What Was the Point/Effect of Mantle?

[soccerballtux]

Um.. this isn't evidence of trade secrets leaking out or being copied by the competitors, this is evidence of AMD directly contributing to the efforts.

AMD fully understood the risks of trying to create a vendor-specific API, because that era of PC gaming was a mess, and going forward that would not have created any lasting benefit for AMD. On the contrary, it was developed to kick-start development of low-level APIs across the board, and it seems that both of the vendor-neutral APIs we will be receiving have direct lineage in Mantle.

AMD made that happen, it isn't an example of Microsoft or Nvidia trying to capitalize on AMD's mistakes in locking down the code.

I'm aware of all of that. I'm saying they shouldn't have contributed. I think you like, totally misread what i said

 

[RampantAndroid]

I'm aware of all of that. I'm saying they shouldn't have contributed. I think you like, totally misread what i said

Why? Look, AMD worked with MS to make the XB1 which has a mutated Direct3D11 on it, which basically already has a lot of benefits of DX12 in it (Phil has said publicly DX12 won't do much for the XB1's perf.) Now, I don't know how much code AMD saw, but MS had a "close to the metal" API in development super early that the DirectX people probably had a lot to do with. That it gets turned into DX12 makes sense. I don't know how much effect Mantle had on MS' schedule for releasing a new DirectX, but I don't think it's that much. You don't go from no code to demo in 6 months.

You're advocating that AMD not support MS' efforts to make a hardware abstracted layer on Windows and instead....what? Fragment the market? Instead of just targeting DirectX on Windows, you want developers to have to make games function on DirectX for Nvidia users and on Mantle/Vulkan for AMD users? Give me a break.

Vulkan will matter on mobile (err, Android. It won't matter on Windows Phone or iOS) and a little on Linux desktop users. It won't matter too much on Windows I don't think...and I have no idea if the PS4 will use it. The XB1 will never use it.

 

[destrekor]

I'm aware of all of that. I'm saying they shouldn't have contributed. I think you like, totally misread what i said

I didn't misread at all. This post confirms exactly what I understood from your previous post.

Them not contributing would have been the greatest slight against the developer and gaming community in a long while. It would have been an absolutely terrible and tragic decision to withhold Mantle for their very own use, saying screw you to Microsoft, Nvidia, and the Khronos Group (OpenGL), among numerous other providers of graphics hardware.

Why? Look, AMD worked with MS to make the XB1 which has a mutated Direct3D11 on it, which basically already has a lot of benefits of DX12 in it (Phil has said publicly DX12 won't do much for the XB1's perf.) Now, I don't know how much code AMD saw, but MS had a "close to the metal" API in development super early that the DirectX people probably had a lot to do with. That it gets turned into DX12 makes sense. I don't know how much effect Mantle had on MS' schedule for releasing a new DirectX, but I don't think it's that much. You don't go from no code to demo in 6 months.

You're advocating that AMD not support MS' efforts to make a hardware abstracted layer on Windows and instead....what? Fragment the market? Instead of just targeting DirectX on Windows, you want developers to have to make games function on DirectX for Nvidia users and on Mantle/Vulkan for AMD users? Give me a break.

Vulkan will matter on mobile (err, Android. It won't matter on Windows Phone or iOS) and a little on Linux desktop users. It won't matter too much on Windows I don't think...and I have no idea if the PS4 will use it. The XB1 will never use it.

This. However, I think Vulkan will matter for all mobile, not just Android. I expect iOS and Windows 10 to fully support Vulkan, so it'll be up to the developer. If they develop only for iOS, then Metal would make sense, but Vulkan would be the ideal choice for multi-platform titles.

 

[Magic Carpet]

This "Mantle 2.0" is actually Vulkan, and yes, it will be a big hit on Linux, this seems all but guaranteed.

Valve is putting a big push behind SteamOS gaming, which is a Linux platform. And they are putting a big push behind Vulkan, announcing plans to completely take up development with Vulkan, with Source Engine 2 utilizing that API.

Linux gaming will be receiving a huge boost in developer support thanks to Vulkan and Valve/Steam.

I just hope, this will be finally getting somewhere. Not the first time

 

[JDG1980]

So now that Mantle is going away, what was the point?

I mean, I get that the idea was to bring AMD's console optimizations to the PC but even they couldn't have hoped that it would be used in more than a few games.

So why did they do it? And was it a success by that reason?

Back in 2013, AMD seems to have gotten word from Microsoft that there would never be a DirectX 12. This, needless to say, would have been bad for AMD's GPU business. Therefore, to give MS a kick in the butt, AMD decided to do their own low-level API, which later became an inspiration for DX12, and even more so to OpenGL (Vulkan is basically Mantle with the serial numbers filed off). It did succeed in that goal, but it didn't do much to spur sales of AMD cards, and it cost precious R&D dollars that AMD could have used elsewhere. Results are mixed at best.

 

[destrekor]

Back in 2013, AMD seems to have gotten word from Microsoft that there would never be a DirectX 12. This, needless to say, would have been bad for AMD's GPU business. Therefore, to give MS a kick in the butt, AMD decided to do their own low-level API, which later became an inspiration for DX12, and even more so to OpenGL (Vulkan is basically Mantle with the serial numbers filed off). It did succeed in that goal, but it didn't do much to spur sales of AMD cards, and it cost precious R&D dollars that AMD could have used elsewhere. Results are mixed at best.

Ha, this started way before 2013. Way before. DX 12 alone started far earlier than 2013.

 

[RampantAndroid]

Back in 2013, AMD seems to have gotten word from Microsoft that there would never be a DirectX 12. This, needless to say, would have been bad for AMD's GPU business. Therefore, to give MS a kick in the butt, AMD decided to do their own low-level API, which later became an inspiration for DX12, and even more so to OpenGL (Vulkan is basically Mantle with the serial numbers filed off). It did succeed in that goal, but it didn't do much to spur sales of AMD cards, and it cost precious R&D dollars that AMD could have used elsewhere. Results are mixed at best.

Did you even read this thread? Why do people keep parroting this? DX9 was released in 2002. DX10 was 4 years later in 2006. DX11 was 2009. DX12 will be...2015? That's 4 years, 3 years and 5 years. No real trend at all - but 5 years certainly isn't out the ordinary given how much of a change we get with 12 (and they may have been busy doing custom work on the XB1 in there? The XB360 for example shipped 9.0C I believe. I would posit that DX12 development is directly tied to the custom DX version on the XB1 so development *at least* began at that point in time.)

Edit: Yup, the 360 shipped with 9.0c - so there's an instance of the XBox and DirectX being tied together in some fashion.)

 

[sandorski]

Part Marketing, part getting an advantage over Nvidia, and part Proof of Concept. They certainly created a lot of buzz and proved the value of the concept. It will be interesting to see what Mantle 2.0 brings and how AMD's offerings look on Linux in the future.

 

[RampantAndroid]

Part Marketing, part getting an advantage over Nvidia, and part Proof of Concept. They certainly created a lot of buzz and proved the value of the concept. It will be interesting to see what Mantle 2.0 brings and how AMD's offerings look on Linux in the future.

I think those are very different questions: "Mantle 2.0" is Vulkan, right? So not in AMD's control (and it'll be hardware abstracted.) If it had been Mantle 2.0, still locked to a subset of all AMD GPUs I'd still be saying it'd fail.

Linux is more an issue with Drivers however, not with Mantle or Vulkan?

 

[geoxile]

The point is that Johan wanted an API that could let DICE do on PC what they do on consoles. So he came up with the idea, did initial design, and then went to hardware vendors to implement it.

http://www.heise.de/newsticker/meld...ber-AMDs-3D-Schnittstelle-Mantle-2045398.html

AMD agreed and helped him implement it.

I can't make it clear enough that Mantle was made for developers BY a developer, bringing it to more vendors was always part of the plan.

 

[youshotwhointhe]

It is clear DX12 and Vulkan are largely based on Mantle, but that isn't really the important point. People somehow fail to realize that the console APIs and these new PC APIs are all similar because they are all low level APIs targeting similar hardware. And the reason Nvidia has no problem supporting these new APIs is because their hardware really isn't that different either. Any differences between modern GPU designs are lower level than any graphics/compute API needs to be.

So what was the point for AMD? To kick legacy DX and, more importantly, OpenGL to the curb. These are APIs designed in the 90's to target the hardware available at that time. Hardware that in no way resembles a modern GPU. Since then these APIs have accumulated new features, but the old ones stick around. It has become a nightmare to create and maintain a driver that fully supports these legacy APIs.

Nvidia has the best drivers in the industry, which is why they are less than thrilled by the move away from OpenGl. Another issue for them is that in order to fully support Vulkan they will need to fully support the latest OpenCl (they both use SPIR-V) which they have avoided in the past in order to prop up CUDA. AMD has the second best drivers, but Intel and the bigger mobile companies are catching up fast. Intel has way more money to throw at driver developers.

Everyone but Intel is happy about the reduced CPU requirements, but I doubt Intel cares too much since they are also switching to low power SoC designs.

Any GPU sales from improved performance in PC games that used Mantle was just icing on the cake.

 

[Noctifer616]

Linux is more an issue with Drivers however, not with Mantle or Vulkan?

And Vulkan fixes that issue because the driver is very thin. You make one good driver and that's it. Most of what makes up the driver today, the optimizations for specific applications, is gone.

 

[Spanners]

The point was to sell AMD cards. The effect entirely different.

This doesn't really explain anything useful though does it? If someone asked what was the point of G-Sync? Then selling more video cards doesn't really explains it's purpose anymore than it would for any product/feature from any for-profit company ever.

The effect being "entirely different" would imply Mantle had caused AMD to sell less cards which I'm sure you don't have anything to substantiate.

 

[3DVagabond]

Mantle is not going away. They are working on Mantle 2.0, and as I recall it will end up on Linux as well.

But Mantle was a very important stepping stone on the path to DX12. It gave developers a chance to use an API that is extremely similar in implementation to DX12. It allowed them to find bugs and architectural issues long before DX12 was released. While DX12 is technically different from Mantle, to say that Mantle had zero influence on DX12 would be extremely short sighted.

Yes it also gave them some marketing hype for sure. And they certainly pushed that. But it also helped consumers. BF4 was unplayable for me on my old system because all four cores on the CPU would be pegged out. With mantle, I got literally double the frame rates compared to DX11.

I'd say Mantle had way more influence than many want to admit. If you look at this what it looks like is MSFT changed enough words to not infringe on copyright, but other than that it's a direct copy.

 

[RampantAndroid]

I give up. Make valid points, people ignore them. OK, fine. Enjoy your little bubble.

 

[3DVagabond]

Ha, this started way before 2013. Way before. DX 12 alone started far earlier than 2013.

Is there a source for this? Other than nVidia saying they had a conversation with MSFT at sometime before that.

 

[NTMBK]

Thinner driver layer = less driver optimization work required = cheaper driver team for AMD, and less chance to screw up compared to NVidia.

 

[SirPauly]

Is there a source for this? Other than nVidia saying they had a conversation with MSFT at sometime before that.

Microsoft did, when AMD offered, " there would be no DirectX 12," Microsoft responded, "We have absolutely no intention of stopping innovation with DirectX."

Obviously AMD is the one with a credibility problem. Microsoft never stopped innovating.

 

[3DVagabond]

Microsoft did, when AMD offered, " there would be no DirectX 12," Microsoft responded, "We have absolutely no intention of stopping innovation with DirectX."

Obviously AMD is the one with a credibility problem. Microsoft never stopped innovating.

That's not very definitive.

 

[SirPauly]

From your point-of-view, but it's AMD that was irresponsible and ridiculous to offer such a statement to promote Mantle.

 

[Spjut]

Or AMD wanted Microsoft to speed up the announcement of DX12:biggrin:

But yeah, it'd be interesting to know when they started writing the code instead of just talking

 

[3DVagabond]

Or AMD wanted Microsoft to speed up the announcement of DX12:biggrin:

But yeah, it'd be interesting to know when they started writing the code instead of just talking

I'm curious when they decided on DX12 not DX11.* and when they decided it was going to have low level access. The quoted statement just says they planned on continuing progressing DX. Not anything definitive.

 

[Spjut]

I'm curious too. We do know though that DX12 requires a new driver model, WDDM 2.0, which must have been in development for quite some time.

 

[Noctifer616]

I'm curious too. We do know though that DX12 requires a new driver model, WDDM 2.0, which must have been in development for quite some time.

What difference does WDDM 2.0 actually make? Mantle works on Win7 and seams to be delivering the same results as DX12, Vulkan is going to work on a massive range of platforms (android, Linux, Windows etc.) so it doesn't require the new driver model.

 

[JEinstein]

It was always my understanding that the point of Mantle was for the OS to have better and more direct communication with the Graphics hardware.