The AMD Mantle Thread

[krumme]

That doesn't mean that they have nothing in common with Mantle.

If you have two different low level API's for the same architecture, there has to be some kind of common code. Also, both the PS4 and XBONE API's were developed with the help from AMD.

Here is the statement from AMD that confirms what I say:

http://community.amd.com/community/...0/17/the-four-core-principles-of-amd-s-mantle

The consoles is not developed with help from amd. Its amd who made them using ip that was even invented and produced before. Same hardware with slight differences.

 

[Noctifer616]

Xbox One uses DirectX and will be a direct copy to PC with DirectX. Mantle however, more work from more or less scratch. The PS4 uses OpenGL, again, not Mantle or close to Mantle either. Its not long ago some people claimed Mantle was the low level API in consoles. And we all know how that vent.

XBONE is using DirectX 11.x, not DirectX 11.2. It's not the same API as we have on the PC, it's specifically designed for the XBONE.

The PS4 has 2 API's, a high level one (a modified OpenGL one) and a low level one (that one should be similar to Mantle to some degree).

If AMD says that the console code is similar to Mantle, than that is a fact unless someone proves us wrong.

 

[Yarn]

I've been asking where people get high end graphics market share from till I am blue in the face. No one seems to know where they pull these random percentages from.

 

[Yarn]

Xbox One uses DirectX and will be a direct copy to PC with DirectX. Mantle however, more work from more or less scratch. The PS4 uses OpenGL, again, not Mantle or close to Mantle either. Its not long ago some people claimed Mantle was the low level API in consoles. And we all know how that vent.

How do you know working with mantle will require starting from or near scratch. And because some people were wrong about it being the exact same api, which really didnt ever seem likely, that means it jumps to the other end of the spectrum and is now the ugly step sister who has nothing in common with dx.?

 

[SiliconWars]

I've been asking where people get high end graphics market share from till I am blue in the face. No one seems to know where they pull these random percentages from.

"Discrete" graphics share is about 60-40 for Nvidia, however it includes cards from the 640 GT and below, which are no better than AMD's APU's (which aren't included).

Without the real bottom end stuff It'll be about 50-50, with AMD selling more cheaper cards like the 7770/7850's's etc and Nvidia more expensive cards like the 770's/780's. That's generally how it's been for a long time.

The big swing is AMD now has the consoles on top of it, for an easy majority of market share. By the end of this year it'll be about 2/3rd's AMD.

 

[Yarn]

A source would be nice but more specifically, I'm encouraging people to use better proxies for the AAA games market then straight gpus sold.

 

[SiliconWars]

http://jonpeddie.com/press-releases/details/add-in-board-market-down-in-q2-amd-gains-market-share/

 

[Erenhardt]

http://jonpeddie.com/press-releases/details/add-in-board-market-down-in-q2-amd-gains-market-share/

Is it taking reference boards only? There are custom board graphics cards, that sure are quite a big part of all AIBs. Or do I understand "AIB" wrong?

 

[Minkoff]

Hi guys, I’m following this tread from the beginning and I just wanted to share something.
The comment is not technical and is a little bit long but the “Mantle hole” might be a little bit deeper. Just scroll to the end if you’re not interested in the opinion of a just registered user. The bellow is just my experience an idea and pure speculations tbh.

The constantly quoted “AMD does not have the GPU market share” thing…is irrelevant now and will be irrelevant in the, not so long, long term as well. Mantle is not about current market share but influencing future decisions when developing/purchasing a GPU and CPU. Most of you, of course, know that. For me personally, it’s the future proof capabilities that might be given to me through Mantle.
1. Assume that we are reaching the limits in shrinking process and the positives of die shrinks are rapidly decreasing. What if there is a serious possibility that this limit will be reached long before the new breakthrough in computing? The short term solution is to make full use of the capabilities of the current hardware, not to mention that the need for more powerful computers is growing as oppose to the hardware sales. Those optimizations are supposed to give us more time, since you cannot constantly make new architectures and new architectures might soon not be possible. So, there is the limit in current technology.

2. The increase in R&D costs (for everything) even now can be considered high as hell, which reflects on price, sales and respectively the adoption of new hardware. Please note, that the full costs cannot be automatically transferred to the customer (there is a subjective and objective limit) not to mention that this will further stagnate the market. So, there is cost related to adopting more powerful hardware.
There is a 3 and 4….simply the above is a long story with many connections to other areas so I’ll stop here.

How this affects devs. Well you have those game (and other) engines that show you how beautiful everything looks but you cannot actually implement in a game and with the small you have implemented in the game even smaller number of gamers will see. And this has nothing to do with the market share of Nvidia or AMD (or Intel). It’s one thing to see that Titan will drive that game at certain (not maxed out) settings and another is when you sit at your PC.

Now, devs are seeing the slow adoption of new hardware on one hand and on the other they see that the majority of gamers do not have high-end gaming rigs to start with. Thus new games do not meet our expectations and every game looks the same. This of course hurts sales (and not only sales). At the same time, the same developers are being held back by the hardware so, there are two things they can do. Keep the current quality (make hard but small optimizations just to include new reflections on the water ) heading to nowhere or request help for optimizing the use of current hardware.

AMD said, that this is something that developers (certainly not all, but some with vision) wanted for years. And they have not requested it from AMD specifically, that’s for sure. The problem with this is that companies such as Intel, Nvidia and AMD will never sit on one table and make a unified architecture, so that developers are eased. Here it comes back to Intel, AMD, Nvidia etc. to do something but again you see actions from no one and fears by everybody. This is where cojones and long term vision starts to matter, as there are only two outcomes. Failing or succeeding.

DirectX has been bloated for years and will not change that in a day. It only makes my heat output higher and I do believe that Microsoft knows that DirectX (as it is now) is going nowhere, plus…It doesn’t actually matter what Microsoft thinks - likes or dislikes. What matters is that certain people are seeing beyond the current status quo and it’s clear that someone has to start this change.

We know that AMD has had problems in the past and even in the present. Drivers, hardware…but truth to be told no company is immune to that. The Will to change that is what matters and most importantly vision. Somehow AMD ended in the better position and saw this as an opportunity.
What Mantle brings to the table will soon be unveiled and I am not going speculate on that, but I’m sure that AMD will throw a lot to make Mantle worth it. And worth it not for the “Titan” type of guys, as their souls are already lost. If I have 200 bucks in my pocket, it would be nice if I can buy a card and game, right?

Some developers have seen the demo presentation and know what Mantle is capable of some have jumped onboard, the rest….they’ll wait and see what others have done with it
Since we already have some devs to support Mantle I’ll put it this way. You will not see a company pushing on a technology so hard, if the possibility of failure is larger than the possibility of success. And you almost certainly will not see a dev supporting Mantle if there is a large possibility of failure or there are large costs related to it.
No one wants to perform a public hara-kiri.

What drives my confidence and hope is that some of the most innovative ideas came from AMD, so they just might have an idea to combine the request from developers with something….they have been developing for a while.
ex. What if, just if, Mantle brings something extra? Not at the beginning, as you cannot implement everything you want at once. Hey, yesterday the GPU was only showing you pictures on the screen, right.
You all must have noticed how AMD is moving in direction of Heterogeneous computing and everything the company does is connected to HSA. Is it possible that Mantle is the start of a larger unification? There are not only games you know

Will AMD succeed? Well time will tell. Is it worth to try? Definitely!

P.S. For those that constantly point to the market share of Nvidia, Intel, AMD - The market has a nasty habit of being uncertain! I’ll just point to you the almighty Nokia and Blackberry

(I am an economist who does not work for AMD, I just like markets )

 

[HisDivineOrder]

That doesn't mean that they have nothing in common with Mantle.

If you have two different low level API's for the same architecture, there has to be some kind of common code. Also, both the PS4 and XBONE API's were developed with the help from AMD.

Here is the statement from AMD that confirms what I say:

http://community.amd.com/community/...0/17/the-four-core-principles-of-amd-s-mantle

"Similar to" != "Same as."

The difference means they are WORLDS apart. One means they spend no extra money to add support. The other means they spend money to add support. The percentage of users that can benefit is incredibly small. The cost is very likely higher than doing any of a number of other things that would benefit PC gamers (resolution changes, higher quality textures, bindable keys, FOV sliders, etc) and many publishers/developers have difficulty mustering support for those.

If they won't spend the pittance to add a menu for something, I sincerely doubt they are going to suddenly be invested in adding a whole second API just to reach the small percentage of users that would suddenly be able to play the game because of Mantle. Btw, that would be a group of low end users who have not bought an APU yet because Kaveri--the only APU that would benefit--hasn't even been released yet.

So we're talking a very small group that would suddenly be able to game that couldn't before. Other than that, you have the majority of Mantle-capable computers having 7xxx products or later. Again, a ridiculously small group of users compared to the rest of PC gamers that will have older AMD products or ANY Intel/nVidia graphics.

The extra cost won't add enough users to be worthwhile for a great many publishers and there WILL be extra cost because as your link says, Mantle is only "similar to" the code they'd be using for low level access to consoles.

Moreover, PC's have higher resolutions and other assorted differences that would require investment, man hours, support, QA, and patching separate from the DX/OpenGL API.

We're talking so much more money than you guys are allowing for.

1) EA was paid $8 million dollars to support it.
2) Oxide is a new startup who needs cash and/or hardware support.
3) Eidos didn't announce any actual games running Mantle.
4) Star Citizen is supporting everything high end on PC, including Mantle, PhysX, etc.
5) Sweeney is the man behind the Unreal Engine. Sweeney is openly antagonistic toward Mantle. The majority of games that came out last gen and seemingly so far are supporting Unreal Engine or Unity. Unity is unlikely to need the high end performance of Mantle while Unreal Engine is managed by a guy who seems to loathe Mantle.

Two API's costs more to support than one unless that second API is wholly identical to one they're supporting on another platform AND that second API that is WHOLLY identical is 100% applicable to the PC. In both cases, it is not.

So it costs more to support and publishers/developers are trying to cut costs on PC ports, not raise them. There are currently NO Kaveri users who could even suddenly play Mantle games that couldn't play DirectX games and it is unlikely Kaveri will sell the numbers to make that a profitable exercise in the next year. Certainly, Llano, Trinity, and Richland didn't.

Again, Mantle seems unlikely to make an impact beyond perhaps compelling MS to push more Xbone API improvements downwind to PC and getting the OpenGL committee to work harder (re: something Valve is probably already doing).

The most important thing that will compel MS to improve DirectX, however, is actually Valve working on making OpenGL faster than DX. That's going to have the Firefox-on-IE6 effect of waking the sleeping dragon.

Mantle is Opera in this analogy.

 

[Spjut]

I think many here are thinking way too hard about this.
Hopefully we'll have our questions answered in a couple of days.

Here's what I think about some of the possible worries about Mantle that have been raised

1) If the graphics engines support Mantle, the studios using the engine won't have to spend time adding support.
And for the creators of the various graphics engines, isn't it actually a selling point for their engines to be supporting many APIs?

2) It's true that Mantle, for now, will be only for GCN. But every GPU and APU AMD releases in the next years are going to be based on GCN.

3) DX9, DX10 and DX11 also required new hardware, and a new OS for many, to use the new hardware features. Mantle not being compatible with AMD's previous GPUs and Nvidia isn't THAT different regarding adoption rates with new tech.

I'm surprised at seeing so much negativity about Mantle on a PC enthusiast forum

 

[psoomah]

SOME game developers don't agree. Many more do agree. Tim Sweeney and Carmack both seemed very opposed to the idea. They're not exactly uninformed and before you say that was at an nVidia event, they also had the guy you worship as a god for building a Mantle version of BF4 there and he seemed to imply that he saw Mantle as a means to get Microsoft to improve DirectX.

This is exactly a line in Dice's final slide at the GPU13 Hawaii presentation - "Super excited about Mantle!".

Andersson was at an Nvidia event, he was being diplomatic. What he was really thinking was 'Mantle is about to lay a 2x4 up against Nvidia's head'.

 

[Minkoff]

I just saw that...
http://www.tomshardware.co.uk/forum/id-1863987/official-amd-radeon-representatives.html#

Quote for AMA on question for Mantle support:
greenhillmaniac said:
First of all, thank you AMD for taking the time to answer our questions. My question is this: nVidia has always been in the upper hand, specially with their proprietary technologies and, their main point of advantage, deals with the most popular game brands (WB Games, 2K Games, Rockstar, ...). With the introduction of Mantle and your Gaming Evolved Program, I would like to know which game companies (aside the ones you already told us) are going to colaborate with your teams to provide better optimized games to the Radeon line in the future?

Thracks 6 November 2013 17:31:36
We're partnered with:
Eidos Montreal, DICE, Square Enix, Rebellion, Codemasters, EA, Crytek, Irrational Games, Nixxes, and more. We have an excellent relationship with almost every major studio out there, and we're in frequent contact to assist with performance optimizations for Radeon.

I'm curious if he was talking only for Mantle support

 

[psoomah]

Will AMD succeed? Well time will tell. Is it worth to try? Definitely!

P.S. For those that constantly point to the market share of Nvidia, Intel, AMD - The market has a nasty habit of being uncertain! I’ll just point to you the almighty Nokia and Blackberry

(I am an economist who does not work for AMD, I just like markets )

That was one looong post. But a good one.

Let's boil it down to it's essence.

If APU13 shows Mantle having game changing performance gains (probable) and wide developer support (now all but certain) Nvidia AIB sales will immediately fall off a cliff and stay there for a very long time.

Past market share charts are utterly irrelevant as a predictor because from that point forward AMD will own the AIB market.

It simply wouldn't make sense to buy Nvidia if AMD provides at least cost/performance parity across all games and substantial cost/performance gains on Mantle games.

Developers will see all those gamers upgrading their PCs to play their next gen games on AMD GCN GPUs and those without a Mantle rendering path will know they will be at a considerable competitive disadvantage, a disadvantage that will grow over time.

For AMD it's a positive feedback loop, the more developers come on board with Mantle, the more pressure on the remaining holdouts to join, and AMD looks to already have a critical mass of developers and game engines on board. In the end even holdouts like Sweeney will be forced incorporate Mantle into the Unreal engine because all the developers using Unreal for their games will demand a Mantle rendering path.

 

[SiliconWars]

Is it taking reference boards only? There are custom board graphics cards, that sure are quite a big part of all AIBs. Or do I understand "AIB" wrong?

No it's all boards - "Add in Boards", ie discrete graphics cards.

 

[psoomah]

Thracks 6 November 2013 17:31:36
We're partnered with:
Eidos Montreal, DICE, Square Enix, Rebellion, Codemasters, EA, Crytek, Irrational Games, Nixxes, and more. We have an excellent relationship with almost every major studio out there, and we're in frequent contact to assist with performance optimizations for Radeon.

I'm curious if he was talking only for Mantle support

He was specifically talking AMD partners that work with AMD to implement AMD hardware optimizations, not Mantle partners, but it is likely they will turn out to be one and the same. And that would be one hell of a foundation for Mantle.

I am intrigued as to the "almost every major studio" bit - Epic?

 

[Minkoff]

And a little bit more...answers from Robert Hallock

Quote:
Question: 6.b.) All we know from AMD as yet about Mantle is that it can provide up to 9x more draw calls. Draw calls on their own shouldn't mean too much, if the scenario is GPU bound. You suggest that it'll benefit CPU-bound and multi-GPU configs more (which already have 80%+ scaling).
That said, isn't Mantle more of a Trojan horse for better APU performance, and increased mixed APU-GPU performance? AMD's APUs are in a lot of cases CPU bottle-necked, and the mixed mode performance is barely up to the mark.

Answer: I suggested that it’ll benefit CPU bottlenecking and multi-GPU scaling as examples of what Mantle is capable of. Make no mistake, though, Mantle’s primary goal is to squeeze more performance out of a graphics card than you can otherwise extract today through traditional means.

Question: 6.d.) Developers have said they'll "partner" with you, however the only games with confirmed (eventual) support are BF4 and Star Citizen. Unreal Engine 4 and idTech don't seem to support Mantle, nor do their creators seem inclined to do that in the near future.
Is that going to change? Are devs willing to maintain 5 code paths? It would make sense if they could use Mantle on consoles, but if they can't...

Answer: The work people are doing for consoles is already interoperable, or even reusable, with Mantle when those games come to the PC. People may have missed that it’s not just Battlefield 4 that supports Mantle, it’s the entire Frostbite 3 engine and any game that uses it. In the 6 weeks since its announcement, three more major studios have come to us with interest on Mantle, and the momentum is accelerating.


Question: How do you convince developers to use Mantle when other APIs are vendor neutral?
For existing games, how hard or easy is it to "port" it to use Mantle?...

Answer: We haven't had to convince them. Every single one of them has come to us and asked for it without prompting!

 

[Paul98]

If the big increase in FPS going from windows 7 to 8.1 is caused by the new DX version, we can expect big things from mantle especially on windows 7 machines. I hate having to wait to see haha

 

[ShintaiDK]

If the big increase in FPS going from windows 7 to 8.1 is caused by the new DX version, we can expect big things from mantle especially on windows 7 machines. I hate having to wait to see haha

In BF4? It was just an increase between 2 beta patches wasnt it. The next one brought win7 performce up to win8.1.

 

[smackababy]

In BF4? It was just an increase between 2 beta patches wasnt it. The next one brought win7 performce up to win8.1.

I thought BF4 Windows 7 performance was due to some kind of thread issue only exhibited in Windows 7. Windows 8.1 performed normally.

 

[Paul98]

In BF4? It was just an increase between 2 beta patches wasnt it. The next one brought win7 performce up to win8.1.

No, windows 8.1 still runs faster for people who had no problems in windows 7 by a good bit. BF4 does take advantage of new DX functions in 8.1 so that very well could be the reason. Though we will see if there are more patches introduced or if dice says something about it.

 

[krumme]

And a little bit more...answers from Robert Hallock

Quote:
Answer: The work people are doing for consoles is already interoperable, or even reusable, with Mantle when those games come to the PC.

This. Notice how he answers the question

 

[MutantGith]

From what I understood (as someone that hasn't played BF4 yet) is that the issue was that many (not all) users with Win 7 were having problems with CPU core parking being overly aggressive when dealing with BF4, leading to their CPU being wildly under-utilized.

Win 8/8.1 handles thread scheduling and load balancing differently, and so was not as prevalently impacted.

There was a fix circulating that used Regedit/a patcher program to disable the aggressive CPU core parking. Once that happened, suddenly people that had that fix work (a fair number, anecdotally) saw much improved framerates, less dips and smoother loading and play.

Interesting, given the big draw from Mantle, so far as anyone expressly knows, is to reduce the API overhead from DX - typically something that is CPU bound - as DX has to interface with lots of other memory systems to make draw calls etc.

Patching the OS registry to more fully use all the available cores suddenly relieved the surprisingly massive CPU bottleneck, and people (again, anecdotally) reported getting very clean and smooth frame rates with even older hardware (570's and 68/6950's).

There are some links over in PC gaming, I think, including some pretty interesting gameplay footage with a CPU usage/Framerate overlay with and without the fix, if people are interested.

 

[MutantGith]

This. Notice how he answers the question

I would also point out that that could really be said of any multi-platform release title going back to the beginning of the last generation of consoles.

It's not as though people that use Unreal Engine 3 to do something like...Bioshock back in the early days had to completely recode the game from scratch three times to release on PS3, Xbox 360 and then PC. They're using an engine to do the bulk of the work, and then taking that base of code and assets, reusing them to output games to each pipeline.

The statement, vague as it is, is just as applicable to a work flow that looks like that as it is to the utopian model of being able to just hit a compile button three times and do no other work to relase on PS4, XBOne and PC that some people are implying with Mantle.

Without specifics, and ideally a post-release summary from a developer that actually used Mantle all the way through, I'm not sure how much we can fairly or accurately speculate on these merits. I think the fact that there is a definite, clear break in PC gaming that this creates between GCN cards and everyone else is the core of some people's interests, and concern.

Nothing I've heard makes that differentiation logically go away, unless we fast forward in time to a point where there is no other manufacturer in PC video cards, other than AMD, and therefore everything is GCN by default. That doesn't seem like a rosey future to me, personally.

 

[psoomah]

Answer: The work people are doing for consoles is already interoperable, or even reusable, with Mantle when those games come to the PC. People may have missed that it’s not just Battlefield 4 that supports Mantle, it’s the entire Frostbite 3 engine and any game that uses it. In the 6 weeks since its announcement, three more major studios have come to us with interest on Mantle, and the momentum is accelerating.


Question: How do you convince developers to use Mantle when other APIs are vendor neutral?
For existing games, how hard or easy is it to "port" it to use Mantle?...

Answer: We haven't had to convince them. Every single one of them has come to us and asked for it without prompting!

Thanks!

I consider that definitive - Mantle is flat out going to enjoy MASSIVE developer and engine support and the vast majority of next gen AAA games will have Mantle at release, or added post release.

"Every single one of them has come to us"

No way Sweeney will be able to hold out on Mantle when the many developers using the Unreal engine for their next gen games start demanding Epic provide a Mantle path for their engine.

I don't see how any major developer will be able to ignore Mantle.