[maximumPC] Nvidia Tom and Rev interview on Contracts, Gsync + more

[Black Octagon]

Mantle is also beneficial on certain systems containing powerful CPUs, e.g.s, those running multiple high-end GPUs in Crossfire.

 

[3DVagabond]

Mantle was AMDs attempt to solve a very personal problem. It has been completely overshadowed and forgotten, especially in this thread.

What, you might ask.

APU performance.

Mantle boost performance most on weak CPUs and can do this very well. This is by design and it does this pretty well. AMD bet their future on APUs. They are really potent graphics bolted onto weaker x86 cores. AMD needed a solution if they wanted to make their APUs shine and thus mantle was born.

Instead of marketing Mantle for what it is (a way to prop up their weak x86 cores), the focus shift to how terrible DX is.

The reality is that DX performance on strong CPUs is not that bad at all. Even mantle offers little improvement here. But on the weak slower cores on AMD APUs, it can really make a difference.

I see people talking out their behinds calling DX a failure and how much better AMD done with mantle. They really dont have a clue what they are talking about. You just cannot compare to two like so. DX is an API that spans across hundreds of different hardware and that was the goal always. All in all, it has done a great job. But if you focus on one architecture alone could you get more performance? I sure would hope so.

So I think the conversation should be on why mantle is really here. AMD set out to solve a very specific problem and I think Mantle does that rather well. I just find it very interesting how well it has been spun and how readily people adopt the misdirection. It is not about how much DX sucks, it was really about squeezing more performance out of their APUs. Once you realize this then you will understand some of the more puzzling aspects. Like why AMD would really want to keep mantle around even after DX12 is out. And you will know the answer when someone asks, "do you think intel will ever support mantle?"
Answer, "No"

Its not gonna happen. I promise

The optimizing that's done for CPU's is done because they are running out of CPU power. We aren't seeing massive gains on Mantle because the games would completely choke DX if all of a sudden they substantially increased the draw calls, even though Mantle could handle it.

 

[ocre]

Never said there is zero mantle does in a high end case. But amd didnt invest the millions in mantle for the smaller gains with a powerful cpu. Amds center focus was to unlock their apus. Just look back at the first unleashing, it is more than evident. Amd not only suggested reviewers showcase mantle with their weaker apus, they sent configurations. These were odd configurations with top end gpus paired with low end apus. And this is also where mantle shines and shines bright.

There is nothing wrong with this, i am not downing the effort or results. I am just saying that people dont seem to realize this was the center focus.

So to 3d,
99% of all games are console ports. Ports that were never limited by draw calls. With this in mind, it seems unlikely things would have ever been different.

I am glad that mantle has come and that the industry is thinking differently. It does open doors for a better future. So the effect is much greater than the original goal.

 

[dacostafilipe]

Amds center focus was to unlock their apus.

I would assume that the main goal was to have people talk about AMD and that's about it ...

And judging by the never ending flow of posts (positive and negative) about Mantle, I think they got it right, no?

 

[3DVagabond]

So to 3d,
99% of all games are console ports. Ports that were never limited by draw calls. With this in mind, it seems unlikely things would have ever been different.

I am glad that mantle has come and that the industry is thinking differently. It does open doors for a better future. So the effect is much greater than the original goal.

One of the comparisons that were given to spotlight Mantle's use was that consoles could support many more draw calls than PC's. so maybe this 99% of games were actually draw call limited on PC which creates some of the "poorly optimized console ports?"

 

[BrightCandle]

I don't think its eight to assume mantle was about low end CPUs. We have been having a problem with CPU performance for years and draw calls and api overhead has been a big thing with dx10 and dx11. Not to mention that gpus are vastly different to what DX was designed for. I think its more a recognition of that changes and the knowledge gained from building the consoles.

I also don't don't think mantle matters. We need a lot more games before anyone cares. Its a next generation card thing more than current gen.

 

[ocre]

Amds development of mantle was all about their apus. The gpus keep getting larger and it becomes an obvious bottleneck. You cannot talk about console niw without talking about apus. This is where and why mantle was born.
That doesnt mean that is the end of it but it absolutely is the beginning

 

[VulgarDisplay]

Mantle was born because the API market is completely stagnant, and AMD did something about it.

There are enough people working at Microsoft to update Direct X with new features for EVERY new GPU series released, but instead we see the same API for 5+ years at a time. Someone needed to put a stop to the Microsoft stagnation stranglehold on PC gaming. Everyone talks about the proprietary nature of Mantle (at this point) being a bad thing but fail to recognize how harmful Microsoft's monopoly on PC gaming is.

Mention OpenGl so I can laugh in your face.

 

[SirPauly]

I think it was created primarily for the potential of APU's moving forward as they may replace discrete revenue! It's a core investment for their future and to compete!

 

[bystander36]

I'd say Mantle was created because they were presented with an opportunity to upstage Nvidia and get free advertising, along with benchmark wins. It also presented a possible solution to CPU bottlenecks in games, which would help their CPU division compete.

 

[ocre]

Just look at what it does for their apus. It has a massive effect, this isnt by accident. Do people really think this? That amd sit around in a room one day and decided to spend millions making an api just because they thought M$ sucked? Then this crazy byproduct was massive performance gains on their apus? Wow

Mantle puts the same picture on the screen as dx. It is just a totally different way to get there. One that is very well suited for their apus. It prevents a lot of back and forthwhich in bandwidth limited and not so powerful cpu cases it has massive returns. This is by design, not accidental. Mantle was never created as an attack on m$. that is crazy.
Amd bet the farm on their apus and obviously this is very very important to them. Mantle helps them shine a lot brighter. To anyone who cant see they are strongly connected, i am in disbelief

 

[BrightCandle]

The big benefit so far with Mantle is with low power CPUs and high power GPUs. No one would buy that today because all the other games would run poorly on it, those that weren't Mantle. However we have been stuck at 9k draw calls for quite some time, GPU performance has continued to increase and that gap of CPU to GPU performance is becoming more apparent. While today the performance is fine and there is little value to a high end CPU and GPU combination (or a low end CPU and low end GPU) in the future low overhead/high draw call will allow games that we can't currently have.

Lower overhead and more multithreading is necessary for the future of gaming, we are hitting the limit already.

 

[frozentundra123456]

Just look at what it does for their apus. It has a massive effect, this isnt by accident. Do people really think this? That amd sit around in a room one day and decided to spend millions making an api just because they thought M$ sucked? Then this crazy byproduct was massive performance gains on their apus? Wow

Mantle puts the same picture on the screen as dx. It is just a totally different way to get there. One that is very well suited for their apus. It prevents a lot of back and forthwhich in bandwidth limited and not so powerful cpu cases it has massive returns. This is by design, not accidental. Mantle was never created as an attack on m$. that is crazy.
Amd bet the farm on their apus and obviously this is very very important to them. Mantle helps them shine a lot brighter. To anyone who cant see they are strongly connected, i am in disbelief

Do you have any benchmarks showing this "massive effect" with APU gaming? According to this very recent article link, mantle has a very minimal effect on gaming on an apu. I am sure if you paired the APU with a powerful discrete card, you would see large improvements, but I dont know why you would buy an APU if you were going to pair it with a powerful discrete card, considering their relatively anemic cpu performance.

 

[BFG10K]

Mantle makes the game development process easier not harder.

Yeah, that explains why BF4 Mantle has been in beta for such a long time and still isn't finished. Because it's soooo easy. :awe:

Yes they can but chose not to, which was a smart call.

Whether they chose to or not is irrelevant to my point. When you use Mantle, you're coding for GCN. This is completely unlike DX12 or any other derivative before it.

Again, if you think coding for a piece of hardware at the metal layer is so great, you must think DOS was the best OS ever made.

 

[BFG10K]

Perhaps its a myth that dx12 started in 2006.

Try looking at the date-stamp at the bottom-left of the slides. AMD didn't even have a graphics division at the time they were made.

And I never said DX12 started then, I said the process of removing overhead did with the introduction of the new driver model, coupled with DX10.

If what you say is true than thank the gaming gods that amd has made a competing api that is better than anything Microsoft has released so far in less than 2 years.

Oh yes, an API that ties you to GCN/AMD with zero backwards compatibility with hardware and 10,000s existing games. Where do I sign up for this "amazing" API?

You are acting like Microsoft's stranglehold on PC gaming features is a good think. AMD had single handedly forced Microsoft to innovate and if they can continually force Microsoft to innovate because they can add features at will it is a great thing.

There's absolutely no evidence of MS being pushed by Mantle. Again, steps were being taken to lower overhead back in 2006. Why do you think Vista completely overhauled the graphics driver stack? If anything, you could say Mantle was pushed by DirectX.

As for the stranglehold, MS is in a neutral position to develop an API given they have no competing hardware division and are just trying to sell Windows. This is in stark contrast with AMD making Mantle, which gives them an inherent bias and conflict of interest with locking out competition.

This is why DirectX (and OpenGL for that matter) will always be fundamentally different to any IHV-produced API.

I still find it hilarious that people don't see how lame the unveiling of dx12 has been.

There's nothing lame about a vendor-neutral API that reduces overhead and allows any IHV to target it without restriction or licensing, while providing full backwards compatibility.

 

[BFG10K]

DX11 multithreading traded better core scaling for increased driver overhead.

Please provide proof of this claim.

Meanwhile even now I can show DirectX 11.2 further reduced overhead and frame latency using real code and live programs: http://channel9.msdn.com/Events/Build/2013/3-062

DX11 MT is almost entirely done within the drivers.

And? The whole point of WDDM is a tighter coupling between DirectX and the driver stack. That's how it was designed.

Again, this is nothing new, and started with Vista eight years ago.

 

[AnandThenMan]

There's nothing lame about a vendor-neutral API that reduces overhead and allows any IHV to target it without restriction or licensing, while providing full backwards compatibility.

You'd might have a valid point if DX12 was out and did what it promises.

There's absolutely no evidence of MS being pushed by Mantle. Again, steps were being taken to lower overhead back in 2006. Why do you think Vista completely overhauled the graphics driver stack? If anything, you could say Mantle was pushed by DirectX.

I find this ridiculous.

 

[ocre]

Do you have any benchmarks showing this "massive effect" with APU gaming? According to this very recent article link, mantle has a very minimal effect on gaming on an apu. I am sure if you paired the APU with a powerful discrete card, you would see large improvements, but I dont know why you would buy an APU if you were going to pair it with a powerful discrete card, considering their relatively anemic cpu performance.

Every one of the original reviews that showcased mantle when it came out.

So the path forward with APUs is what? A 7870 GPU with steamroller cores? hmmm. What does mantle do in that case?

If you read my post on this, from the first one, you will see what i am talking about.
Mantle as a way to solve a problem. The path forward with APU and how Mantle plays into this. AMD can not keep packing on GPU cores without diminishing returns. Mantle reduces overhead and a lot of back and forth between the GPU and CPU. AMD is a CPU company, you really cannot see how these come together? AMD is a CPU company that bought ATI. Why did they buy ATI?

I think the issue here is most people are looking at this from the wrong angle. The discrete GPU. Mantle performance comes from the CPU, by solving congestion, back and forth, overhead, and stream lining. It makes total since that AMD as a CPU company would focus their time and money in this area. Especially considering their APUs keep getting better and better graphics paired with not so powerful cpu cores.

Look at the APU in the PS4 and the performance it is capable of. AMD as a CPU maker would love to have this kind of performance in the PC world. Can you really not see how Mantle is all about CPU? AMD bet the farm on APUs. They are totally vested in this and there is no going back. Mantle is a multi-multimillion dollar effort and you would really have to be snowballed to think they spent this money just to show out DX. They did not make mantle for bragging rights or for a peeing contest.

Most of the cash flow into AMD comes from their CPU division. Mantle is a way to get more performance out of their CPUs. This is important because they are pairing larger and larger graphic cards to weak steamroller cores. Mantle sits right in the middle of AMDs most ambitious project ever. The APU.

The APU was the result of a very costly purchasing of a graphics card maker, ATI. AMD sank billions into the purchase and with the plan always being to merge the CPU and GPU to give them an edge. The bet the farm on this. It was extremely expensive and AMD could hardly afford it. They went into tremendous debt in the purchase and it took them years to finally get their APU to market. So much so that Intel had already merged the GPU and CPU. But AMDs APUs had much more powerful graphics. Their only draw back was the weak CPU cores. One which still exist to this day.

AMD is still 100% behind their APUs. Their success is vital to the company. Can you still not see how Mantle fits in this? Shall i go on?

As a CPU company, as a company that makes APUs like the ones AMD are making, what does Mantle offer? Do you really think this is a byproduct or coincidence?

 

[3DVagabond]

Yeah, that explains why BF4 Mantle has been in beta for such a long time and still isn't finished. Because it's soooo easy. :awe:

Now I know you comprehend what a beta is.

Whether they chose to or not is irrelevant to my point. When you use Mantle, you're coding for GCN. This is completely unlike DX12 or any other derivative before it.

You would have to have seen Mantle source and DX12 source to know how similar or different they are. People in the industry have lead us to believe they are very similar.

Again, if you think coding for a piece of hardware at the metal layer is so great, you must think DOS was the best OS ever made.

This is not direct to metal. There is a driver and abstraction layer.

 

[SirPauly]

You'd might have a valid point if DX12 was out and did what it promises.

I would place more merit with DirectX and Windows than an API from AMD! And there could be a potential backlash if the market feels that AMD is too focused on Mantle and not enough focus on DirectX!

And this classic from AMD :

AMD’s Roy Taylor Started the Obituary of DirectX, but Microsoft Says Not So Fast

http://www.hardwarecanucks.com/news/amd-roy-taylor-directx12/

 

[BFG10K]

Now I know you comprehend what a beta is.

Relevance to the quoted statement: Mantle makes the game development process easier not harder?

You would have to have seen Mantle source and DX12 source to know how similar or different they are. People in the industry have lead us to believe they are very similar.

There's some similarity given AMD copied Microsoft's DirectX HLSL, which again makes the claim that Mantle is forcing Microsoft's hand laughable.

Outside of that, I'm not sure how anyone can claim an API targeting GCN is in the same league as DirectX which was designed to be hardware-agnostic. DirectX works on tablets, consoles, and desktops and has mechanisms to promote scaling for different environments.

This is not direct to metal. There is a driver and abstraction layer.

Try an elementary Google search of "DOS device driver" and VxD.

 

[BFG10K]

You'd might have a valid point if DX12 was out and did what it promises.

Given every version prior achieved that, what evidence do you have that version 12 won't?

 

[MathMan]

One big difference between Mantle and GameWorks is the kind of problem that it tries to solve:
Mantle targets reduction of CPU overhead to increase the efficiency of the GPU. GameWorks is a middleware that encapsulates complex visual effects into a library and move it to the GPU.

The problem for Mantle is that this is a one-time deal: if your GPU runs at 80% because it's hampered by an overly heavy driver model, the best you can do is improve the performance by 20%. Very nice to have, but not earth shocking. What's more is that others can do something similar as a one time thing too: enter DX12. It doesn't even have to improve as much as Mantle, it only has to do most of it to close the gap to an unnoticeable level. Nobody is going to lose sleep over one GPU running at 100% and the other at 98%. This is similar to the problem of Creative Labs: you very quickly get to a situation of 'good enough for most'. Once Intel integrated sound into its chips, they became a footnote in history.

Because of this, Mantle is inherently a short term thing: as soon as there is a similar API that's not controlled by one of the 3 heavy weights (AMD/Nvidia/Intel), it will be game over. Compare it to Nvidia's Cg shader language vs. standard DX shader language. It was a great innovation initially, but ceased to be useful as soon as Microsoft entered the game.

GameWorks is completely different: they don't tackle a deficiency like Mantle, they're trying to expand functionality. There's really no limit to what can be incorporated in it, as long as there's a desire to get every more visual oomph out of a GPU.

It is much harder and more work to do this and, by extension, to compete with it. They are also pretty complex things to implement by a game developers, who are always on a tight schedule and on tight budgets, so it's a very appealing option for them too.

Just go to the AMD website and compare the visual effects library that they provide (hint: it's only TressFX) with the effects that are provided in the GameWorks library. I expect this gap to only get worse. AMD is right to be very concerned about GameWorks: it's a much bigger threat to their business than Mantle is to Nvidia's.

 

[ocre]

I honestly think that Gameworks is Nvidia trying to differentiate PC from console. They are bolt on effects that are really only being applied in cross platform ports. PC gaming is very important to Nvidia (and very much rightly so) and gameworks makes implementing advanced features a whole lot easier from one game to the next. If there was no gameworks, these games would come out pretty much exactly as their console counterparts. But i guess people really think otherwise.

Basically it has been villainized to death, just like pretty much anything Nvidia does. I do not really get all the widespread aggressiveness towards the company but apparently its a smaller group making a lot of noise. Reading forums, review post, blogs, you tube comments, etc it would appear that every one absolutely hates Nvidia and all that they do. Release a tablet, and it is the worse tablet ever-pointless and a waste of everyone's time. They make game works and they are trying to sabotage PC gaming (when actually they are trying for the opposite, to take it up a notch). Gsynsc, Physx, you name it. Terrible terrible nvidia, plaster it all over the net.

The most astounding thing is that as the nvidia hate has been spreading and increasing everywhere on the net, Nvidia themselves are as popular as ever with gamers. As much as there are claims that everything nvidia does is a failure, they somehow continue to make decent profits. Surely something isnt adding up here.

Perhaps it is because the real gamers who buy their cards are out enjoying them playing games. They dont care or feel like they have nothing to prove or defend. I mean, no matter how loud these nvidia nay sayers get, Nvidia keeps trucking along, doing what they do. So the people that make the most noise, well they just want to be heard...... I guess

 

[BrightCandle]

Gameworks is used on the consoles, as well as the PC. Its designed to solve the problem that the GPU manufacturers created cards capable of these effects, spent time writing sample code on how to produce them and then the games developers didn't do anything with it. We all remember the awesome looking demos AMD and Nvidia have been pumping out for the last 15 years or so and yet a lot of it never made it into games. Gameworks is solving that by providing all those routines in a production ready package that you can drop the DLL into your game and write a call to the appropriate feature, tweak the settings a little and its working and looking great from the beginning. For bonus points it works on AMD, Nvidia and Intel hardware because its calling into DirectX using the standards set there by the 3 vendors. It has the potential to be bad for everyone but Nvidia, but so far they haven't appeared to cheat using it, there is no smoking gun showing Nvidia hurting their competitor using this API. If it happens they will ruin what they are building, its not in their best interests for game devs to not use gameworks, they get paid for it.