The AMD Mantle Thread

[blackened23]

Whoa there buddy. You're the one that said DICE had financial obligations to deliver Mantle for BF4. I just happen to agree with you on that point. Not sure why you're so upset. Calm down. Personally I don't give a crap what AMD paid DICE for mantle. If DICE got an 8 million $ payday good for them. If nvidia pays random developer 8 million bucks - hey - why would I care. I don't care.

As far as moving the discussion forward, i'm all for that - you seem stuck on this point for some reason. I was just pointing out that you told us all that AMD paid DICE for BF4 mantle as an "obligation" last month. I don't disagree with your statement.

So can we move on with the thread now instead of being stuck on this? Don't you want to talk about Mantle instead? Let's just move on buddy. As far as i'm concerned we're done talking about this, I won't respond about AMD's payments to DICE henceforth. DaZeeMan's point about Mantle and asymetric crossfire seemed interesting. On the Kaveri. Maybe we'll find out more on that tomorrow. Or maybe Mantle will be released tomorrow? Any ideas? Thoughts? Mantle coming tomorrow?

 

[VulgarDisplay]

Dice is obligated to release mantle because it as much their baby as it is amd's.

 

[3DVagabond]

Whoa there buddy. You're the one that said DICE had financial obligations to deliver Mantle for BF4. I just happen to agree with you on that point. Not sure why you're so upset. Calm down. Personally I don't give a crap what AMD paid DICE for mantle. If DICE got an 8 million $ payday good for them. If nvidia pays random developer 8 million bucks - hey - why would I care. I don't care.

As far as moving the discussion forward, i'm all for that - you seem stuck on this point for some reason. I was just pointing out that you told us all that AMD paid DICE for BF4 mantle as an "obligation" last month. I don't disagree with your statement.

So can we move on with the thread now instead of being stuck on this? Don't you want to talk about Mantle instead? Let's just move on buddy. As far as i'm concerned we're done talking about this, I won't respond about AMD's payments to DICE henceforth. DaZeeMan's point about Mantle and asymetric crossfire seemed interesting. On the Kaveri. Maybe we'll find out more on that tomorrow. Or maybe Mantle will be released tomorrow? Any ideas? Thoughts? Mantle coming tomorrow?

Once again you say whatever you want to and then tell others to stop. More hypocrisy.


If you and Blackened23 continue with this nonsense you BOTH will be kicked from this thread.

-Rvenger

 

[TeknoBug]

I don't see any evidence there of AMD paying DICE 8million or any other developer to use Mantle.

So I will continue with my baseless claims that Sweeny and Carmack are only hostile towards mantle cause they are in nVidia's pocket.

Isn't it the other way around? DICE's Johan Andersson approached AMD about the Mantle project. He also approached Nvidia, but they turned it down.

 

[0___________0]

Most games are ported from console API's.

I don't think this can even be called a half-truth. "Porting" according to the original meaning basically never happens anymore. Modern engines build for every platform without any additional work from the dev. Obviously you use lower resolution textures, poly count, tessellation, etc. on the console version, but that's irrelevant to this. You construct your game once, the engine is already designed to build for every API. Grab a copy of the CE, UE, or Unity, you don't need to port from your target platform (EDIT: I mean target as in the platform you anticipate being the largest audience).

I suppose you could argue that AMD simplifies engine creation with Mantle, and you'd probably be right, but that would only be relevant if it completely replaced DX. As it stand right now, you now have to maintain another API.

When people say "crappy console port" what they should be saying is a studio's engine devs only gave a half baked effort at PC DX support. Nowadays, everyone is using "porting" to refer to when a console is the primary development platform, but it's not really accurate because nothing was ever ported. Today it is about as useful as "coding to the metal".

Also, most bugs are difficult to fix because it requires the Devs and the hardware vendors to adapt the game and the hardware to DX, and according to the devs DX is a black box they can't see inside of.

I'd careful with your usage of "most". I think that most bugs in release products are results of the game programming and not issues with the engine. I also don't think that console versions are any less plagued by bugs, even though they aren't bound by DX. BF4 was probably even more glitchy on consoles, it certainly hasn't been better than on PC. The console launch is most likely the biggest factor for the lawsuit against EA too.

Bug are always going to exist, I don't think that AMD, DICE, or anyone has claimed there's going to be a lot less bugs run using Mantle. The primary problem with bugs is devs not doing things the right way and releasing a product that needs more work done. Every game is going to have problems, but the ones that cause rioting on the internet on launch are consequences of doing things in a hurry or just not caring, or ineptness.

I think people are still too caught up with the notion of the magical console API ported to PC.

I'm not bashing mantle, I like it, but I find myself avoiding this thread because it seems to be just a mix of people who don't know what they're talking about but try to convince others that they do, people who have made their mind up before even seeing anything, and others who take this way too seriously. Obviously there's people in here genuinely contributing, but I'm just making a point. People need to give it a rest until some tangibles start coming out. We're down to arguing over contractual obligations and companies paying off studios, c'mon guys, this is sad.

 

[MutantGith]

...

I think people are still too caught up with the notion of the magical console API ported to PC.

I'm not bashing mantle, I like it, but I find myself avoiding this thread because it seems to be just a mix of people who don't know what they're talking about but try to convince others that they do, people who have made their mind up before even seeing anything, and others who take this way too seriously. Obviously there's people in here genuinely contributing, but I'm just making a point. People need to give it a rest until some tangibles start coming out. We're down to arguing over contractual obligations and companies paying off studios, c'mon guys, this is sad.

(emphasis added)

Indeed. I couldn't agree more.

Every once in a while, when reading this thread, it seems like someone tries to make a point, ask an honest question, or bring up some thoughts about what this hubbub might mean in regard to PC gaming as a whole going forward. Those comments are almost universally drowned out in short order by a wash of what boils down to the same posts being rehashed again and again. While there is a potential for critical analysis and investigation, I'm finding sorting through the number of posts that are full of contradictory information and cheer leading (either way) extremely tiring.

It's flat out amazing that the hype machine is so wound up (again, on both sides of the equation), given the short amount of time the whole Mantle announcement has even realistically been made. I think it might be a great idea for everyone to take a deep breath and realize that almost everything in this thread is mostly speculation based on marketing slides and vague statements. Everyone (again, on both sides of the equation) should think about the front page news item on this very web site before they start making bold proclamations as fact. You know, the one where one of the technical writers for this site made a statement that Mantle was almost surely exactly the API for the XBOX One ported to windows... only to have both AMD and Microsoft refute to varying degrees of strength some days later. That, from someone who is a professional industry correspondent with access most everyone on this board lacks.

I, and I think IMHO that other people who were looking to read the thread for information might appreciate it if the lack of information actually released at this point gave people pause and instilled caution, rather than being seen as a license to engage in wild speculation.

 

[Barbatruuk]

I really hope the Mantle patch will be released very soon. It's very clear people need some hard results, because this thread is derailing every single time it's been reopened (lack of facts and numbers).

I must admit, things were looking decent since this thread reopenend last time, but starting to look like a mud fight again haha.

The faster Mantle gets released, the sooner we can discuss actual numbers.

 

[DamnedLife]

http://www.techpowerup.com/mobile/1...ormance-improvement-in-battlefield-4-amd.html mind you they got their hands on some newly released apus and considering amd demoed mantle at their ces booth (not 30 seconds demo but a laptop with eyefinity setup running bf4 on high @30fps average) they may also got the same mantle demo to test it out themselves with their own setup before anyone else. 280x=~gfx780ti even if its partially true i am sold.

 

[Barbatruuk]

From the same article:

the only known game with planned support for Mantle

I thought there are a lot more games planned with support for Mantle, including Thief, which is to be released in february if all goes well.

I still think it's impressive running a 3 monitor setup on a laptop at those settings and getting 30 fps.

 

[96Firebird]

http://www.techpowerup.com/mobile/1...ormance-improvement-in-battlefield-4-amd.html mind you they got their hands on some newly released apus and considering amd demoed mantle at their ces booth (not 30 seconds demo but a laptop with eyefinity setup running bf4 on high @30fps average) they may also got the same mantle demo to test it out themselves with their own setup before anyone else. 280x=~gfx780ti even if its partially true i am sold.

That article (if you even call it that) just rehashes what has been shown on AMD's slides. The whole 280x = 780Ti is assuming the 45% increase from AMD's slide. The A10-7850K APU providing acceptable FPS at 1080p assumes 45% increase from AMD's slide. The reason they use the word could so often.

BTW - I'm not one to nit pick about these things, but that first sentence is such a bad run-on sentence, I'm shocked the writer didn't catch that.

 

[zlatan]

From the same article:
I thought there are a lot more games planned with support for Mantle, including Thief, which is to be released in february if all goes well.

More interestingly, somebody has some extra information.

http://www.techpowerup.com/forums/t...ent-in-battlefield-4-amd.196842/#post-3048853

While I don't know anything about Galactic Civilization III with Mantle support, but the other titles are mostly the same as I heard.

 

[Pottuvoi]

From the same article:
I thought there are a lot more games planned with support for Mantle, including Thief, which is to be released in february if all goes well.

All games currently using Frostbite should get support.

 

[Barbatruuk]

More interestingly, somebody has some extra information.

http://www.techpowerup.com/forums/t...ent-in-battlefield-4-amd.196842/#post-3048853

While I don't know anything about Galactic Civilization III with Mantle support, but the other titles are mostly the same as I heard.

Exactly, those are the titles I've read about, which are supposed to have Mantle support as well.
And if Mantle delivers, it wouldn't be surprising more titles will be announced in the future.

 

[tonyfreak215]

It's the dual graphics support in the new Catalyst patch that has me intrigued. The asymmetric APU thing (iGPU + dGPU) was mentioned in one of the Mantle slides during APU 13 as I remember.

Asymetric crossfire will likely be bunk, because uneven performing GPUs results in microstutter. Prior APUs with asymmetric crossfire had microstutter so horrible that it was worse than simply using the APU alone - There is no way around this.

In the slides they mention that developers would have control over how the 2nd gpu would be used. It could be used for rendering or it could be used for compute. We'll just have to wait and see.

As for crossfire micro stuttering; they have a hardware fix. The 290's have improved on it immensely. I'm sure those fixes are also implemented in their new APUs.

More interestingly, somebody has some extra information.

http://www.techpowerup.com/forums/t...ent-in-battlefield-4-amd.196842/#post-3048853

While I don't know anything about Galactic Civilization III with Mantle support, but the other titles are mostly the same as I heard.

That does make sense as Oxide is a spin off of Stardock.

 

[ultimatebob]

The Mantle hype machine continues... it looks like AMD is now promising "Up To" 3X performance on their next generation of processors:

Mind you that 8 fps is basically unplayable for basically any game in existance, but whatever.

Man... I hope this API works at least half as well as promised, or nobody will ever take AMD seriously again.

 

[BrightCandle]

In the slides they mention that developers would have control over how the 2nd gpu would be used. It could be used for rendering or it could be used for compute. We'll just have to wait and see.

Of all the things mantle offers this is the one I feel has the most potential. Right now if you go crossfire and you target the same FPS as a single card then you double the latency from the GPUs, because each is now taking twice as long to render because of the more complex scenes. Being able to use both GPUs for the same frame is the holy grail of multiple card setups allowing higher quality images or lower latency ones when using multiple cards.

But it will be a lot more effort by the developers to do this instead of using basic AFR. I think in many ways what is really needed is an API with an abstraction that completely hides the 1 or 2 or N GPU distinction completely, allowing the GPUs and their drivers to choose good performing options. Its funny I think there are two ways to solve the current issue, go low level like AMD has or go even higher level so that there is no difference between AMD/Nvidia hardware and no difference between 1 or N GPUs.

 

[RampantAndroid]

Exactly, those are the titles I've read about, which are supposed to have Mantle support as well.
And if Mantle delivers, it wouldn't be surprising more titles will be announced in the future.

Coming into this thread late, but even if it delivers, isn't it going to come down to how many GCN cards are owned? If the majority of people have nVidia and non-GCN AMD cards, Mantle will be a hard sell since you need to support BOTH OpenGL/DirectX as well as Mantle?

Just sounds like needing to support D3D and Glide back in the day.

 

[smackababy]

Coming into this thread late, but even if it delivers, isn't it going to come down to how many GCN cards are owned? If the majority of people have nVidia and non-GCN AMD cards, Mantle will be a hard sell since you need to support BOTH OpenGL/DirectX as well as Mantle?

Just sounds like needing to support D3D and Glide back in the day.

It doesn't even have to be a majority of GCN cards; it has to be such an overwhelming majority they can drop support for DX, before it becomes a situation where they don't need two code paths.

No matter how trivial people want to believe it is to have a Mantle code path and a DX code path, you have to fully test both. Regardless of the development time, you're double testing time.

 

[TerryMathews]

you have to fully test both

Why start now? Games ship broken all the time.

Just saying, there's no way Mantle is going to damage the stellar QC that currently happens on PC.

 

[TerryMathews]

how many GCN cards are owned?

Youre forgetting GCN-based APUs

 

[Paul98]

People over estimate how hard it will be to port to mantle from other API's. It's going to be the hardest right now in alpha or beta with little documentation, having to learn how to use it and write functions from scratch. But it still uses the HLSL from directX, the math and much of that code will be reusable.

 

[ASM-coder]

And people over estimate how hard it is to code "closer to the metal"

 

[dacostafilipe]

The whole 280x = 780Ti is assuming the 45% increase from AMD's slide. The A10-7850K APU providing acceptable FPS at 1080p assumes 45% increase from AMD's slide.

If we believe hardware.fr, the 45% from the slides are not on Kaveri (as some people might think) but on R9 290/x hardware.

Source: http://www.hardware.fr/news/13525/ces-45-version-mantle-bf4.html

 

[ASM-coder]

Maybe, but I would think Kaveri is the perfect poster child for Mantle.
An underpowered CPU with a GCN GPU, with the possibility
of leveraging shared memory(hUMA)

 

[smackababy]

If we believe hardware.fr, the 45% from the slides are not on Kaveri (as some people might think) but on R9 290/x hardware.

Source: http://www.hardware.fr/news/13525/ces-45-version-mantle-bf4.html

But, even if it was with a 290, the scenario isn't even fleshed out enough to mean anything. They could be running a 290 and eyeFinity with 4k monitors, so their 45% FPS increase is a total of 4 frames per second better or something.

We need real numbers and real information. Not these marketing slides "best case scenario in our canned test made explicitly to make us look good, while providing zero real information". I am not saying just AMD does this, but they are doing it quite a bit for Mantle.

Once the API is actually out, we can get a real idea of just how involved it is and what it really takes to get performance improvement.