider recent AMD GPUs and APUs when buying new stuff.

I never knew AMD developed tessellation, nor did I know about this Mantle business. It sounds more exciting than the usual die shrink; I definitely like the "work smarter instead of harder" philosophy .

I hope to see good results out of this, and more developments in parallel computing.

Tessaltion was championed by Matrox in 2002.

http://www.pcmag.com/article2/0,2817,1155954,00.asp

ATI (now AMD) had their "tessellation" card out in 2001 -- R8500. It was the first ATI card I ever bought (had owned Nvidia --- and 3dFX before that). TruForm (tessellation) really was a huge improvement at the time -- like night and day. However, Nvidia and Microsoft would not support it.



TruForm

Only one game I found and bought used tessellation. Although ATI tessellation worked flawlessly only if "designed for", any game could use it just by setting a "switch" in the drivers (the customer could turn it on or off). That actually worked. And though some items became distorted (when not designed for tessellation), it was only a few in the game I played (Serious Sam) that used the "switch". So, to me at the time, it did not make much sense that games did not at least provide the "switch" for their customers.

According to the articles published over the period of years following, ATI/AMD continued improving tessellation. In the years following R8500, some of their cards had some of the basic technology inside, but it was not supported by anything and not advertised by ATI/AMD.

In the long haul, games just upped the mesh complexity to give nearly the same result as tessellation -- however, it was still nowhere near as good as early tessellation to my eyes. Now, Microsoft and Nvidia support it -- and, we see it is better even now -- but, it was better all along -- for the customer.

As posted above, it seems Mantle has a chance to have a better fate.

ATi Had Tessellation First.


MS is Still committed to blocking Mantle on Xbox One. (Supposedly doesnt "Support it"), Wants everyone to continue to Use DirectX 11.2 API

I wonder how much money AMD can afford to toss around to developers to get the support for Mantle going and maybe to keep it going?

Developers are going to have to include great DirectX support anyhow by default just like they always have, otherwise they will miss out on all the customer base that is not AMD (latest Steam H/W Survey shows top 13 cards are non-AMD):
http://store.steampowered.com/hwsurvey/videocard/

So any Mantle specific development by game publishers will be extra development and QA costs above DirectX that AMD will need to fund somehow since they won't do it just for coolness sake, at least initially unless and until the developers/publishers see a return on Mantle investment.

Call me a skeptic, but with XBOne not supporting Mantle and SteamOS hardware consoles going Nvidia direction, ultimately Mantle has a big uphill fight that will need to be funded by AMD paying development costs directly until they can show enough customer return benefits for publishers spending dev and QA time for it.

Reduced costs for publishers was a key reason DirectX became the defacto standard and we can see even by Nvidia's own attempt to make Physx be some hardware specific benefit, that the publisher adoption rate on that has been minimal unless they pay the developer specifically for it, and even then the games rarely benefit from it because otherwise the developer catches backlash from the customers not supported.

My guess is Mantle might be like Physx, a nice marketing bulletpoint on the video card packaging that comes at direct cost to the company with little actual difference for the gaming industry.

Just my $.02 as a neutral skeptic who buys both Nvidia/AMD but doesn't drink the marketing coolaid from either side.

Mantle allows developers Direct Access to the GPU, thus removing the Overhead from DirectX. it should be like 10-20% Performance increase depending on hardware and DirectX Version.

Alot of developers are limited by what is in the DirectX SDK, so to give them the option to program directly to the GPU Functions, removes that Limitation.

So, Reduced hardware Overhead, and no longer Limited by DirectX Features.....
Programmers will have a field day, Games that use Mantle will prolly have better looking effects and stuff in a few years.

PhysX is a Joke, any GPU/CPU has the ability to calculate physics ....
The Origin of PhysX was to have a Separate PPU to remove strain from CPUs, which were Single/Dual Core at the Time.

Now that CPU's are up to 16 Cores,
I Think Dedicating a Core to Physics Processing removes the requirement for a dedicated PPU,

Ageia Died off when Dual/Quad Core(HT) CPUs Came out, nVidia bought them and moved PhysX to CUDA.

PhysX is an Engine Plugin for companies Stupid enough to not program Physics into their Game Engine.
PhysX is Mainly used for: Draping Fabrics, Explosion Debris, Character Animation (Jointed Geometry, etc), Fog/Smoke that react to objects. (Basically Animation and Interaction).

All of which can be easily done without PhysX. Using Bone Animations and Correctly Programmed Collision/Reaction Models.

Everything you write is rational. I'm surprised to read that even a small number of developers have taken on Mantle.

Along those lines, the competition (Intel/Nvidia) got where they are partly (sometimes largely) by paying developers to use their stuff in preference to their competition (a different aspect of industry history). I really don't think AMD can outbid Intel or Nvidia (except in special cases -- exceptions to every rule). Moreover, Intel and Nvidia have good reputations with the average person.

So, mostly, AMD technology has to provide obvious "bang per buck" to get used.

I don't know if the current crop of Mantle users use it because they really want to -- or because they were paid. I do think that AMD will provide them with help in implementing Mantle (as Nvidia does for their products). I won't be surprised if Mantle powered PC games are still a year or two away from release.

Regarding PhysX: I too have read that it is very inefficient (at least in the recent past -- maybe its better now).

A couple or so years ago, I read that it was noticed that the PhysX CPU only driver (provided by Nvidia) was mildly crippled -- to make PhysX faster on the Nvidia GPUs of the day -- without crippling, it was faster on a good CPU.

Yeah don't get me wrong, I think Mantle could be very cool and definitely would push the DirectX gaming graphics bar that Microsoft (conveniently) seems to have allowed to remain stale on the PC for years now so as to not compete with Xbox (I won't even mention how the "nextgen" consoles can't seem to do 1080p/60fps very easily like our 2 year old cards can on PC).

The main thing is AMD will have to really push/fund much of the Mantle development for it to get large traction, otherwise I don't think we will see many publishers or devs financially jump into it on their own, above and beyond the DirectX development/QA they already have to pay for.

In software development there are always cool new technologies that many devs want to explore, but often in the short term they end up having to stick to the current tried and true baseline that the development/QA budget will support unless there is some specific business return on doing it or it can be included at little additional dev time/cost/risk.

Hopefully AMD is ready to fund and push this for the next 12-24 months and not just rely on their normal (some might say limited to specific titles) AMD developer support.

So far I think the best news is the adoption by Star Citizen, with the billions of crowdfunding and hype that title has and lack of publisher budget oversight, the pursuit of cool stuff like Mantle is definitely likely in that title so it could prove to be a great example for it.

Today an article speculates in one or two sentences that AMD Mantle does NOT exclude Nvidia. Nvidia cards could benefit from Mantle. How that works is unclear to the writer.

Meantime, more games will use Mantle -- over 20 now, it is rumored.