AMD GPU will power all 3 major next gen consoles?

[AnandThenMan]

Im talking about Consoles not PC or workstation, and as i have said before i believe next gen consoles will use Tessellation thats why i have mentioned it.

The Xbox360 supports tessellation. I have not seen it implemented to any great degree though. And let's face it, even PC titles with insane levels of tessellation showed little "wow" factor, the improvement was subtle. Until Devs take tessellation seriously, it's going to be a bit of a gimmick.

 

[nonameo]

The Xbox360 supports tessellation. I have not seen it implemented to any great degree though. And let's face it, even PC titles with insane levels of tessellation showed little "wow" factor, the improvement was subtle. Until Devs take tessellation seriously, it's going to be a bit of a gimmick.

I'm not sure if there's support for it, but ati has had tess for a while, though it didn't have good takeup.

http://en.wikipedia.org/wiki/Truform

not that it is really relevant 😀

 

[Genx87]

We are probably at least 3-4 full GPU generations away from being able to use PhysX and Tessellation properly in videogames. PhysX might as well not exist since it's a proprietary standard. That just leaves us with Tessellation. On that front, it would make little difference for PC game development if they put a GTX580 or an HD6970 in the PS4. NV needed 3 GTX480s to run their lighthouse tessellation demo....

I would say 2 at most. Which puts us into 2014. When the next consoles are due. Tesselation is a major pillar of DX11. Most high end cards even 2 generations past the introduction of DX9 were doing fine in most intensive games. I said Physics, not PhysX. I think that is where the next frontier in realism will happen in gaming. Ragdoll effects are really pathetic today. We need destructible environments, objects that truely interact ect.

 

[nenforcer]

If this is true then I think it is a serious blow to nVidia to be locked out of an entire console generation. There was speculation that Tegra might power the new Playstion Vita but that didn't come to be.

I would be surprised if nVidia didn't somehow find a way onto the Playstation 4.

 

[videoclone]

whatever they get i hope its a 28nm part...

Moving to a 65nm part to a 28nm will be a huge step and better futureproof the consoles.

 

[darkswordsman17]

If this is true then I think it is a serious blow to nVidia to be locked out of an entire console generation. There was speculation that Tegra might power the new Playstion Vita but that didn't come to be.

I would be surprised if nVidia didn't somehow find a way onto the Playstation 4.

We'll see how the GPUs turn out, but if they build on current AMD designs then it won't be a huge deal as both AMD and Nvidia will have fairly different GPUs going forward. However, if AMD can get Microsoft and Sony to use something based on GCN then it could turn out to be a coup.

whatever they get i hope its a 28nm part...

Moving to a 65nm part to a 28nm will be a huge step and better futureproof the consoles.

I believe both the 360 and PS3 are on 45nm. The 360 even integrated the CPU and GPU. Certainly it would still be a significant improvement, and I think it would be best for them to start on the smallest manufacturing node they can.

 

[Mopetar]

The process really doesn't have much to do with making it future proof. The Xbox 360 GPU is currently made on a 45 nm process, after being shrunk from a 60 nm process, and being originally made with a 90 nm process. It still has the same number of transistors, only now it's smaller, more energy efficient, and gives off less heat.

If they want something "future proof" they'll have to adopt designs based on the latest architectures from AMD or Nvidia. The only benefits of starting with a 28 nm process are that they can be a little more aggressive with the design, but they'll have to balance that against potentially lower yields or other factors. Otherwise they'll just design a chip to be fabricated on a more mature process and drop down to 28 nm when it becomes practical.

 

[GaiaHunter]

I would say 2 at most. Which puts us into 2014. When the next consoles are due. Tesselation is a major pillar of DX11. Most high end cards even 2 generations past the introduction of DX9 were doing fine in most intensive games. I said Physics, not PhysX. I think that is where the next frontier in realism will happen in gaming. Ragdoll effects are really pathetic today. We need destructible environments, objects that truely interact ect.

So far I've been disappointed with what Tessellation brought to the table - generally it is a minor increase (in many cases it is more a question of looking different rather than objectively better) in IQ at a huge performance cost.

Maybe it is a question of games not being built from the ground for tessellation since the tessellation ready user base is small?

Hardware PhysX, at first look, seems to take quite a big blow - all potential hardware physics effects seems to be locked on AMD platform, which won't be using hardware PhysX.

I'm not seeing developers changing their console hardware physics implementation to physX for their PC ports, which leaves NV with only whatever exclusives for PC decide to go hardware physX or more games where hardware physX is an afterthought.

 

[Blitzvogel]

The way things costs now, it would be stupid for any of the big three to switch their major chipsets to something completely new and unfamiliar, especially the PS4 and Xbox 3. Obviously developer know how will be important in the transition, and leveraging that with more powerful but familiar architecture would only serve to benefit everyone, especially, if a minor upgrade in power is the plan (like 3x or so the graphics and computing prowess).

Sony certainly seems the most mysterious at this point, and to be honest, I think they should stick with a new upgraded Cell variant as has been rumored. Improved SPEs, hopefully 16 of them, and Power7 based PPUs (x2?) would make it much more powerful, yet developers would automatically understand the architecture. As for the GPU front, I would be very surprised to see Sony cock block Nvidia. Fermi is a good architecture, and even something as "minor" as GF106 (GTS 450) would be a good 4x the capability (probably much more) of the RSX, with no Cell assistance needed. The similar hardware would make semi-hardware backwards compatibility a reality too, with the extra Cell hardware working to translate RSX functions into GF10-whatever compatible code.

Decent backwards compatibility and developer familiarity will be more important than ever for the next round. Nintendo could afford it's fresh start. Microsoft could too (barely), but they know they need to keep things simple and just get the games out more than anything which means giving developers the means to make good software without too much of a rise in cost. Sony easily is in the biggest hole here, and I'm beginning to think it's the beginning of the end for them in the home console business. I think there are two or three major rounds left in the industry before everything just goes into phones, tablets and practical computing platforms which will be just as good at delivering entertainment in the home as well as on the go via wireless functions like sending video wirelessly. If anything consoles are just reverting into fixed function boxes like media systems and DVRs, and that is a likely course for them. They won't be "real gaming platforms" anymore. Just true entertainment media systems running software that is analogous (yet proprietary) with what will be in other devices like they are now, but with less focus on gaming.

 

[tviceman]

I thought Microsoft and Sony have said over and over again that the current generation is only half way through (or slightly more now). If so, then to have the hardware specs banged out already for next gen MS and Sony consoles sounds a little silly to me.

 

[(sic)Klown12]

I thought Microsoft and Sony have said over and over again that the current generation is only half way through (or slightly more now). If so, then to have the hardware specs banged out already for next gen MS and Sony consoles sounds a little silly to me.

They can each support two console simultaneously. Sony had this happen when the PS3 first released as the PS2's sales remained strong for a fairly long time afterwards. The first generation of games of the new consoles will probably come out on X360 and PS3 also.

 

[3DVagabond]

So far I've been disappointed with what Tessellation brought to the table - generally it is a minor increase (in many cases it is more a question of looking different rather than objectively better) in IQ at a huge performance cost.

Maybe it is a question of games not being built from the ground for tessellation since the tessellation ready user base is small?

Hardware PhysX, at first look, seems to take quite a big blow - all potential hardware physics effects seems to be locked on AMD platform, which won't be using hardware PhysX.

I'm not seeing developers changing their console hardware physics implementation to physX for their PC ports, which leaves NV with only whatever exclusives for PC decide to go hardware physX or more games where hardware physX is an afterthought.

They'll just throw some money at the game producers to have the PC version a TWIMTBP title. They'll give them some code that runs better and/or, in the case of PhysX, exclusively on nVidia hardware. Along with all of the other marketing tactics. They'll maintain the status quo in the PC graphics card market.

 

[RussianSensation]

I thought Microsoft and Sony have said over and over again that the current generation is only half way through (or slightly more now). If so, then to have the hardware specs banged out already for next gen MS and Sony consoles sounds a little silly to me.

You are right on that. Remember 6 months ago so many rumors and articles of Sony and Microsoft execs coming out with statements that they are only half way into this generation? I think they were either downplaying the next generation or got caught with their pants down when Nintendo announced Wii U to launch in Q1 2012. It's possible they sped up their plan to release the consoles 2+ years earlier. I am not sure why it would take them 5 years to design a console. The hardest decisions probably revolve around the controller/motion sensing capabilities, media discs, etc. Picking hardware should be fairly easy and fast. You only have 2 GPU vendors to choose from and outside of the GTX580, neither offers a performance advantage. So you just pick based on efficiency per watt and cost you can get it at.

I think Nintendo should have released a Microsoft / Sony style marketing campaign and claimed that Wii U would have 5x faster GPU than PS3. Then we may have seen a 28nm GTX580 in the PS4. But with Wii U only having an RV770 GPU, PS4 / Xbox3 can just put a $120-130 GTX460 / HD6850 in there - not really setting the bar too high.

I like to look at mobile GPU offerings as a better predictor. The current HD6970 M only has 960 SPs. At the current time, that's my upper limit for how fast a GPU we can expect in the next generation of consoles if they are to launch in 2012.

 

[3DVagabond]

I like to look at mobile GPU offerings as a better predictor. The current HD6970 M only has 960 SPs. At the current time, that's my upper limit for how fast a GPU we can expect in the next generation of consoles if they are to launch in 2012.

I think that's very likely in the ballpark. :thumbsup:

 

[badb0y]

If these consoles are due in 2013-2014 they have to start banging the specs out right now.

 

[Saico]

Microsoft and Sony may suddenly jump to ARM by 2013-2014. If this happens - Nvidia will be the only choice.

 

[Cerb]

"It is highly likely that Microsoft's 3rd generation Xbox will be sporting a new IBM cell processor as well although it is slightly, ever so slightly possible, that this could change."

The Cell was all but abandoned by IBM, and MS' Xenon (weak Power6-type cores) easily won in real performance for a minimal amount of developer-hours, every single time. It would be good for MS to keep going with IBM, but with a more general-purpose chip. I would seriously hope that is a typo (IE, should say Sony), or misinformation by MS. Power7 looks to be a perfect CPU to rip parts out of, and they will have a nice compute-capable GPU.

wouldn't that mean removing a ton of xtors that exist for CUDA reasons? it would need it wouldn't it to get it anywhere near AMD in thermals and power/watt?

Wouldn't that be cutting on their nose to spite their face? CUDA should be a specific selling point, key features of CUDA are integrated into the HW, and CUDA is far more mature than any software support GCN will have, much less AMD's VLIW support.

If they choose AMD gpus then clearly the need for highly advanced tess is low.

It's practically nonexistent. It may have been better for AMD to make tess that wasn't tied to memory channels, but the reality is that it takes unplayable scenarios, or very expensive cards, for nVidia's tess to show itself being really great. Neither should be applicable to consoles, nor the majority of PC games, either. Tesselation is going to take a few generations of both HW and games, to really be useful/needed. That it is required for DX11 means that it will be there, though, and will get used.

Microsoft and Sony may suddenly jump to ARM by 2013-2014. If this happens - Nvidia will be the only choice.

On top of that being highly implausible (ARM designing CPUs that will not be good for mobile devices), I don't get how it would make nVidia their only choice. In fact, I don't see how moving to ARM would affect their GPU choices one little bit.

 

[AtenRa]

That advantage only comes into play in the GTX470/480/570 and 580. The lower end chips have no Tessellation advantage over the 6900 series because they have far fewer Tesselation engines. The problem is NV's chips with the Tessellation performance consume a lot more power to begin with. HD6950 2GB consumes less power than even the GTX560 Ti.

At each price category NV GF100 (and derivative chips) has superior performance/watt in Tessellation.

The link you provided for the power consumption, tests Crysis and not a DX-11 Tessellation game or benchmark. I clearly talked about DX-11 Tessellation performance/watt and in order to see that, you really have to test a real Tessellation game or benchmark not Crysis in DX-10.

Have a look at TessMark OpenGL, http://www.geeks3d.com/20110408/download-tessmark-0-3-0-released/

You really need to understand that Tessellation in DX-11 API is not programmable (Hull-Shader and Domain-Shader are programmable) and the only way to get more performance is to install more Tessellators in the architecture of your Graphics card, that’s why NV GF100 is a multicore (Multi-tessellation cores) design and only 69xx has dual core design from AMD, HD58xx and HD68xx series are single-Tess Core designs.

Next gen AMD architecture GCN (Graphics Core Next) will utilize a multi-Tessellation core design much like NV GF100/110.

It gets even worse for NV from there as we go down the food chain. HD6870 consumes about the same power than the GTX460 but has superior performance to begin with. Alternatively, HD6850 has similar performance to the 460, but consumes less power.

AMD simply has superior performance per watt at nearly every price level below a GTX580. For the consoles this is probably the most important metric since fitting a GTX570/580 is just not an option based on their power consumption.

Im talking about Tessellation and you keep saying that AMD cards have better performance/watt in general use, have you a problem understanding what Im talking about ??

Why not test HD6870 against GTS550 in Tessellation and observe witch card has better performance/watt ?? Even GTS450 can keep up against HD5870 or HD6870 in Tessellation.

http://www.gpu-tech.org/content.php...sellation-punch-does-the-Geforce-GTS-450-have

 

[3DVagabond]

^^^We are not just rendering triangles.

 

[Bateluer]

This is good for AMD. They need to invest the profit made in CPU R&D. 😛

 

[Gikaseixas]

Much needed income for AMD if true. If a BD APU makes its way to consoles it means even bigger $$ for them.

 

[Carfax83]

NV needed 3 GTX480s to run their lighthouse tessellation demo....

You mean this lighthouse demo?

According to the Nvidia rep @ 1:50, it was rendered using one GPU, and not three.

 

[Dribble]

Tessellation, even the relative power usage merits of the geforce's vs the radeon's probably aren't key here - both are close enough to each other that either could be used.

I suspect this has more to do with profit margins - basically the console makers want to pay the gpu makers the minimum per gpu possible. Whoever bids lowest wins. Chances are AMD bid lower. Nvidia have a history of refusing to sell their hardware too cheap which is why they lost the xbox 360 contract (not I suspect because of any xbox history but because AMD offered a better deal).

They were lucky to get the PS3 really - Sony originally was just planning to use the cell, when they realised that wasn't going to work they had to go begging. They probably couldn't have used ati as MS already had ati's custom gpu. If Sony had gone for an off the shelf one it would obviously have been weaker and that's very bad marketing. Nvidia probably cost more but because it was different they could at least claim it was better and the PS3 cost a fortune anyway so what's a dollar or two more 🙂

This does lead to the obvious question - how can two competitors releasing about the same time in the same "high powered" console market use the same gpu company? Either they both get the same gpu which is bad for marketing as it makes it harder for them to compete, or one gets a better gpu which is really bad for the other.

Sony might not want to use nvidia but I'd be surprised if they were to use ati.

 

[RussianSensation]

^^^We are not just rendering triangles.

Ya, seriously. If as a console developer I am faced with a decision of choosing between a card that has better performance per watt in games vs. a card that only wins in 2% cases once you enable Tessellation, I am choosing a card that has better performance per watt in games. Modern cards just can't use Tessellation unless you are talking about GTX470/480/GTX570/580, which are not going to find their way into a console.

 

[zephyrprime]

"It is highly likely that Microsoft's 3rd generation Xbox will be sporting a new IBM cell processor as well although it is slightly, ever so slightly possible, that this could change."

The Cell was all but abandoned by IBM, and MS' Xenon (weak Power6-type cores) easily won in real performance for a minimal amount of developer-hours, every single time. It would be good for MS to keep going with IBM, but with a more general-purpose chip. I would seriously hope that is a typo (IE, should say Sony), or misinformation by MS. Power7 looks to be a perfect CPU to rip parts out of, and they will have a nice compute-capable GPU.

I agree. I expect cell will be dropped by sony and all three console makers will be using IBM Power derivatives of some sort. The next generation of consoles will be odd in that they will be very similar in architechture.