AMD GPU14 Tech Event Sept 25 - AMD Hawiian Islands

[DarkKnightDude]

If Mantle really does what AMD advertises, would really help out the low end computers, or at least somewhat anyways.

 

[Rikard]

Nope, seems completely fine to me. Mantle is exactly what PC gamers have been crying out for - better graphics and performance than consoles (and less reliance on CPU). This will lead to much cheaper systems being viable gaming machines.

Oh, if you're an Nvidia fan? You got two options, either switch to red or get in the DX11 slow lane. Your choice, I'm benefiting from Mantle and you can as well. This is the crux of the issue - Nvidia never had an overwhelming product that made people switch from ATI. CUDA or PhysX, we didn't care.

Do you care enough about Mantle to switch? If so, AMD has won. Only the most lost of Nvidia fans will stay loyal to green in the face of overwhelming fps superiority.

Lets not get too carried away, ok? IF Mantle really works as well as we hope, AMD will have an advantage this generation. It is not like people without GCN will not be able to play in DX, they will just not get as good performance and perhaps image quality. It is is a bit like you need certain cards for DX11, but DX9 still works. Then NV will either make their own API or get onboard the Mantle train some time in the future.

 

[BD2003]

Nope, seems completely fine to me. Mantle is exactly what PC gamers have been crying out for - better graphics and performance than consoles (and less reliance on CPU). This will lead to much cheaper systems being viable gaming machines.

Oh, if you're an Nvidia fan? You got two options, either switch to red or get in the DX11 slow lane. Your choice, I'm benefiting from Mantle and you can as well. This is the crux of the issue - Nvidia never had an overwhelming product that made people switch from ATI. CUDA or PhysX, we didn't care.

Do you care enough about Mantle to switch? If so, AMD has won. Only the most lost of Nvidia fans will stay loyal to green in the face of overwhelming fps superiority.

Or you know what, maybe Nvidia could start selling cards for $100 cheaper than their AMD counterparts instead of $200 more? Imagine that!

And what do you think happens to AMD going forward with that kind of FPS dominance? You think you're still going to see the same year over year improvement, or maybe they'll just bump the clocks a few MHz, give it a new name, and pretend like its a major improvement over last year?

 

[PrincessFrosty]

This talk about GCN is interesting.

Obviously as a gamer I like the idea of removing abstraction and coding more directly to hardware to optimise performance in games.

But I think I'm correct in saying that APIs such as DirectX aren't deliberately inefficient, their lack of efficiency comes as a trade off where multiple architectures can be supported, the layers of abstraction cause a drop in performance but an increase in compatibility.

So Mantle is creating a trade off by increasing performance and decreasing compatibility. But what does this mean for future AMD graphics hardware, does it mean to continue to support Mantle they're locked in to the GCN architecture?

What repercussions could that have for AMD, and gamers (both AMD and Nvidia users) in the future? I'm very aware that there MUST be some kind of trade off here, it's not just that DirectX is crap and Mantle is great, it's that they trade off different benefits with each other and obviously AMD want to peddle the benefits and aren't going to discuss any potential drawbacks.

I'm interested to see how people think this will play out long term, by long term I mean 5+ years, when GPU architecture changes significantly.

One possible market response I can forsee is Microsoft being rather displeased with Mantle (being OS compatible) and wanting to drag it down, wouldn't the best move be to evolve DirectX in such a way that future DirectX compliance requires low level hardware architecture decisions which directly conflict with GCN, pressuring AMD to either change architecutres and drop GCN compatibility.

I see it going one of two ways, either it will fail to be widely adopted outside of AMD sponsored titles, in the same way that PhysX failed as a GPU accelerated physics engine. OR it might get widespread adoption and really harm DirectX but then set us up for a long term fail being tied to GCN. Normally I would say it's the next PhysX, but factoring in native console support and potentially steam machine support...as well as developers constant battle to reduce their development to 1 build fits all (max audience, min effort) I think there's a distinct danger this could go either way.

 

[GaiaHunter]

nVidia supports OpenCL and DirectCompute. Cuda is there that they can leverage their feature set much faster. For graphics they use OpenGL in the same way with their own extensions.
Cuda would be same if we compare Cuda to x86 and saying that nVidia is much faster with GPGPU than AMD GPUs while using x86...
Mantle is here that developers dont use DX/OpenGL anymore to give AMD an advantage. If their hardware would be so much better they would not need an own API.
And Cuda is only usefull for the professional market. People want to solve problems. Cuda has no feature in the consumer market in which developers want to sell their software to as many people as possible.

And AMD supports DX, Mantle is just there so developers can leverage their feature set much faster.

Notice how it states that it allows full access to graphics hardware capabilities.

Then market will decide what is best, like they do with CUDA, DirectCompute and OpenCL

And how ist Mantle resolving this problem? :hmm:

I could swear I saw a dude from Dice saying how they worked with AMD for this and how it allows them to use the hardware in a much more interesting way.

 

[sontin]

And AMD supports DX, Mantle is just there so developers can leverage their feature set much faster.

We have OpenGL for the second part.

Notice how it states that it allows full access to graphics hardware capabilities.

Lol, yes. Like Glide. Back to the 90s.

I could swear I saw a dude from Dice saying how they worked with AMD for this and how it allows them to use the hardware in a much more interesting way.

Yes, a Dude who got paid for saying he loves PC gaming while introducing a proprietary graphics API. D:

 

[SirPauly]

I have no problem with a company that tries to offer more and go beyond standards for their customers! Curious to see how much more efficiency AMD can muster with Mantle and their Radeon family!

Ultimately, the market decides and nice to see AMD invest and trying to improve their customers gaming experiences -- hopefully, more awareness to build their brands!

 

[GaiaHunter]

One possible market response I can forsee is Microsoft being rather displeased with Mantle (being OS compatible) and wanting to drag it down, wouldn't the best move be to evolve DirectX in such a way that future DirectX compliance requires low level hardware architecture decisions which directly conflict with GCN, pressuring AMD to either change architecutres and drop GCN compatibility.

I see it going one of two ways, either it will fail to be widely adopted outside of AMD sponsored titles, in the same way that PhysX failed as a GPU accelerated physics engine. OR it might get widespread adoption and really harm DirectX but then set us up for a long term fail being tied to GCN. Normally I would say it's the next PhysX, but factoring in native console support and potentially steam machine support...as well as developers constant battle to reduce their development to 1 build fits all (max audience, min effort) I think there's a distinct danger this could go either way.

It can also be used by developers that build and then license their engines to others to make that proposition more interesting.
They could create wrappers for AMD, Intel and NVIDIA, allowing each low level API to work with their engine, making it more attractive for the customers of their engines, that would otherwise have to either code for a high level APi like openGL/Direct X or spend resources coding for each individual render path.

 

[PrincessFrosty]

Nope, seems completely fine to me. Mantle is exactly what PC gamers have been crying out for - better graphics and performance than consoles (and less reliance on CPU). This will lead to much cheaper systems being viable gaming machines.

Do you honestly believe that in the long run this will lead to "much cheaper systems"? If AMD dominate and take the market, and Nivida are forced to pull out or cease to make a competing product, what do you think will happen to AMDs prices?

What about willingness to invest in R&D and push to have faster and faster hardware, is that going to get better or worse?

All gamers who have been on the scene since the early days of 3D gaming, know that a single dominant player in the market with a monopoly is just about the worst of all possible long term outcome, for absolutely everyone.

That's one of the reasons DirectX was widely adopted, because it eliminated hardware favoritism and allowed a free market of hardware devices to compete fairly, competition creates low prices and high value.

 

[GaiaHunter]

Do you honestly believe that in the long run this will lead to "much cheaper systems"? If AMD dominate and take the market, and Nivida are forced to pull out or cease to make a competing product, what do you think will happen to AMDs prices?

What about willingness to invest in R&D and push to have faster and faster hardware, is that going to get better or worse?

All gamers who have been on the scene since the early days of 3D gaming, know that a single dominant player in the market with a monopoly is just about the worst of all possible long term outcome, for absolutely everyone.

That's one of the reasons DirectX was widely adopted, because it eliminated hardware favoritism and allowed a free market of hardware devices to compete fairly, competition creates low prices and high value.

Is Intel going somewhere that I don't I know about?
DX also created basically a duopoly concerning discrete PC GPUs out of a pool of 6+ makers.

If any company gets a monopoly that is why we pay taxes to keep States and their armies. (although it is possible politicians will lack guts or prefer to be paid off instead of breaking such a monopoly)

 

[Revolution 11]

This is way too good to be true. AMD seems to be outmaneuvering Nvidia for once. First consoles, and now this? This can't be happening.

There has to have been a serious shift over in More-Cores Land for this to be possible. :hmm:

I think those announcements last year and this about important people being hired by AMD like Raja Koduri is finally paying off. Old AMD management was not this aggressive.

 

[RussianSensation]

My main issue with this is it most likely will cause NV to bring out their own version of mantle. Once that happens then it will really pollute the gaming environment. Some games will support mantle, some will support nv-mantle. Now what?

I agree with you. Problem is the gaming market ruled and didn't buy AMD cards. Time to change strategy or go bankrupt. Adapt or die.

A common API (DX, even though developers want it to be better) is better for the gamer (who isn't a video card company groupie) and isn't that what WE want?

Yes, I would want that too but NV and AMD are now locked in a battle of exclusive features as differentiators. In the past, MS's DX gave them both a level playing field. I guess AMD is thinking outside the box on this one and said what the heck, we can make a more efficient API than DX, so let's do it. By very nature, I don't see how you can have a lower-level API on a metal level for both AMD and NV since lower-level API of this nature is GPU architecture specific. I don't think it's realistic. At best you can create a more efficient API on a driver level for AMD/NV that beats DX11 but AMD decided to go even further down to open their hardware level access to developers.

Whether this fails or not, it's an interesting development in the gaming industry for sure. The most surprising part of this all is that AMD did it first despite far lagging NV in financial resources and historically weak developer/gaming relations.

If you read my comments when AMD entered the GE program, I already feared the market is going to get even more segregated. Of course NV can simply out muscle GCN with Maxwell and then what? Imagine if AMD could run games at 20-30% faster with Mantle, but Maxwell 20nm is 50% faster than GCN 20nm without Mantle. Mantle would keep AMD competitive and without it, NV would raise prices like they did this round. The main thing I want is for AMD and NV to be competitive and if Mantle is a way for AMD to keep up with 550mm2 die chips, so be it because I am not paying $1000 for flagship single GPUs. Think about it, it took AMD 5-6 months to catch up to 780 and 780 is a nearly 4 year old Kepler architecture. If anything, AMD should be shaking once 20nm Maxwell drops because their GCN will be 3 years old competing against a brand new arch tech from NV. They need every advantage they can get or they'll get run over by a 550mm2 20nm Maxwell.

In an alternative case, if NV's cards are 20-30% slower, NV fans will still buy them anyway. We know already - 5870 vs. 285, HD7970 vs. 580. If AMD is 20-30% slower than NV, AMD users will buy NV cards.

 

[wand3r3r]

The funny thing is that we have yet to see a single concrete number that means anything. So draw calls are 9x faster, what does that translate to for the end user? Before everyone gets to disturbed you might want to relax and see what it actually means.

 

[SiliconWars]

Lets not get too carried away, ok? IF Mantle really works as well as we hope, AMD will have an advantage this generation. It is not like people without GCN will not be able to play in DX, they will just not get as good performance and perhaps image quality. It is is a bit like you need certain cards for DX11, but DX9 still works. Then NV will either make their own API or get onboard the Mantle train some time in the future.

Even if Nvidia makes their own API they're going to have to pay the devs an awful lot of money to code for their GPU's. This is the genius part of AMD's plan here - if the API is basically the same as in the XBone and PS4 all it will take is a few bucks "incentivizing" for console ports and they'll have a huge advantage.

And what do you think happens to AMD going forward with that kind of FPS dominance? You think you're still going to see the same year over year improvement, or maybe they'll just bump the clocks a few MHz, give it a new name, and pretend like its a major improvement over last year?

No because there are more reasons than Nvidia to continue to progress. Mobile for one, and Intel isn't standing still either.

 

[skipsneeky2]

Just like cude/phsyx. I don't know if fighting fire with fire is the best option, but it's funny to see the nvidians squirm and switch goalposts anyway.

Despite the fact i own a nvidia card,i'm not bothered by all the talk about this Mantle thing period.The 7970Ghz and the 290x/280x will have the lead over my gtx770 and that is fine but i don't feel like my card turned into a pos overnight.

Like the slides about Mantle,frostbite 3 and how somehow it will utilize 8 cores.The end of quad core gaming as far as mainstream performance goes perhaps?

 

[sontin]

Even if Nvidia makes their own API they're going to have to pay the devs an awful lot of money to code for their GPU's. This is the genius part of AMD's plan here - if the API is basically the same as in the XBone and PS4 all it will take is a few bucks "incentivizing" for console ports and they'll have a huge advantage.

Lol, the UE4 is optimize for Kepler. nVidia only need to put their own "Mantle" API in it and AMD is screwed. No difference. Nobody is developing multi port games with an low level API in mind. They using Windows PCs. Right now only DICE is using it. Other companies have nVidia machines (Ubisoft) or using them for PC only games (Witcher guys).

 

[BD2003]

Do you honestly believe that in the long run this will lead to "much cheaper systems"? If AMD dominate and take the market, and Nivida are forced to pull out or cease to make a competing product, what do you think will happen to AMDs prices?

What about willingness to invest in R&D and push to have faster and faster hardware, is that going to get better or worse?

All gamers who have been on the scene since the early days of 3D gaming, know that a single dominant player in the market with a monopoly is just about the worst of all possible long term outcome, for absolutely everyone.

That's one of the reasons DirectX was widely adopted, because it eliminated hardware favoritism and allowed a free market of hardware devices to compete fairly, competition creates low prices and high value.

Exactly. If you're some sort of graphics card fanboy you get to celebrate a temporary victory, and then your worst nightmare comes true. We've seen this happen SO MANY TIMES in so many industries. I don't care how much performance it can unlock, there's literally no realistic scenario where this results in any tangible benefit to anyone but AMD shareholders.

If AMD kills Nvidia, they'll stop innovating and their quality will slip, just like what happens to EVERY COMPANY IN EVERY INDUSTRY that's ever been in that position - gamers lose.

If nvidia creates a competing proprietary API, gaming will become super fragmented, to the point where you might need to buy two separate graphics cards to play everything with full features - gamers lose. If mantle fails, it will have resulted in a years long distraction where AMD spent a ton of time and resources working on it, rather than spending that time and money on making the products better in an open competitive market.

 

[Final8ty]

Lol, the UE4 is optimize for Kepler. nVidia only need to put their own "Mantle" API in it and AMD is screwed. No difference. Nobody is developing multi port games with an low level API in mind. They using Windows PCs. Right now only DICE is using it. Other companies have nVidia machines (Ubisoft) or using them for PC only games (Witcher guys).

The difference is that AMD is in the consoles as well.

And optimizing for one vender in sponsored titles on the PC is nothing new and NV has been doing just that far longer than AMD.

 

[SiliconWars]

Lol, the UE4 is optimize for Kepler.

And strangely there is a lack of noise about this...

nVidia only need to put their own "Mantle" API in it and AMD is screwed. No difference.

If Nvidia is still around in the 2+ years it will take them to develop it, you mean.

Nobody is developing multi port games with an low level API in mind. They using Windows PCs. Right now only DICE is using it. Other companies have nVidia machines (Ubisoft) or using them for PC only games (Witcher guys).

So why the fear? Is it because you know that the devs have been asking for it? Surely you must realise that Crytek are bound to follow next...then it could just be like a stack of dominos.

In the end this is enabling better graphics and better performance. Those that move with it will prosper and those who don't, won't.

 

[ShintaiDK]

If AMD kills Nvidia, they'll stop innovating and their quality will slip, just like what happens to EVERY COMPANY IN EVERY INDUSTRY that's ever been in that position - gamers lose.

Thats incorrect. It would kill themselves as well.

 

[RussianSensation]

The end of quad core gaming as far as mainstream performance goes perhaps?

:thumbsup: I hope so. I can't wait. If games start using 6-8 threads effectively, a lot of us will jump to $500 level 6-core & 8-core Intel/AMD CPUs for example. As Intel/AMD fatten up their margins through higher ASPs, they will devote more resources to desktop CPUs than they do now where the focus is largely mobile and performance/watt.

If there are games that will run 35-40% faster on 6-8 core CPUs over quads, many of us will actually consider dropping $500 on a CPU vs. 4770K.

 

[RussianSensation]

Exactly. If you're some sort of graphics card fanboy you get to celebrate a temporary victory, and then your worst nightmare comes true. We've seen this happen SO MANY TIMES in so many industries. I don't care how much performance it can unlock, there's literally no realistic scenario where this results in any tangible benefit to anyone but AMD shareholders.

Not true. We can't predict the future, but it could end up that the ONLY way AMD stayed in the game with Maxwell and Volta is through a 20-30% advantage via Mantle. We don't know but you can't rule out a possibility of NV releasing a 550mm2 Maxwell/Volta that beats AMD's best 20nm by 30-35% again. And like I said, history already shows that people will buy NV cards if they are slower, but history shows that when AMD's cards are slower, AMD's brand equity, profitability and market share tank. Not saying it's going to happen but it rebuts your theory that there is absolutely no case to be made why Mantle could be good for us gamers.

 

[5150Joker]

And strangely there is a lack of noise about this...



If Nvidia is still around in the 2+ years it will take them to develop it, you mean.



So why the fear? Is it because you know that the devs have been asking for it? Surely you must realise that Crytek are bound to follow next...then it could just be like a stack of dominos.

In the end this is enabling better graphics and better performance. Those that move with it will prosper and those who don't, won't.

Wow talk about delusional fantasies.

 

[flopper]

:thumbsup: I hope so. I can't wait. If games start using 6-8 threads effective, a lot of us will jump to $500 level 6-core & 8-core Intel/AMD CPUs for example. As Intel/AMD fatten up their margins through higher ASPs, they will devote more resources to desktop CPUs than they do now where the focus is largely mobile and performance/watt.

If there are games that will run 35-40% faster on 6-8 core CPUs over quads, many of us will actually consider dropping $500 on a CPU vs. 4770K.

if a 8core amd cpu would run as a game as BF4 in the better or similiar to a 4core ht Intel one, expect a huge amd boost in sales.

 

[Rikard]

Whether this fails or not, it's an interesting development in the gaming industry for sure. The most surprising part of this all is that AMD did it first despite far lagging NV in financial resources and historically weak developer/gaming relations.

This is what amazed me how smart AMD have been, despite their aweful presentation techniques and tradionally lackluster public relationships. They might have had Mantle in mind for years, and when the new consoles come they invest an undisclosed fortune to make sure they own 100% of that market. That, together with a pretty substantial market share of discrete graphics cards, allows AMD to push through Mantle as a new defacto standard to replace DirectX. It could never have been feasible if they did not own the consoles. It will be very interesting to see how this developes. I think Microsoft has a lot more to fear than Nvidia.