AMD GPU14 Tech Event Sept 25 - AMD Hawiian Islands

[taserbro]

If I made a game and I had the chance to add low level access support for the game with mantle, and add 2x performance to any gcn card, I jump at the chance.
Now the game bundle and such all make sense.

The sad part is that this could have happened a long time ago.
There's a lot of devs who've had to pull miracles out of their hats to make modern games run on outdated consoles and all their efforts would have been better used if hardware vendors were to give them any choice.

That said, I'm quite optimistic about all of this.
Making it easier to code closer to the metal isn't voodoo magic that only amd can do. I expect all their competitors to step up their game and eventually, this news to turn into a great step forward for PC gaming in general.

 

[sontin]

That is not an artificial limitation, that is a natural technical limitation, that was his point.

Sorry, but that is wrong.
A "natural technical limitation" is something like no support for DX11 because your hardware dont support Tessellation up to factor 64.
Mantle is an "artificial limitation" because is an API for a vendor specific hardware.

Yes, and before they already had to do that for PS3, Xbox360, and PC. I'm not sure why you think they won't put in the word for multiple code paths.

Xbox and PC are using DX. That's the reason why we have so many multi port games. PS3 is using OpenGl which makes is much harder to port a PS3 optimize game to other plattforms.

Yes there is. AMD specifically mentioned that the proliferation of GCN in the consoles is what allows this to make sense for game developers.

They mentioned the experience they used to create Mantle.

 

[Freaksterz]

Was the conference finished when they cut the stream or did they cut it for the NDA'd conference?
*Sorry offtopic *

 

[BD2003]

Agree with a lot of this. Understand why AMD is going in this direction, but long term for gamers it's harmful. Problem is, DirectX is just really inefficient (not to mention Windows) so from that perspective I welcome anything that alleviates the wasted processor cycles.

The perfect solution to me would be a close to the metal, open graphics API on Linux. And no OpenGL is not good enough, in fact it's a bit of a mess the last few years.

Those efficiency gains aren't worth it. From a price/performance perspective, we'll get better performance for our dollar over time by having the two of them fight it out for dominance. In a perfect world we wouldn't need inefficient high level APIs like DX, but that's just wishful thinking. Be careful what you wish for!

 

[AnandThenMan]

How is Nvidia going to "purpose" Mantle in the same way it will utilize AMD hardware? Saying Mantle is open sounds like lip service from AMD to me.

Be careful what you wish for!

I'm wishing that Linux replaces Windows completely, if I'm being blunt.

 

[sushiwarrior]

I, like most people, will watch where it goes. If it is truly increased performance, good for them. The market will sort it all out.

If something is sacrificed in order to obtain that performance, then it will fail badly.

The sacrifice is dev time I suppose, it takes work to optimize. But it increases the performance of console ports, and means that PC performance is better than anything possible with DirectX ideally.

 

[EightySix Four]

Was the conference finished when they cut the stream or did they cut it for the NDA'd conference?
*Sorry offtopic *

NDA'd

 

[EightySix Four]

Sorry, but that is wrong.

A "natural technical limitation" is something like no support for DX11 because your hardware dont support Tessellation up to factor 64.

Mantle is an "artificial limitation" because is an API for a vendor specific hardware.

You don't even realize that it is actually similar to not supporting DX11 because your hardware doesn't support Tessellation up to a factor of 64. They are fundamentally different architectures, this wasn't some business decision AMD reached to screw nVidia over... It's like being an ARM customer and being angry that Intel offers the Intel compiler...

 

[RiDE]

Was the conference finished when they cut the stream or did they cut it for the NDA'd conference?
*Sorry offtopic *

It's NDAd... probably to keep people from getting an aneurysm if the new card isn't faster than the Titan.

 

[sushiwarrior]

Sorry, but that is wrong.
A "natural technical limitation" is something like no support for DX11 because your hardware dont support Tessellation up to factor 64.
Mantle is an "artificial limitation" because is an API for a vendor specific hardware.

Nvidia hardware does not support GCN architecture. I feel like you really don't understand this

 

[Despoiler]

How is Nvidia going to "purpose" Mantle in the same way it will utilize AMD hardware? Saying Mantle is open sounds like lip service from AMD to me.


I'm wishing that Linux replaces Windows completely, if I'm being blunt.

Nvidia would have to write code that gives low level access to their cards. They would use common calls that AMD uses and perhaps have some additional ones created to support their unique architecture. Again, I think this is a big middle finger to MS DX.

 

[stahlhart]

Two people have just gotten one-day vacations for obscene language in this thread. This is the last warning to the rest of you.
-- stahlhart

 

[Arkadrel]

Sorry, but that is wrong.
A "natural technical limitation" is something like no support for DX11 because your hardware dont support Tessellation up to factor 64.
Mantle is an "artificial limitation" because is an API for a vendor specific hardware.

Is CUDA a "artificial" limitation put in place by Nvidia, since it only works on their GPUs? or does it only work on nvidia gpu's because of the way the GPU was designed? (thus not artificial)?

Its the same thing with GCN design and this Mantle API, its just coded close to the metal, like CUDA is for nvidia cards.

Only this is not ment for GPGPU, but for getting much higher performance out of games.

 

[BrightCandle]

Seems like everyone is ganging up on Microsoft Windows. We now have the start of an alternative API, sound innovation in the GPU and linux support for games being pushed via Steam. Its a perfect recipe for finally removing the need for Windows as gaming machines in a few years time, driven mostly by AMD as they leverage the monopoly given to them by the consoles.

Presumably Nvidia will write a driver for mantle, it might never be as efficient or it might work great. But it says a lot about how AMD feels about the DX updating process, they were saying there were no more updates 6 months ago and now we understand the agenda behind that statement, they are the ones trying to kill it.

 

[EightySix Four]

Nvidia would have to write code that gives low level access to their cards. They would use common calls that AMD uses and perhaps have some additional ones created to support their unique architecture. Again, I think this is a big middle finger to MS DX.

It depends on how close to the hardware this is. If it's really right next to the metal than it may not actually be able to do that. If it's further abstracted, the advantage over DIrectX or OpenGL won't be that big... We'll have to see.

 

[YBS1]

Mantle is going to be the best thing since TruForm! :awe:

 

[Riceninja]

given the rebrand going on with the r9s and mantle api, i feel like the 7950 and 7970 just got a life extension.

 

[Freaksterz]

Seems like everyone is ganging up on Microsoft Windows. We now have the start of an alternative API, sound innovation in the GPU and linux support for games being pushed via Steam. Its a perfect recipe for finally removing the need for Windows as gaming machines in a few years time, driven mostly by AMD as they leverage the monopoly given to them by the consoles.

Presumably Nvidia will write a driver for mantle, it might never be as efficient or it might work great. But it says a lot about how AMD feels about the DX updating process, they were saying there were no more updates 6 months ago and now we understand the agenda behind that statement, they are the ones trying to kill it.

About time things shake up alittle bit if you ask me.

 

[Arkadrel]

Mantle is going to be the best thing since TruForm! :awe:

I know your trolling everyone with that comment and it made me smile, when I read it,
however today every discrete pc GPU sold has "truform" (aka tessellation).

Seems like everyone is ganging up on Microsoft Windows.

^ this, wasnt too long ago there where alot of developers speaking out against windows.
Now it seems there are a few pushing for Linux.

 

[sontin]

Is CUDA a "artificial" limitation put in place by Nvidia, since it only works on their GPUs? or does it only work on nvidia gpu's because of the way the GPU was designed? (thus not artificial)?

Cuda was the first attempt for GPGPU. But yes, it's the same.

Only this is not ment for GPGPU, but for getting much higher performance out of games.

Cuda is on the same level like OpenCL. It is not really low level.
The difference here is that there is no vendor independet low level API or a work between Dice and nVidia.

 

[Fastx]

given the rebrand going on with the r9s and mantle api, i feel like the 7950 and 7970 just got a life extension.

Yes those were also my thoughts (w/nice possible speed bump in perf in supporting games like BF4 & future games) early on!

 

[YBS1]

I know your trolling everyone with that comment and it made me smile, when I read it,
however today every discrete pc GPU sold has "truform" (aka tessellation).

That was exactly my point actually. This stuff tends not to catch on unless it's supported by both of the players. Developers aren't going to want to develop for two different user bases unless they have some motivation (being paid to do so).

 

[BD2003]

How is Nvidia going to "purpose" Mantle in the same way it will utilize AMD hardware? Saying Mantle is open sounds like lip service from AMD to me.


I'm wishing that Linux replaces Windows completely, if I'm being blunt.

I agree completely, Windows has outlived it's usefulness. It's time to move on to *truly* open platforms and standards, enough with this proprietary BS already. It's fine in the early stages of a technology, but to make an attempt to close down what was previously open...it's just ridiculous.

Next thing you know, Cisco is going to announce a proprietary form of WiFi that's faster than anything else out there, but only works with Cisco gear. Can you imagine what a nightmare that would be?

 

[24601]

I agree completely, Windows has outlived it's usefulness. It's time to move on to *truly* open platforms and standards, enough with this proprietary BS already. It's fine in the early stages of a technology, but to make an attempt to close down what was previously open...it's just ridiculous.

Next thing you know, Cisco is going to announce a proprietary form of WiFi that's faster than anything else out there, but only works with Cisco gear. Can you imagine what a nightmare that would be?

It'd be more like if Cisco implemented a form of TCP/IP that only worked on Cisco hardware with a specific architecture.

 

[jvroig]

(Ok, since you brought this up, maybe you'll realize your mistake if I try one more time by leading from your own example).

A "natural technical limitation" is something like no support for DX11 because your hardware dont support Tessellation up to factor 64.

That is EXACTLY what CUDA and Mantle (if Mantle really is a close-to-the-metal API*) are!

In your example, you can't support DX11 Tess up to 64 because maybe your hardware doesn't have Tess units. Your hardware physically has zero Tess units, therefore it can never support DX11 Tess. Correct.

In CUDA and Mantle (again, if it is truly a close-to-the-metal API), the same logic applies. Because CUDA and Mantle will keep making calls to X, Y, or Z hardware units (whatever hardware units they are expecting to be present) and only NV arch (for CUDA) or GCN (for Mantle) will have those hardware units. It is exactly like how your hardware can't support DX11 because of the lack of Tessellation fixed-function hardware.

Do you realize your earlier mistake now, in calling CUDA and Mantle artificially locked down? Or will you prove that you are just trolling here?


Different topic, footnote:
*On whether Mantle truly is a close-to-the-metal API: I am getting mixed signals from AMD, or the press is misreporting. If Mantle is truly a close-to-the-metal API, then it cannot be "open" or cross-ISA/arch, because being locked to the specific arch is inherent to bare metal coding. Even now (decades since they arrived), we don't have assemblers that produce one type of machine code that can be run on any ISA like x86, ARM flavor, PowerPC, etc. It is simply impossible, because that machine code has to have a 1-1 target on the hardware side, and that is defined by the specific arch. But now there is talk and "confirmation" of Mantle being open or some such silliness.

You can't have it both ways. Either you are truly a close-to-the-metal API (therefore locked down to your target hardware arch), or you are still an abstraction layer (therefore can be "open" and used by other competing hardware archs). It can only be one or the other. If the press reports of being "open" are true, then this definitely isn't close-to-the-metal as we understand it, and simply another abstraction layer that supposedly has lower overhead than DX (i.e., far less useful and slower than if it were truly close-to-the-metal as initially thought). If it's just another abstraction layer that is abstracted high enough to accommodate multiple archs instead of just GCN, then it is a waste of time as the performance benefit is probably miniscule and not worth it.

I guess we'll only find out if they release more data or when the BF4 Mantle-powered version arrives, whichever happens first.