• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Halts Optimizations for Mantle API

Didn't AMD also say that they didn't know if Mantle would offer any benefits to GPUs with architectures other than GCN? Which suggests to me it was written for GCN.

As for Mantle pushing Microsoft to make DX12, could you not use the same logic to deduce that Nvidia had a programmable GPU architecture before AMD thus pushing AMD to come up with GCN? Without which AMD wouldn't have come up with Mantle.
So Nvidia are indirectly responsible for Mantle and by extension Vulkan and DX12!


Mantle was written for GCN, their own marketing slides show that quite clearly:
http://images.anandtech.com/doci/7889/MantleBenefits.jpg
 
Mantle was written for GCN, their own marketing slides show that quite clearly:
http://images.anandtech.com/doci/7889/MantleBenefits.jpg

Mantle was designed with Gcn architecture in mind but is not tied to it.

https://twitter.com/amdgaming/status/400742321957773312

The trouble is multi-vendor support. Oddly enough Frostbite head honcho Johan Andersson said Mantle is actually not tied to AMD’s GCN architecture and it’s forward compatible. This obviously means AMD’s post-GCN GPUs will support it, but it also means Nvidia could embrace it as well, as DICE claims “most Mantle functionality can be supported on today’s modern GPUs” – unless DICE thinks Nvidia doesn’t make modern GPUs, this more or less means Nvidia GPUs could support Mantle sometime in the future.
 
I can find many openGL names that are equivalent, does that mean Mantle is just a copy of OpenGl? I'm sure someone who know DX can find similar.

It is a graphics APi, of course many commands are going to be the same.
I can find commands with the same name in Javascript as I can in C, Qbasic, python and R. does that mean they are all the same?

Yup they are all pretty generic function names for similar features - can't get much more generic than get object info, queue submit, etc. heh.
 
GPU's were already heading towards general compute architectures. Just nvidia got there 2 years before AMD when fermi dropped. :P

which is why fermi can support DX 12 driver overhead reductions.

So when AMD do something first (well, they didn't do low level APIs first did they, but maybe on the PC) then it forces everyone else to keep up.
When someone else does something before AMD, then that's just the way things were going?
Yeah, that seems like a totally unbiased viewpoint...

Mantle was designed with Gcn architecture in mind but is not tied to it.

https://twitter.com/amdgaming/status/400742321957773312

So DICE didn't think AMD's 6000 series cards were modern GPUs but he thought Nvidia's 400 series were? Ooooo burn AMD!
 
So when AMD do something first (well, they didn't do low level APIs first did they, but maybe on the PC) then it forces everyone else to keep up.
When someone else does something before AMD, then that's just the way things were going?
Yeah, that seems like a totally unbiased viewpoint...



So DICE didn't think AMD's 6000 series cards were modern GPUs but he thought Nvidia's 400 series were? Ooooo burn AMD!

Nvidia's Fermi is a more modern compute focused architecture compared to the 6000 series.
 
So when AMD do something first (well, they didn't do low level APIs first did they, but maybe on the PC) then it forces everyone else to keep up.
When someone else does something before AMD, then that's just the way things were going?
Yeah, that seems like a totally unbiased viewpoint...

So DICE didn't think AMD's 6000 series cards were modern GPUs but he thought Nvidia's 400 series were? Ooooo burn AMD!

And everyone had their own Low abstraction APi's for their hardware in the early 90's :P

Lol. Gpu's WERE heading in the direction of becoming general computer. just a few features were still fixed function on the 6000 series which prevented them from being fully general compute. And thus being capable of DX12/MAntle etc in the way they wanted the API's to work.

Lol, not my fault that Nvidia never bothered to capitalise on theie general compute architecture. You should be asking them why they didn't release their own API sooner than later. Considering they like to make their own proprietary software.
 
Even if Microsoft don't give a sausage about gaming they still had plenty of incentive to include DX12 in Windows 10. Microsofts biggest problem of the last few years has been trying to find ways to get people to upgrade their old versions of Windows, DX12 is a good incentive for people to make the upgrade.

DX12 wasn't a way to move ppl to Win10 because they were giving it away anyways. It was more a reaction to Mantle, because Mantle was more or less OS independent. What happens to MS when their one grapple hold on gaming is not so strong because Mantle gives ppl the option of running whatever OS they want, whether it be linux or windows 7. DX12 is not any incentive except for gamers but that only came after Mantle's breakthrough. Also remember back before Mantle's release, there was hardly any push or movement in any direction. Talk about stagnant. And now it's all anyone can talk about.
 
DX12 wasn't a way to move ppl to Win10 because they were giving it away anyways. It was more a reaction to Mantle, because Mantle was more or less OS independent. What happens to MS when their one grapple hold on gaming is not so strong because Mantle gives ppl the option of running whatever OS they want, whether it be linux or windows 7. DX12 is not any incentive except for gamers but that only came after Mantle's breakthrough. Also remember back before Mantle's release, there was hardly any push or movement in any direction. Talk about stagnant. And now it's all anyone can talk about.

So why do older cards that pre date Mantle have support for DX12 ?

Because DX12 has always been the upgrade path.
 
And everyone had their own Low abstraction APi's for their hardware in the early 90's :P

Lol. Gpu's WERE heading in the direction of becoming general computer. just a few features were still fixed function on the 6000 series which prevented them from being fully general compute. And thus being capable of DX12/MAntle etc in the way they wanted the API's to work.

Lol, not my fault that Nvidia never bothered to capitalise on theie general compute architecture. You should be asking them why they didn't release their own API sooner than later. Considering they like to make their own proprietary software.

An Graphics APIs were evolving too.
I think my logic is as sound as the "we wouldn't have DX12 without Mantle" logic.

Also, yeah, Microsoft didn't come up with it for 8 years, but neither did AMD!
AMD could've done it 5 years ago just as easily as Microsoft.
 
So why do older cards that pre date Mantle have support for DX12 ?

Because DX12 has always been the upgrade path.

GCN was first released in 2011, it is a more general purpose architecture that can do both compute and graphics well and was AMD's answer to Nvidia's Fermi which was released first and already followed these principles: http://www.anandtech.com/show/4455/amds-graphics-core-next-preview-amd-architects-for-compute/1
and:
https://community.amd.com/thread/171008. Newer GPUs are an evolution of these basic architectural principles used in 2011.

These lower level APIs are written for this type of architecture (Nvidia and AMD's are more similar than you'd think) which is why VLIW4/5 (pre radeon 7000)which is vastly different, is not supported. I'm not an expert on this so people are welcome to correct or expand, but this is what I've understood from various talks and articles on Mantle/Vulkan/DX12.
 
Last edited:
So the cards were developed for DX12 rather than DX12 being developed to support majority of cards. Not sure which is more likely lol.

Neither. DX has always been co-developed between MS and manufacturers. Stuff gets added to DX to support hardware and stuff gets added to hardware to support DX.
 
I can't see it swinging a lot of the established developers around to it. If it did offer better support/documentation and a more intuitive learning curve than DX12 though it could pick up a lot of ground swell from "bedroom" programmers and smaller companies, etc.
 
I've read many arguments from both sides and I don't buy that DX12 is mantle and there would be no DX12 without mantle. There was always going to be a DX12 and they were always going to refine the process just like they did with DX11 over DX10. DX12 is also meant to bring some new eye candy as well so its more than just refinement process. A new DX is also a good way to sell a new OS and make gamers upgrade, it also helps sell gpus so generates income and keeps customers buying.

Personally I think its a shame AMD have binned off Mantle so early as its going to take many years for the pc gaming scene to jump over to DX12. AMD could have kept the branding of Mantle going as another feather to its bow even if it was only for a few games a year a bit like how Nvidia did with phsyx. It was a good selling point for them and unique to their product line. Dropping it was a bad move IMO.
 
A lot of us did say Mantle would die very very quickly, ridiculed we were.

Unfoortunately blind loyalty is a large cause for ignorance and delusion. I understand the loyalty to AMD, I really do as I used to use them heavily, but at some point you have to wake up and smell the roses and look at things objectively.

AMD are in trouble, they do not have the resources or stability to develop and support frivolous things like API's while their revenues are falling and their staff are continuously decreasing. They are being battered on two fronts, CPU and GPU, and money is weeing down the toilet.

Sadly, some people will deny this until the bitter end.
 
Back
Top Bottom