By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - PC Discussion - AMD unveils 'Mantle' API to optimize GPU performance - Debute with BF4

Machiavellian said:
fatslob-:O said:
Machiavellian said:
This really just sounds like a direct API to the AMD chip which would be no different if Sony did the same thing for their chips. Basically what this means is that developers have 4 APIs to choose from. This API, Direct X, OpenGl and whatever Nvidia bring to the table.

Will nvidia even bring their own api ?

I believe Nvidia already have their own APIs.  This is really sounding like we are going back to the 3dfx days where each card have their own APIs and extensions to openGL where developers had to decide which one to use and support.  DX was brought into scope to fix that problem.  Since Nvidia and AMD cards still will be sold in bunches, developers concentrating on GPU specific features that are not supported on both cards is the same as 3rd party developers doing the same thing with Sony and MS console.  

Cuda doesn't count just so you know because it is not meant to be a graphics language like directx or mantle so do you have anything else saying that nvidia has a graphics API ?



Around the Network
fatslob-:O said:
Machiavellian said:
fatslob-:O said:
Machiavellian said:
This really just sounds like a direct API to the AMD chip which would be no different if Sony did the same thing for their chips. Basically what this means is that developers have 4 APIs to choose from. This API, Direct X, OpenGl and whatever Nvidia bring to the table.

Will nvidia even bring their own api ?

I believe Nvidia already have their own APIs.  This is really sounding like we are going back to the 3dfx days where each card have their own APIs and extensions to openGL where developers had to decide which one to use and support.  DX was brought into scope to fix that problem.  Since Nvidia and AMD cards still will be sold in bunches, developers concentrating on GPU specific features that are not supported on both cards is the same as 3rd party developers doing the same thing with Sony and MS console.  

Cuda doesn't count just so you know because it is not meant to be a graphics language like directx or mantle so do you have anything else saying that nvidia has a graphics API ?

I did some checking and it appears Nvidia does not use a seperate API, instead they have a SDK for D3D.  In other words they show how to leverage Nvidia hardware using DX D3d but they do have specific API calls for their PhyX and other features.

When all is said and done, this is just a low level API to AMD hardware.  Just because AMD is making it open source does not make this an API that other vendors will use since its targeted at AMD GPU.  This has the pontential to fragment the industry again and I am not sure this is a good thing.  Its not like Nvidia is going away anytime soon and its not like Nvidia would not bring their own solution to the mix thus stirring the pot.



The fact that we're drifting away from M$ more and more brings me joy!!!!!



UnknownFact said:
The fact that we're drifting away from M$ more and more brings me joy!!!!!

You know people are saying on other websites that this tech comes from MS partnership with AMD.



My god this coming gen is going to be interesting. Mantle, Steambox, boxes streaming to each other, Mobile gaming (ew) and three new home consoles. This is going to get crazy!



Around the Network
Locknuts said:
My god this coming gen is going to be interesting. Mantle, Steambox, boxes streaming to each other, Mobile gaming (ew) and three new home consoles. This is going to get crazy!


That we can all agree on!



Machiavellian said:
fatslob-:O said:
Machiavellian said:
fatslob-:O said:
Machiavellian said:
This really just sounds like a direct API to the AMD chip which would be no different if Sony did the same thing for their chips. Basically what this means is that developers have 4 APIs to choose from. This API, Direct X, OpenGl and whatever Nvidia bring to the table.

Will nvidia even bring their own api ?

I believe Nvidia already have their own APIs.  This is really sounding like we are going back to the 3dfx days where each card have their own APIs and extensions to openGL where developers had to decide which one to use and support.  DX was brought into scope to fix that problem.  Since Nvidia and AMD cards still will be sold in bunches, developers concentrating on GPU specific features that are not supported on both cards is the same as 3rd party developers doing the same thing with Sony and MS console.  

Cuda doesn't count just so you know because it is not meant to be a graphics language like directx or mantle so do you have anything else saying that nvidia has a graphics API ?

I did some checking and it appears Nvidia does not use a seperate API, instead they have a SDK for D3D.  In other words they show how to leverage Nvidia hardware using DX D3d but they do have specific API calls for their PhyX and other features.

When all is said and done, this is just a low level API to AMD hardware.  Just because AMD is making it open source does not make this an API that other vendors will use since its targeted at AMD GPU.  This has the pontential to fragment the industry again and I am not sure this is a good thing.  Its not like Nvidia is going away anytime soon and its not like Nvidia would not bring their own solution to the mix thus stirring the pot.

They have what you would call libraries and physx is one of them. I don't think mantle is completely targetting gcn architechture otherwise they wouldn't claim it "open source". Oh we all know this won't kill nvidia imediately but once it starts gaining some crazy traction this is going to be one hell of a ride for nvidia. Even if nvidia does bring their own solution it will most likely be rejected because their solution does not favour AMD designed consoles. 



CGI-Quality said:
fatslob-:O said:
CGI-Quality said:
fatslob-:O said:

It works alright but how about the performance impacts ? Nvidia also has to worry about devs choosing light indexed deferred renderers instead of tiled deferred or regular deferred renderers. Just because all these effects can be ran on nvidia cards don't mean that it will run good. :)

We've only heard one side of things, while nVIDIA hasn't said anything of their upcoming plans. It's pointless to claim "who's in trouble" based on that.

As for TressFx, like most things that are system taxing, if you have the hardware to run it, and the feature is optimized (which TressFx was about 75-80% at the time), you'll get solid performance. On lower end cards, both AMD and GeForce, TressFx can cause significant system impacts. On my GTX 690, a GeForce card, no such problem.

And, just an FYI:  I never said that all AMD features will run "good" on nVIDIA hardware, I said it can be done. 

Any card that is basically not a GCN card will encounter some performance issues with tressfx (This includes AMD's older HD6000 series and under too) but seeing as how you have a gtx 690 that was mostly no problem for you. Depends on how you define "trouble" and I mean trouble in the sense that AMD can now play extra dirty against nvidia potentially. Got to remember all of these consoles are now under AMD's mercy so if anything I can see AMD calling alot of shots LOL. I dearly worry for nvidia users too seeing as how this is a big threat to their experience on geforce. I seriously hope AMD was about open source and integrity otherwise they were all just talk because what AMD could start doing is having games exclusively use the mantle api only therefore possibly locking out nvidia owners too which sucks. 

Yeah, we'll see. I'm not counting the chickens before they've hatched, because with technology, there are remedies to many "limitations".

But their are no remedies to property ownership is there ? :P



so, we are back to Glide eh? .................................



CGI-Quality said:
fatslob-:O said:
CGI-Quality said:

Yeah, we'll see. I'm not counting the chickens before they've hatched, because with technology, there are remedies to many "limitations".

But their are no remedies to property ownership is there ? :P

Which means nothing, especially in regards to GeForce vs Radeon.

I don't know but having an API under your control would mean something.