The AMD Mantle Thread

[ShintaiDK]

My guess is that since it's low-level AMD already has planned and taken into account Linux. PS4 is also Linux based and devs will be able to use Mantle there as well for low-level control on GCN.

FreeBSD is not Linux.

 

[Lonyo]

http://vr-zone.com/articles/mantling-alliances-ritchie-corpus-amd-interview/58215.html



So, DICE only. No specs or SDK until...who knows when. Not exactly something that will be adopted anytime soon. Assuming it ever will be.



No Mantle for SteamOS.

SteamOS is just Linux.
And considering Valve will have a hell of a job just getting AAA games developed for Linux, I doubt having AMD also try and push Mantle would be of any use at the moment.
Plus, Valve are still running on DX9 level tech, they wouldn't care about Mantle anyway because their engine is so antiquated.

Not sure how Valve have any ability to dictate whether Mantle will be available for Linux games that require Steam, since they don't make the engines for most of the games on Steam.
Their own engine is out of date and wouldn't make use of Mantle anyway.

Remember, "SteamOS" is just skinned Linux running Steam as far as has been made apparent. Valve don't control Linux, Valve don't control the engines of other games.

 

[Shivansps]

Its funny consoles keep being mentioned. Yet nowhere is it stated that Mantle is used for consoles.

Not to mention there isnt even tools for Mantle out or a spec for companies to design after. And its not coming anytime soon either as stated by AMD.

Personally ill like to know how its even possible to share the same low level API when you have hUMA on PS4, ESRAM on xbox, and pcs with dgpus have have neither... no way optimisations will be the same.

The console thing was just a misinformation, the slide from DICE says "console style", meaning "low level", no compatible or same api.

 

[Shakabutt]

Personally ill like to know how its even possible to share the same low level API when you have hUMA on PS4, ESRAM on xbox, and pcs with dgpus have have neither... no way optimisations will be the same.

The console thing was just a misinformation, the slide from DICE cleary says "console style access", meaning "low level", no compatible or same api.

We are talking about the graphics parts of the console API...which uses GCN cores the same as the GCN cores in desktop radeons.

If console devs push heavy optimizations for GCN on consoles there is a high probability that those optimizations will come over to Mantle .

If there was some other provider for GFX tech on consoles other than AMD then i would probably be highly skeptical about Mantle, but since they got their fingers in both consoles with GCN tech, i think you guess what their strategy was all along, they are not chasing the profits from the chips on consoles, they want their tech at heart of the gaming, imo the console contracts are a big win for them in the long run.

Funny thing i noticed since the reveal of Mantle, Geforce cards are selling like crazy on my local gaming forums/Comunities, 3 geforce gtx 680 are up for sale from yesterday, for like 200 EUROS.

 

[Lonyo]

Personally ill like to know how its even possible to share the same low level API when you have hUMA on PS4, ESRAM on xbox, and pcs with dgpus have have neither... no way optimisations will be the same.

The console thing was just a misinformation, the slide from DICE says "console style", meaning "low level", no compatible or same api.

HUMA isn't a PS4 only thing. No one is entirely clear on what the HUMA status is of consoles/etc. AMD stated that when someone said only PS4 had HUMA, she was incorrect.

Also, as people have said, you don't need the same memory architecture to be able to take advantage of the GPU core.

 

[Haider]

2) To my eye, Mantle is like Glide in that the company that perceives itself to have dominance in GPU market share is trying to capitalize on that dominance by shutting the other people out of the "high performance" niche. High performance space is where the marketing for the lower end products happens; people hear the R9 290X is the best of the best, so the R7 260 sells better, etc. Mantle is about selling the high end AMD card as the only card capable of hitting the highest of the high end on the "benchmark" titles that AMD has been targeting for the last year with Gaming Evolved.

4) That brings us to Mantle and DirectX/OpenGL. Two is too many codepaths. You speak of porting from Xbone/PS4 to PC like it's nothing, but it is not nothing. Even if the exact same code from Xbone low-level to Mantle could work, it would not be an inconsequential amount of money, time, or support to maintain a second, separate codepath for a new API. Sure, let's say hypothetically it's COMPLETELY identical to the Xbone's low level API (this is speculative and unlikely as "completely identical" seems iffy), you'd still have to code for PC optimizations, right? I mean, lots of talk about Mantle implies that GCN 2.0 is the same as GCN 1.0, that the Xbone version of the game is as good as the PC version is going to be (resolution? better textures? farther draw distance? faster framerate?) with no customization or improvements.

Somehow, I doubt that EVERY self-respecting PC gamer is going to take an untouched, unimproved port of Battlefield 4 running at the Xbone's resolution of 720p/1080p and be content with that. Certainly, PC gamers are going to want more. That doesn't just magically appear after you code for Mantle. You've got to code even more for Mantle to make that happen and work well. Who works on the Mantle code? The same guy working on Direct3d for PC? Or the guy working on the Xbone low level API? What if the PC version gets a new feature, does the Mantle version get it? So the PC version guy does it, right?

What if the low level Xbone API gets further optimized? I guess now the Xbone API guy goes in and alters Mantle, but what if that breaks some of the PC-specific code in Mantle for the PC version of Battlefield 4 Mantle? Who fixes what? Clearly, money is invested to maintain different parts of the code and since Mantle is a whole new API, you have to maintain it somehow. At best, a developer might have a whole other group handling Mantle smoothing and at worst you'll have the same overworked employee doing both Direct3D and Mantle, splitting his time between them both rather than just plain optimizing Direct3d alone.

5) This brings us back to the past and Glide. Even developers who favored Glide in a Glide vs Direct3d vs OpenGl threeway were forced to eventually code to the standards. In due course, it became clear that holding such a place of dominance with Glide had led 3dfx to avoid changes that might imperil their dominance-assuring technology, Glide. But those changes led to inferior technical leaps when they long, long clung to ideas for 3d tech that were far behind the curve. Developers who'd long stayed with Glide eventually realized they had to move to DirectX or OpenGL to keep up with the changing pace.

Suddenly, 3dfx's reliance on Glide and their reluctance to leave it behind had led to them being way, way behind in technical design. In a panic, they then made a series of boneheaded business decisions that ended them abruptly.

But the lesson to be learned is still that letting one house control an API in the interests of "superior performance" leads to stagnation and a reliance on that specialized API to maintain their lead and eventually to keep up.

Regardless, that matters little. Glide made sense not because other companies had their own API's back then, but because Direct3d and OpenGL were not ready when it first showed up. Glide was a way of bringing OpenGL into the modern before it was done. That's the only reason it existed.

Mantle is not about bringing 3d to gaming like Glide was. In this way, AMD is worse than 3dfx. AMD wants to own the market, plain and simple. They don't want to spend obscene amounts of money maintaining a pace with companies that are far richer than them: nVidia and Intel. They know they will lose that war eventually as they'll run out of money long before they get there. It's cheaper to give MS and Sony good deals on next gen hardware, then use that as a way to win over more developers to their own specialized API across all platforms. It's clever, if disastrous in the long term since nVidia and Intel will respond in kind. Any money saved by a developer/publisher by "just coding" for Mantle could be more than compensated by either nVidia or Intel, both of whom have a lot more money than AMD.

That's assuming any developer would want to support two API's.

So they want to change the rules of the game. They get to spend less and still be far ahead. Plus, they can finally have a chance of having games and PC's use HUMA, which was heretofore unlikely when nVidia and Intel have little reason to support it. But Mantle sidesteps that entirely, for the first time offering a compelling reason for gamers to buy into an all-AMD PC. Again, it's a clever workaround to the fact that Intel and nVidia refuse to work with them, plus MS is throwing their hands up about PC gaming in general.

6) Direct3d and OpenGL were a response to multiple API's. Everyone who talks about Mantle acts like nVidia and Intel couldn't make their own, initiating an API war. If you think Intel doing it is a joke, you really need to look at how much money that company is sitting on and how many GPU's it's selling per year. Do you look forward to a day when publishers are outright paid to focus only on the Intel GPU codepath, leaving the Direct3d one to be the "safemode?" Or unsupported? Perhaps nonexistent? How many gamers right now with a discrete GPU made by nVidia or AMD have an Intel GPU on their CPU, waiting for its day? Lying in wait, like sleeper agents.

AMD is doing a clever move with Mantle, but if you think developers/publishers aren't going to support other GPU vendors because Mantle's so much easier, you really need to remember that the 360 version of games were often the first version and that had AMD hardware for eight years now. How many of those AMD-first games were then ported over to nVidia hardware and suddenly became TWIMTBP'ed? It was only in the last year-ish that Gaming Evolved actually evolved into something worthwhile and it wasn't because of b/c with the AMD GPU in the 360 (or Wii). It was because money talks. Suddenly, these companies had deals for thousands and thousands of licenses of games from these companies. AMD paid them money and they made Gaming Evolved for AMD.

So what makes you think that the pittance of savings of some code that Mantle offers will be more tempting than the truckloads of money Intel or nVidia could hypothetically offer? It may be easy to forget AMD is in extreme financial straits, but there's still a lot of talk about them going out of business and even if that is (somewhat) overblown, they don't have a ton of money to push Mantle with.

And their competitors do.

7) But all that comes back to the obvious. If money is not a factor (because AMD's competitors are far richer) in AMD's favor, then ease of use and performance improvement would have to be incredible to be compelling to developers to use a SECOND additional API for PC gaming than the seemingly required DirectX/OpenGL one.

Even if a developer went out of his way to use Mantle for the next gen consoles (or if Xbone's low level API were based on Mantle), that doesn't mean they'd go to the trouble of applying that to PC gaming. "Why wouldn't they, it's easy as pie?" It's not. It's really not. Anyone who says that doesn't get what it takes to make a game.

More than that, the performance improvement would have to be incredible to warrant such an investment for any developer, which almost assuredly locks out most developers who just want compatibility. Of the select few who are making high end niche benchmark titles, you've got some that are nVidia-friendly and some that are mercs, up to the highest bidder. Money, again, speaks louder. If nVidia senses a real threat on its homefront, nVidia will turn around and start spraying money like it's confetti.

So the market that will use Mantle remains small. It may be (somewhat) easier to code to from the next gen consoles, but it's still a second API. It's not the primary one, not like Glide was. The other API's are mature, stable, works with everything. Not like when Glide was king. And games are already in the stable with DirectX.

AMD has to make a case for why anyone should chase after such a small percentage of the market as AMD represents. Hell, even their own APU's won't support Mantle by and large yet. That's the one area where Mantle makes more sense since it could be used to create PS4/Xbone-like optimized games for low end PC hardware...

8) AMD wants Mantle for the same reasons that 3dfx clung to Glide, but Mantle is going to fail for all the reasons 3dfx was supplanted by DirectX. Not because Glide suffered for all the alternate API's. In most cases, it won out because it was the first and rather than deal with the chaos, developers just went with the most mature (Glide). No, Glide just didn't keep up with the times and maintaining its features kept 3dfx from evolving.

That's what's going to happen to AMD if they aren't very, very cautious because when you code games "to the metal," then you can't really change much about the layout of said metal or things start breaking.

9) Mantle is a move of desperation by a company that doesn't see itself changing much in the way of its technology over the next ten years. It's clever and it's going to send some shudders through the PC gaming space, but in the end I think something like Valve's Steambox based on Linux will have more impact on finally freeing us of the tyranny of an OS run by a company obsessed with gaming consoles and not PC gaming.

You make some valid points but the balance is bit biased against AMD.


SteamOS/box is interesting but can they gain developer support? What does it bring to the party? It has to do more than Windows. Why should I move to SteamOS? Will my games run faster and better? Will games like GTA V come out at the same time as the console release? If GTA V comes out on SteamOS before Windows I will install it. Are you listening Valve? There is disaffection in PC gamer world with MS, strike while the iron is hot. I am not happy with MS. MS Windows division should be talking to Rockstar, give them some money for Windows 8 exclusive eye-candy/features...


Agreed Mantle won't want to be used by every title just the AAA titles, the headline acts. Too true.


AMD want to be the top-dog or the only-dog that matters in the graphics market. Whilst it is true but it is disingenuous to suggest nVidia and Intel don't...


nVidia and Intel could come out with a low-level API. This is the key difference AMD believe that they can get the AAA software houses to use theirs in the AAA titles. Time will tell there is conjecture for why this may happen.


AMD want Mantle to deliver performance that outstrips their competitors on headline games. The big headline grabbing games not every game. They believe they have developed that technology. Proof of the pudding's in the eating. Let's see what performance advantage for GCN BF4 delivers - £/fps , consistency of frame-rate, any extra eye candy etc it will be interesting to see what is delivered...

I have no brand loyalty just to my wallet. if AMD can deliver it so be it, they are no better or worse than any other corporation, don't kid yourself by thinking nVidia is anymore altruistic.

 

[Stuka87]

nVidia and Intel could come out with a low-level API. This is the key difference AMD believe that they can get the AAA software houses to use theirs in the AAA titles. Time will tell there is conjecture for why this may happen.

I think AMD being in the consoles is what they are hoping will propell this on desktops. nVidia and Intel an both develop something similar, however it wont be the "push button" porting that it will be for mantle (No it isn't as simple as that, but compared to porting to DirectX, it may as well be).

I too am very interested in performance figures. I will not be surprised one bit if there is an average 20% performance boost.

 

[NIGELG]

Don't hype this up too much Kenmitch (and others) as you are likely to be let down if expectations are as high as Everest. I mean, learn.
It'll be given a chance just like anything else, and I hope they can even do half as good as they are pimping.

Do you use words like 'pimping' when Nvidia is concerned?

 

[wand3r3r]

I'm not sure if this was posted.

"AMD’s Mantle is the biggest change to gaming in a decade"

AMD’s GPU14 public presentation had one big surprise at the end called Mantle, and it will change gaming. Mantle is one of those shots across the bow of the industry, the biggest industry upheaval in a decade.

Charlie's opinion.
http://semiaccurate.com/2013/09/30/amds-mantle-biggest-change-gaming-decade/

 

[bgt]

I'm not sure if this was posted."AMD’s Mantle is the biggest change to gaming in a decade"AMD’s GPU14 public presentation had one big surprise at the end called Mantle, and it will change gaming. Mantle is one of those shots across the bow of the industry, the biggest industry upheaval in a decade.Charlie's opinion. http://semiaccurate.com/2013/09/30/amds-mantle-biggest-change-gaming-decade/

Good find. Looks very promising:thumbsup:

 

[Keysplayr]

n/m

 

[Kenmitch]

n/m

Dont sweat it. If anything pimping was a bad choice of words.

The internet is pimping Mantle currently not AMD.

 

[Keysplayr]

Dont sweat it. If anything pimping was a bad choice of words.

The internet is pimping Mantle currently not AMD.

Nigel did not like that at all. take it back. :sneaky:

Seriously though, I've never seen something pushed so hard yet not a single result seen nor heard of. It's like pre-ordering a 290X.

 

[VulgarDisplay]

Nigel did not like that at all. take it back. :sneaky:

Seriously though, I've never seen something pushed so hard yet not a single result seen nor heard of. It's like pre-ordering a 290X.

The 290x is hardly being talked about. Its just caught up in the mantle hysteria.

 

[SiliconWars]

Nigel did not like that at all. take it back. :sneaky:

Seriously though, I've never seen something pushed so hard yet not a single result seen nor heard of. It's like pre-ordering a 290X.

You forgot about Fermi already? What about the numerous annoucements of reveals of Fermi?

 

[Haider]

Saylick, its more like 2.5x the performance boost of console hardware versus PC going through DX, that figure is touted by big name devs as the DX overhead.

The question is how much of that 2.5x will translate to PC running GCN. I am not overly optimistic because a lot of the gains on the console is due to lifting CPU bottlenecks as well as GPU optimizations. On the PC, the CPU is rarely the bottleneck. Still, I expect it will be significant, else why would devs be interested in changing their game engine to support a new API?

Yep I have come across that figure from respected developers too. I'm interested to see what Mantle delivers. DirectX needs improvement. Frankly it's an absurd situation. Mantle should not shift the focus from this. Consumers need to put pressure on Microsoft, nVidia, AMD and Intel.

 

[smackababy]

Yep I have come across that figure from respected developers too. I'm interested to see what Mantle delivers. DirectX needs improvement. Frankly it's an absurd situation. Mantle should not shift the focus from this. Consumers need to put pressure on Microsoft, nVidia, AMD and Intel.

Does anyone have any real quotes from "respected" developers saying this? Because I've talked with quite a few developers and they don't believe DirectX is really that slow, especially compared to the ease of use and standarification it brings to PC development.

"2.5 times performance" is a rather bold claim, especially if you don't have any meaningful test to back it up.

 

[blackened23]

Does anyone have any real quotes from "respected" developers saying this? Because I've talked with quite a few developers and they don't believe DirectX is really that slow, especially compared to the ease of use and standarification it brings to PC development.

"2.5 times performance" is a rather bold claim, especially if you don't have any meaningful test to back it up.

Both John Carmack and Timothy Lottes have spoken outright disgust as recently as this year and last regarding DirectX 11. Both stated that the capabilities of PC would widen significantly with direct hardware LibGCM style APIs - As far as the actual performance gains of Mantle, who knows. That variable is very much still in the air.

 

[3DVagabond]

Yep I have come across that figure from respected developers too. I'm interested to see what Mantle delivers. DirectX needs improvement. Frankly it's an absurd situation. Mantle should not shift the focus from this. Consumers need to put pressure on Microsoft, nVidia, AMD and Intel.

The only way to put pressure on corporations is by hitting them in the pockets. If this gives AMD the performance boost it should, you'll see those companies respond. You'll see real performance growth.
What really needs to be done is M$ needs to do some serious upgrading to Windows. This opportunity exists solely because M$ has gotten complacent. Their new products lack vision.

Strange that those who are against this as anti competition and fragmenting the market don't have any issues with SteamOS. Devs are already writing for Mantle. Why is nobody concerned/objecting to devs having to do "oh so much work" to port to SteamOS, but it's just going to be such a damned effort to port the console optimizations over?

 

[VulgarDisplay]

If anyone should be pissed about this its valve. They talked about how steam os will be great because it will do away with windows/DX overhead. Now and can do the same thing and I don't have to worry about dual booting. I was excited for steam is until I heard the mantle news.

For that Microsoft should thank them. No has a reason to abandon windows anymore.

 

[Keysplayr]

You forgot about Fermi already? What about the numerous annoucements of reveals of Fermi?

Think it's similar to this do you? Glad you remember because I surely do not.

 

[Fastx]

Just thought I'd post this FWIW dated 9-30-13 5:04pm.


VR-Zone had a chance to chat with the man that heads up AMD’s developer relations unit, Ritchie Corpus, about some of the developer relations focused initiatives AMD announced at its keynote.


VR-Zone: Proprietary standards don’t have a great track record of success , historically speaking. Why will Mantle be different?

I think at this stage it makes sense for us to develop Mantle, at least in its current form, because nobody knows our hardware at the lowest level best than we do. So for us to have to do that for alternative graphics hardware [would be] almost impossible.
The plan is, long term, once we’ve developed Mantle into a state where it’s stable and in a state where it can be shared openly [we’ll make it available]. The long term plan is to share and create the spec and SDK and make it widely available. Our thinking is: there’s nothing that says that someone else could develop their own version of Mantle and mirror what we’ve done in how to access the lower levels of their own silicon. I think what it does is it forges the way, the easiest way.
If you think about it, Mantle is truly a collaboration and I’ll tell you [DICE’s Johan Andersson] was at the forefront of that at the very beginning. A lot of the feedback on the development of Mantle came from him. We also solicited feedback from a lot of other partners that we haven’t announced yet. At this stage, Battlefield 4 and FrostBite 3 are the closest to deliver something today. I think, as I mentioned before, the goal would be to provide the spec and SDK publicly.


VRZ: Any timeline to Mantle’s SDK being made public?
It could be as early as sometime next year or maybe the year after.


VRZ: A lot of today’s keynote was about AMD’s hegemony in the gaming sector. How are you sure that Mantle and AMD’s place as the chipmaker for next-generation consoles will make it the dominant platform for PC gaming in the future?


I think there is a lot of similarities in the architectures of the [next-generation] consoles and PC side. I think one of the benefits that game developers are going to realize is that because of their strong familiarity with the PC already, and developing on our hardware, we anticipate development budgets to be somewhere less or lower because they won’t have to spend as much development time and the learning curve is much shorter. We don’t think it’s going to take them as long to extract all of the features and the maximum performance benefits out of our hardware.

http://vr-zone.com/articles/mantling-alliances-ritchie-corpus-amd-interview/58215.html

 

[smackababy]

Both John Carmack and Timothy Lottes have spoken outright disgust as recently as this year and last regarding DirectX 11. Both stated that the capabilities of PC would widen significantly with direct hardware LibGCM style APIs - As far as the actual performance gains of Mantle, who knows. That variable is very much still in the air.

The problem is, outside of Carmack and Lottes, who can actually code the high level API DX processing to interact with the low level access very good? Not the $40k a year code monkeys that the majority of game studios hire. The best comparison I can come up with off the top of my head is Java and hibernate. Sure, a great programmer can make something that has performance increase, but it sure isn't entry level and a lot of developers aren't going to be reusing it.

What I'd hope is that MS will respond with DX12 allowing direct hardware access and allowing you to skip unneeded steps in their stack.

 

[BoFox]

I don't know how many times this has been said, but if you buy AMD you don't get their X86 license.

ARGH, thanks for the reminder! Maybe there could be another way to do it - a mutual joint-venture and "invasion" of AMD by Qualcomm.. :ninja: (like Time Warner and AOL partnership that required an overhaul of AOL's board with TW execs - although it was a stupid move)

 

[Shivansps]

What I'd hope is that MS will respond with DX12 allowing direct hardware access and allowing you to skip unneeded steps in their stack.

Guys it cant be done, thats what the abstraction layers are for. All hardware whould be have to be exactly the same to share any kind of low-level code.

Now Mantle works on PC and PC only, until information indicate otherwise, its PC only, its impossible to share low level code with xbox because of the ESRAM, its a critical GPU resource not present on pc.
Also no one will code for Mantle for consoles if its not inside the xbox/ps4 sdks and documentation.

Mantle CANT run on Nvidia or Intel and at the SAME TIME, share the exact same low level code as AMD, other hardware, other registers, other resources, etc, in order to do that you need an abstraction layer, its what DX/OpenGL are for.

Pure and simple, DX its not "high level and inefficient" for nothing, there is a reason.
You just cant allow "low level" access whiout breaking compatibility.

Its the basics of understanding low-level code, assembler and abstraction layers.

Low level api may seem to be a good thing, but think about Nvidia/Intel response to this, we really want a pc gaming world when Intel, Nvidia and AMD have their own low level apis? think about the devs a second, we all know where this is going. To devs choosing 1 low level API and having a very crappy DX/Opengl implementation.