Mantle API presentation by AMD, DICE and Oxide - AMD Summit 2013

[TheoneandonlyMrK]

Well I think I disagree on how much nvidia matters to mantle to say the least.
I did recently note a verdant nvidia owning friend has refused to buy his fave ever game bf4 and he loves it and Nv interesting.
Shit hurry up and switch mantle On already in bf4.

 

[HTC]

The way i understood it is that any GPU with GCN tech will be able to use Mantle immediately while others will have to have "help" be it from special drivers or, worse, from hardware changes.

If this "help" comes from special drivers, i expect a performance hit because, instead of the game "using" Mantle to "talk" to the graphics card, it will have to use "a translator" because, without CGN tech, it can't "talk" to it directly. This, to me, is what makes the most sense as it explains why Mantle is multi-vendor and also why it works best with CGN tech: if there's a need for "a translator" for other vendors, time will be lost, hence the performance hit.

If this help requires hardware changes, then i don't see Mantle surviving very long in what will be, for all intents and purposes, a proprietary API due to the CGN being AMD tech only.

 

[bpgt64]

Show me numbers and I'll care. Or Wizzards review.

 

[RoutedScripter]

I find it odd after all the people that have pointed out things you're not aware of or understanding of in this thread you still insist we're clueless. No one knows the exact numbers percentage wise of devs in the industry that have made a specific request for AMD or anyone else to come up with an alternative to Direct3D, and I'm sure most of them, unlike you, are understanding of the fact that if not implemented carefully, the same thing could happen that did with Glide. You keep glossing over that. Like I said earlier, you're ignorant of history in the industry in this regard. If you can't even grasp that simple concept, then no, you have not in fact "researched" this so well. I find it odd you'd even use the word research as naive as you sound.

If you had a shred of sense you'd angle this thread toward things that AMD/DICE have done to ensure Mantle will not suffer the same fate as Glide, but instead you insist it's a non issue. There really isn't enough proof that they've done much to make devs think Nvidia and their customers will support this, and THAT'S what matters most, not your narrow minded perception. If all that mattered were the opinions of a few naive AMD fanboys there'd be nothing to worry about, but there's a whole lot more to it than that. You keep insisting we're pessimistic skeptics. No, we're just realists that aren't easily convinced by a few stage presentations.

You and xenocide definitely didn't spend 5 hours watching all the presentations, slides, and even all the other research and knowledge about how much of a difference low-level APIs make.

A good experienced man can predict things, you are inexperienced, you need hard proof to be able to see that. Just like the obamacare supporters, they had to see it written in their mail that they would have to pay 3 times as much, but we who were telling this would happen for months prior, because we done the research, they didn't take it seriously.

You can keep saying whatever you want, but it's only your perception, which obviously doesn't affect reality.

Well I think I disagree on how much nvidia matters to mantle to say the least.
I did recently note a verdant nvidia owning friend has refused to buy his fave ever game bf4 and he loves it and Nv interesting.
Shit hurry up and switch mantle On already in bf4.

I can't even read that ...

The way i understood it is that any GPU with GCN tech will be able to use Mantle immediately while others will have to have "help" be it from special drivers or, worse, from hardware changes.

If this "help" comes from special drivers, i expect a performance hit because, instead of the game "using" Mantle to "talk" to the graphics card, it will have to use "a translator" because, without CGN tech, it can't "talk" to it directly. This, to me, is what makes the most sense as it explains why Mantle is multi-vendor and also why it works best with CGN tech: if there's a need for "a translator" for other vendors, time will be lost, hence the performance hit.

If this help requires hardware changes, then i don't see Mantle surviving very long in what will be, for all intents and purposes, a proprietary API due to the CGN being AMD tech only.

Wrong

Nvidia would have to code in support, has nothing to do with mantle being on GCN only currently.

Show me numbers and I'll care. Or Wizzards review.

Here are the numbers

CPU boost; ~600%
GPU boost: 100%-300%

or in other words:
"amazing"


But as I said, this will be spent on better visuals, more draw distance, more complexity, so if you check back 2 years this post, this numbers will not match, but if you had same visuals, same complexity, then it would hold, this can be done great with something like 3DMark, they should keep improving the performance optimizations, without any fidelity and load stuff, to see clearly see the difference.

 

[Frag_Maniac]

You and xenocide definitely didn't spend 5 hours watching all the presentations, slides, and even all the other research and knowledge about how much of a difference low-level APIs make.

This just makes it all the more obvious you are not paying attention and narrow mindedly focused on one thing. I never argued that such an API can improve performance. Hell, I doubt most devs have. The lesson learned from Glide was not just focusing on performance but rather it being universally beneficial, meaning including Nvidia customers. No matter how many times that is said it just goes right over your head.

Only a fool studies the obvious and ignores the most pertinent details. It's as if someone needs to draw you a picture at this point.

 

[HTC]

The way i understood it is that any GPU with GCN tech will be able to use Mantle immediately while others will have to have "help" be it from special drivers or, worse, from hardware changes.

If this "help" comes from special drivers, i expect a performance hit because, instead of the game "using" Mantle to "talk" to the graphics card, it will have to use "a translator" because, without CGN tech, it can't "talk" to it directly. This, to me, is what makes the most sense as it explains why Mantle is multi-vendor and also why it works best with CGN tech: if there's a need for "a translator" for other vendors, time will be lost, hence the performance hit.

If this help requires hardware changes, then i don't see Mantle surviving very long in what will be, for all intents and purposes, a proprietary API due to the CGN being AMD tech only.

Wrong

Nvidia would have to code in support, has nothing to do with mantle being on GCN only currently.

You just said what i highlighted, dude: that's what i was referring to as "translator".

Let me put it this way:

• Mantle uses CGN to "speak" with the game
• Since nVidia doesn't have CGN, it'll need a translator which is where the support code comes in
• Since nVidia will have to use an extra step for Mantle, it has to process more data then AMD which means there's a hit in performance

 

[Zakin]

This just makes it all the more obvious you are not paying attention and narrow mindedly focused on one thing. I never argued that such an API can improve performance. Hell, I doubt most devs have. The lesson learned from Glide was not just focusing on performance but rather it being universally beneficial, meaning including Nvidia customers. No matter how many times that is said it just goes right over your head.

Only a fool studies the obvious and ignores the most pertinent details. It's as if someone needs to draw you a picture at this point.

If you haven't noticed, he notoriously replies to what he wants to read, he did it with my posts a few pages back. When I pointed this out he flat out ignored me, I'd just drop it and be happier forgetting about this honestly. It's incredibly rude the tone he has when he responds to people.

 

[RoutedScripter]

This just makes it all the more obvious you are not paying attention and narrow mindedly focused on one thing. I never argued that such an API can improve performance. Hell, I doubt most devs have. The lesson learned from Glide was not just focusing on performance but rather it being universally beneficial, meaning including Nvidia customers. No matter how many times that is said it just goes right over your head.

Only a fool studies the obvious and ignores the most pertinent details. It's as if someone needs to draw you a picture at this point.

It sounds that way it really does, I don't ignore details, I just maybe do admit I am kind of getting sick so It sounds like I'm closed minded, but because of your perception it's a spin you add on your conclusion that is the problem, and everything can be spinned out of context and in a certain way, just not getting some of the stuff right makes your story totally different.

And not everything, I'm particularly pointing to the jumping of conclusion and constant comparing to Glide, it's not 1998 anymore, it's totally differend landscape of the industry, there are developers that will use this thing to push the standards up and that will pull up the rest of the industry, the other part is the extra benefit of consoles being on GCN so developers would not be wasting time re-writting the optimization code, some of it yes, some converted but far easier than from scratch, that's what is the money&time saver right there.


I wouldn't be saying any of this if I wasn't sure.

You just said what i highlighted, dude: that's what i was referring to as "translator".

Let me put it this way:

• Mantle uses CGN to "speak" with the game
• Since nVidia doesn't have CGN, it'll need a translator which is where the support code comes in
• Since nVidia will have to use an extra step for Mantle, it has to process more data then AMD which means there's a hit in performance

No, i didn't meant that.

If Nvidia had to do that, makes no point using the new API.

It doesn't use GCN to speak with the game, GCN is an architecture, it uses Mantle to speak with the game. Mantle supports GCN currently and It may later support more GPUs because it's open standard and Nvidia can put their code straight beside AMD's

 

[Frag_Maniac]

It sounds that way it really does, I don't ignore details, I just maybe do admit I am kind of getting sick so It sounds like I'm closed minded, but because of your perception it's a spin you add on your conclusion that is the problem, and everything can be spinned out of context and in a certain way, just not getting some of the stuff right makes your story totally different.

And not everything, I'm particularly pointing to the jumping of conclusion and constant comparing to Glide, it's not 1998 anymore, it's totally differend landscape of the industry, there are developers that will use this thing to push the standards up and that will pull up the rest of the industry, the other part is the extra benefit of consoles being on GCN so developers would not be wasting time re-writting the optimization code, some of it yes, some converted but far easier than from scratch, that's what is the money&time saver right there.


I wouldn't be saying any of this if I wasn't sure.

But that's just it, there's really no way to be sure. Sure a lot of things have changed in the industry, but mostly technically. The one thing that hasn't changed is that AMD and Nvidia still take their own separate paths. This in reality is a discussion of potential, POTENTIAL mind you, tech sharing. Name ONE piece of tech as significant as Mantle that AMD and Nvidia have shared that was created by either side? There's no previous model for how this will play out other than Glide, so obviously many are using it as an example. You can call it jumping to conclusions all you want, but blindly assuming Mantle will overcome the obstacles that shot down Glide is where the conclusions are being monumentally leapt to.

Look, as I've said before I am currently an AMD GPU owner. Of COURSE I want it to succeed. I really get sick of seeing DX11 games suffer from lack of optimization. It's why a LOT of devs don't even bother writing in such features in their games. I'm a realist though, I know a lot of Nvidia customers out there are not likely to accept this, it's obvious from the chat that exists already.

If you wanted to argue your points better you'd have examples to showcase WHY you feel we are jumping to conclusions, ones that show how AMD and Nvidia have shared tech successfully. No one can though because in reality the only tech they actually share is that which is neutral, such as the horribly inefficient Microsoft written API they already use.

 

[RoutedScripter]

Look, I scour the web and all the mantle threads out there, almost everyone who compares Mantle with Glide in a negative way is wrong in most of their opinions on other things around mantle.

What examples to showcase, the first post of this thread has it all, I'm just talking what I pulled out of that stuff man, I don't have to make stuff to show you to know, I could make it and you'd understand, but I don't have time to do that anymore.

Because, all you guys on the forums out there, think that it's about YOUR opinion that these threads want to influence, I didn't post this thread to try to convince you as a primary goal, no, whatever your opinion is has no influence on reality, and is totally irrelevant, everyone in this thread can be negative, you won't change anything, but if you think you'll feel better after doing that, go ahead.

I am interested to explain to a certain point, but this is ridicolous, it just means, you're opinion is set, and that is a sign that I don't have to waste more time with you.

I can see some people having posts like "When I see the hard numbers, i'll care" .... well, who cares about that guy, he is a mainstream gamer, he is not interested in the topic as a tech thing to research it on his own,

Who do you think the people in forums mostly are, avreage gamers, they don't have a wide attention span, they are consumers, they consume what is feed into them, and if you don't bring it to their attention, I have done my research on Mantle from a personal interest, without any gaming interest for the moment, I will not play BF4, or Thief or whatever, I don't even have a mantle-GPU, and I definitely won't get a new GPU just to play a bugfest in terms of gamplay and balance.


---------------------------

Going back to those who bring up developer interest and the glide comparison, they don't know much about, so they psychologically default to negative, because they speculate on the "developer interest" and then their mind makes up excuses to fit that belief, and from a small speculation you end up 9 pages bringing up new arguments which are actually the stuff that fits your view

So I've realized that mostly it's all from the root of "developer interest"

That's why I'll link the most important stuff below, one article mentioned it a few days ago, but the real deal stuff is years old, I've already posted this on this thread, guess I just need to do it again.


http://www.bit-tech.net/hardware/graphics/2011/03/16/farewell-to-directx/

http://www.polygon.com/2013/12/13/5170612/amd-mantle-interview

 

[Frag_Maniac]

What examples to showcase, the first post of this thread has it all, I'm just talking what I pulled out of that stuff man, I don't have to make stuff to show you to know, I could make it and you'd understand, but I don't have time to do that anymore.

Your first post has nothing but AMD claims on performance, and that's all good if that were all that mattered, but it's not, and it's what you don't understand.

So what you "pulled out of that stuff" is nothing new, it's what most already know and won't mean beans unless the majority of devs in the industry don't see it as a problem with their Nvidia customers.

It's not just about performance on AMD spec, in fact if it's too good, it may be even harder for Mantle to succeed because that very well may mean an even bigger differential between performance of Nvidia and AMD hardware for gamers, and those games could very well sell poorly with Nvidia customers because of it.

You talk like you don't have the time and have more than once said you're not going to participate here anymore, but all you do is repeat the same copy and paste info that anyone can obtain, it's common knowledge, not research.

What you SHOULD be researching is what made Glide fail, instead of insisting it's a non issue, because you still don't seem to get that this is more about overall industry reception than performance gains on just AMD's side.

 

[natr0n]

Is mantle getting released this month still?

I haven't read in here much.

 

[RoutedScripter]

Your first post has nothing but AMD claims on performance, and that's all good if that were all that mattered, but it's not, and it's what you don't understand.

So what you "pulled out of that stuff" is nothing new, it's what most already know and won't mean beans unless the majority of devs in the industry don't see it as a problem with their Nvidia customers.

It's not just about performance on AMD spec, in fact if it's too good, it may be even harder for Mantle to succeed because that very well may mean an even bigger differential between performance of Nvidia and AMD hardware for gamers, and those games could very well sell poorly with Nvidia customers because of it.

That might mean bad for nvidia, not mantle, you're twisting things. Also, I don't care about sideffects it will cause to others, it's coming, it will be great, it will uplift the industry into the correct path.

--------------------

Is mantle getting released this month still?

I haven't read in here much.

SDK is released early 2014, the only thing released this year is BF4 update, expect on xmas, probably not later, they're going home to holidays.

 

[xenocide]

You and xenocide definitely didn't spend 5 hours watching all the presentations, slides, and even all the other research and knowledge about how much of a difference low-level APIs make.

A good experienced man can predict things, you are inexperienced, you need hard proof to be able to see that. Just like the obamacare supporters, they had to see it written in their mail that they would have to pay 3 times as much, but we who were telling this would happen for months prior, because we done the research, they didn't take it seriously..

We didn't need to spend 5 hours watching AMD's PR team direct executives through talking points about a new product of theirs to be skeptical. You know why? History. I've seen this kind of song and dance before, and it didn't work out so great then. AMD has always overstated the benefits of their products; HD3xxx series, Bulldozer, Phenom I, etc. We know how much difference low level API's can make, hell they are currently around, the problem is they don't succeed unless they are truly 100% open, and nobody has actually done that yet. PhysX is a low level API, and it works great on Nvidia hardware, but terrible on AMD's cards. It has pretty much floundered since Nvidia bought up Aegia because of this fact. Lets look at past examples, Glide was great for 3dfx customers, but terrible for Nvidia, AMD, S3, and Matrox customers.

The basic fact is to have a low level API you need to be able to talk directly to the hardware, in the case of Mantle, apparently GCN. If you cannot speak directly to the hardware you need to add abstraction layers to translate, this causes noticable performance loss. Either AMD\Khronos will have to update Mantle to support Nvidia's architecture, or Nvidia will have to change their architecture to match GCN, and I guarantee that will never happen. For Mantle to succeed it needs to be supportable by Intel and Nvidia, and I just don't see that being possible without those companies paying a fortune.

Even though it is not relevant I'll touch down on Obamacare--simply put, you're wrong. Any increase in cost is actually saving people from financial ruin. In previous years you could have health insurance plans that gave nearly 0 benefits and cost next to nothing, and because most people aren't diagnosed with serious ailments it gave the illusion of an effective plan. Sure, at the lowest level (Bronze Plans) you're paying more, but your plan is orders of magnitude better, Insurance companies can no long drop you the second you get sick, and pre-existing conditions don't force you to pay 5x as much if you can even get insurance. Cost will come down over time as healthy people start essentially fronting the cost for those who have serious health problems.

You can keep saying whatever you want, but it's only your perception, which obviously doesn't affect reality.

Don't worry, you're distorting enough reality for everyone here.

If Nvidia had to do that, makes no point using the new API.

When I read this I was so certain we had broken through, it was like a TPU rendition of The Miracle Worker, but then you just kind of shrugged it off and it was like Requiem For A Dream...

It doesn't use GCN to speak with the game, GCN is an architecture, it uses Mantle to speak with the game. Mantle supports GCN currently and It may later support more GPUs because it's open standard and Nvidia can put their code straight beside AMD's

The point of low-level API's is that they exist within the engine and communicate with hardware. They aren't some kind of intermediary software that works wonders and just improves performance. It's basically like this--The engine is programmed with the API within it, the API directs the Engine in how to directly communicate with the hardware, if the API does not recognize the hardware it has to translate it so the hardware can understand it which takes time and causes a loss in performance. Engine->Mantle->Hardware. If the API does not adequately communicate with the hardware the whole purpose of the API is lost.

Nvidia can't just plop a bunch of code in there either, and it's not easy to create a low-level API, it takes years of development. There's also the fact that unless I'm missing something, adding tons of code to add support for various GPU architectures would bog down the API, just like it has with DirectX in the past.

 

[Frag_Maniac]

That might mean bad for nvidia, not mantle, you're twisting things. Also, I don't care about sideffects it will cause to others, it's coming, it will be great, it will uplift the industry into the correct path.

You're still not getting it. No wonder you dismiss Glide's failure, because you still don't seem to get WHY it did. This is not just a matter of AMD customers selfishly enjoying better performance without giving a damn about what Nvidia customers think. Developers have to write for ALL those customers. If you, God forbid, were a developer, do you really think you could afford to not care what half your customers think? Try that attitude in ANY business and see how long it lasts.

I mean jeez put your thinking cap on if you have one. Anything "bad for Nvidia" does not stay bad for long because they have the money and the customer base to easily compete, whereas AMD have very little of the former to throw around. They simply can't easily afford to spend lots of time and money on projects only to have to extensively rework them or have them fail.

You may not realize it, but in a way Nvidia can potentially be in the driver's seat on this, because they can afford to sit back and let AMD be the guinea pig testing this API, and if it has success they'll probably just come up with their own, and they'll have WAY more money to spend on talented developers for it.

In reality you're only seeing AMD and DICE's slant on this, and that is but a tip of the iceberg. It's far from proven yet, because proof of concept with this type of tech means acceptance just as much as technical achievement, esp if that tech is one sided. Furthermore, being able to compete with any such future endeavors by the completion.

THAT'S what will define success with this project, not merely what AMD customers think. There's too many unknowns to be calling this an easy win for AMD. That is an extremely naive and overly optimistic viewpoint.

 

[RoutedScripter]

Update:

"Secret" OXIDE Demo now Available!!!

 

[xenocide]

And so it continues...

I'll check it out when I get home from work.

 

[RoutedScripter]

Another, but just a bit older story:

http://gamingbolt.com/nitrous-engin...-gen-engines-utilize-eight-core-architectures

 

[RejZoR]

@Frag Maniac
Low level and open generally don't go together...

 

[xenocide]

@Frag Maniac
Low level and open generally don't go together...

That's because it requires high amounts of access to the hardware to implement, and different companies aren't exactly keen to share trade secrets...

 

[RoutedScripter]

Another Update: but much older from 2nd december, I didn't notice:

Johan interview with TomsHardware (Mantle API on 7 and 8 page)

http://www.tomshardware.com/reviews/johan-andersson-battlefield-4-interview,3688-7.html

 

[freeleacher]

Have Battlefield 4′s Woes Delayed AMD’s Mantle Roll-Out?

http://www.hardwarecanucks.com/news/video/have-battlefield-4s-woes-delayed-amds-mantle-roll-out/

Typical delay...

 

[RoutedScripter]

Johan tells on twitter they have their own site for all things frostbite engine, I don't think it takes all of them to fix bugs that are mostly game logic stuff , and core programmers usually never fix bugs unless it's it's a deep engine bug, that's all handled by support and balance teams.



Well ... here we go again https://twitter.com/repi/status/412728565101113344

More awesomeness every day ...

 

[freeleacher]

awsome

 

[The Von Matrices]

Have Battlefield 4′s Woes Delayed AMD’s Mantle Roll-Out?

http://www.hardwarecanucks.com/news/video/have-battlefield-4s-woes-delayed-amds-mantle-roll-out/

Typical delay...

I never expected them to make the deadline. Even disregarding previous technical issues, releasing a new product just before your development team goes on holiday vacation for one to two weeks seemed like a ridiculous proposition. You need your team to be ready to fix launch day issues.