(GURU3D RUMOR) AMD Polaris 10 GPU To Offer Near 980 Ti Performance For 299 USD?

[flopper]

If you're a gamer...you're not going to care about self driving cars. You might care a bit about HPC, but not much and Deep Learning is something you wouldn't give a rats arse about.

So when folks mention, but but NVIDIA are heading into cars, I immediately realize that they're NVIDIA fans. Because the only reason one would mention this is if they have some sort of soft spot for NVIDIA as a corporation.

What's good for NVIDIAs financials, isn't something I care about (or AMDs for that matter). What's good for gamers, on the other hand, is something I care about as a gamer.

VR will push scalability and a need for more scalable designs and architectures not just on the hardware front but on the API and game engine front. This is good for gamers. It doesn't matter if you're a console' mobile or PC gamer due to the fact that modern APIs allow for portability between all segments of games, applications and software in general.

With AMD having locked in the console business, where an overwhelming amount of the market is to be found, coupled with Vulkan/DX12 means that AMD have locked in the PC business as well. People who reference NVIDIAs last big hurrah (80% market share of PC discrete DX11 GPUs) haven't realized or refuse to realize what is happening.

What is happening is that AMD is rising, far from being pushed into bankruptcy, they will instead dominate the gaming market in short order. I figure that 2016 is the beginning of their rise, 2017 is the establishment of their superiority and 2018 is the end game with Navi.

I've been saying this since August of 2015 or so. It sounded crazy back then but today...I think that more and more people are realizing that this is most likely what will occur.

By the end of 2017, we won't recognize Intel, AMD or NVIDIA. I keep saying it but just you wait and see... It's the truth.

Polaris 10 doesn't need to be high end because the software (games) are already tailored for it. NVIDIA, on the other hand, need big GPUs in order to keep competing and we're going to see this more and more as we head towards 2018.

Great minds comes to the same conclusions.
AMD played the long game, created Mantle that pushed Microsoft to create Dx12 or Vulkan would own them. AMD Owns the console market, guess what games are made to, first?

Once more Dx12 engines comes out and the story seems as far a 390 now beats the 980ti....ouch.....

VR while it can become great with a 360 degree camera education, gaming and other fields will love it and then we want better hardware that can do low latency and AMD can already do low latency as 980ti cant due to a 980ti makes you womit with VR as the latency hit is horrible.

AMD did a long plan and those of us that seen it for a long time knew one thing, its going to be great for AMD.

 

[renderstate]

What is happening is that AMD is rising, far from being pushed into bankruptcy, they will instead dominate the gaming market in short order. I figure that 2016 is the beginning of their rise, 2017 is the establishment of their superiority and 2018 is the end game with Navi.

AMD is certainly well positioned for growth in the gaming market after losing so much ground to NVIDIA, but your forecast sort of assumes competition will just sit and watch, which is quite unlikely. The reason for which AMD and NVIDIA are still around, while many other gfx companies aren't anymore, is that they showed over and over again resiliency and capacity to adapt to an ever changing market.

I've been saying this since August of 2015 or so. It sounded crazy back then but today...I think that more and more people are realizing that this is most likely what will occur.

You know what they say about a broken clock being right twice a day..

 

[Qwertilot]

If you're a gamer...you're not going to care about self driving cars.

This is going very off topic indeed, but I feel oblidged plead for some perspective here

Self driving cars are almost certainly coming and will change society much more than mildly faster GFX cards! Deep learning is the same.

Frankly they're also both vastly more interesting topics than GFX brand fights.

 

[USER8000]

AMD is certainly well positioned for growth in the gaming market after losing so much ground to NVIDIA, but your forecast sort of assumes competition will just sit and watch, which is quite unlikely. The reason for which AMD and NVIDIA are still around, while many other gfx companies aren't anymore, is that they showed over and over again resiliency and capacity to adapt to an ever changing market.




You know what they say about a broken clock being right twice a day..

But it is also equally ridiculous to expect AMD to not be able to get decent marketshare from Nvidia either. I am not sure why people are so sold that AMD won't be able to get back to around 30% to 40% share again - if Nvidia can gain marketshare from AMD,so can AMD gain marketshare from Nvidia.

I mean for some of us who have been enthuisiasts for a longtime,ATI managed to out sell the Nvidia 6000 series for a while with the inferior X800 series which had worse DX9 support.

 

[mohit9206]

I expect
980ti performance at $99
390 performance at $69
380 performance at $49
370 performance at $29
Come on AMD now don't disappoint me!

 

[Magic Carpet]

I expect
980ti performance at $99

I'll give fifty, tops

 

[Raising]

Gonna be hilarious to see AMD fanboys disappointed once more, rebrand after rebrand, the failure of the "overclockers dream" and now recently an overpriced dual gpu card that released almost a year too late.

 

[renderstate]

But it is also equally ridiculous to expect AMD to not be able to get decent marketshare from Nvidia either. I am not sure why people are so sold that AMD won't be able to get back to around 30% to 40% share again - if Nvidia can gain marketshare from AMD,so can AMD gain marketshare from Nvidia.

I mean for some of us who have been enthuisiasts for a longtime,ATI managed to out sell the Nvidia 6000 series for a while with the inferior X800 series which had worse DX9 support.

As I said AMD is well positioned to gain market share. But Mahigan is making hyperbolical "end game" statements that I don't think have much basis in reality.

 

[maddie]

I expect
980ti performance at $99
390 performance at $69
380 performance at $49
370 performance at $29
Come on AMD now don't disappoint me!

You forgot the "Ebay used cards line".

 

[airfathaaaaa]

As I said AMD is well positioned to gain market share. But Mahigan is making hyperbolical "end game" statements that I don't think have much basis in reality.

but nvidia is out of console market and mobile markets since no one will give them x86 licence and also no one wants to hear anymore about tegra chips on smartphones..
how you suggest that they will go upfront on those areas that jrp says that it will produce along with the pc market roughly 150 billions till 2019?

 

[Qwertilot]

Honestly? Probably fine.

There's always optimism of course, but the markets they are firmly established in are a lot more lucrative than mobile/consoles.

The neural network/deep learning stuff could be a massive market for them in the medium term. In the long term they've got some very serious competition looming - Intel of course but also plans for absolutely dedicated architectures, but that will all probably take a bit of time to get established and they're here now.

It is allowed for tech companies to all do well at once!

 

[railven]

Gonna be hilarious to see AMD fanboys disappointed once more, rebrand after rebrand, the failure of the "overclockers dream" and now recently an overpriced dual gpu card that released almost a year too late.

Glad I'm no longer an AMD fanboy

 

[sweetusernames]

... the failure of the "overclockers dream"...

What happened with Fiji I suspect was probably the same thing with Hawaii

http://techreport.com/news/27996/4gb-gtx-960s-trickle-into-retail-channels?post=893388

I mean the Nano is just proof of that, isn't it? 100W lower TDP with only 10-18% performance difference vs the Fury X

AMD just proved by themselves their own cards overclock horribly tbh.

 

[renderstate]

but nvidia is out of console market and mobile markets since no one will give them x86 licence and also no one wants to hear anymore about tegra chips on smartphones..
how you suggest that they will go upfront on those areas that jrp says that it will produce along with the pc market roughly 150 billions till 2019?

There is not much money in the console market for custom SoCs, which is why Intel and NVIDIA exited it and left the whole pie at AMD, that need it much more than they do. NVIDIA also exited the mobile market because it's another razor thin margins game, where only Apple and Samsung make real money with their SoCs (Intel should have followed NVIDIA in that decision instead of waiting for so long).

If AMD can profit in those markets that's good for them but let's face it, their situation is quite different. x86 is not in growing markets, except servers, so it's not really the end of the world not having a license, there are plenty of other means to make lots of money.

 

[JDG1980]

What happened with Fiji I suspect was probably the same thing with Hawaii

http://techreport.com/news/27996/4gb-gtx-960s-trickle-into-retail-channels?post=893388

I mean the Nano is just proof of that, isn't it? 100W lower TDP with only 10-18% performance difference vs the Fury X

AMD just proved by themselves their own cards overclock horribly tbh.

I think it would have dramatically improved the image of Hawaii if AMD had released a version focused on energy efficiency ("R9 290E"?) They could have given it one 6-pin connector and a power limit of 150W. Even if they'd used that crappy blower on the reference model, it would have little enough heat to dissipate that it wouldn't be too loud. Clock speeds, as indicated by that link you posted, would probably be around 800-900 MHz during normal gaming, maybe lower in FurMark and other extreme stress tests. This would have been competitive with GM204 in perf/watt, and would definitely have been the best mainstream card with one 6-pin connector.

 

[airfathaaaaa]

apparently there is if you have the tech and the licence for it...(also since x86 is pretty much everywere but mobiles i fail to see how its not making any money but ok)

also nvidia didnt left the mobile soc it got kicked out because they didnt had any good enough chip to compete at anything else than the graphics

thats why the infamous "x1" has only goten one win to just showcase its capabilities(and it has one of the biggest batteries on a tablet out of stock)

 

[PPB]

What happened with Fiji I suspect was probably the same thing with Hawaii

http://techreport.com/news/27996/4gb-gtx-960s-trickle-into-retail-channels?post=893388

I mean the Nano is just proof of that, isn't it? 100W lower TDP with only 10-18% performance difference vs the Fury X

AMD just proved by themselves their own cards overclock horribly tbh.

Nope, they are clocked past their perf/watt sweet spot. Hawaii because of the crammed design favoring perf/mm2 and its powerhog 512bit PHY, fiji because its unbalanced design making it lackuster in perf/transistor to begin with. Factor your competition in, if there is any perf disadvantage when you have your a0 silicon back compared to their product stack, you take the clocks up to 11 to compensate as making more masks and respining the silicon is uneconomical. You give power consumption the middle finger in the end.

At 28nm, gcn's sweetspot mhz/perf/watt lies in the 900 mhz area. Saying they badly overclock is faux argument because when AMD respected this clock range (launch 7xxx products) they all clocked in this mhz region and had +20% OC headroom. AMD had a less performant product at their sweetspot speeds and thus had to bin better since 2012 to this day. This is a k8 vs conroe situation to a much lesser degree. K8 in rhe end overclocked "badly" because AMD was binning the cpus to their limits to begin with so the performance deficit wasnt as big.





Sent from my XT1040 using Tapatalk

 

[renderstate]

also nvidia didnt left the mobile soc it got kicked out because they didnt had any good enough chip to compete at anything else than the graphics

What does it mean "it got kicked out"? I wasn't aware there was a supernational mobile SoC club that kicks out its members

NVIDIA didn't have much success in that market and progressively exited it. For instance this story, that highlights the beginning of a change of strategy, is 2 years old http://www.droid-life.com/2014/05/2...smartphones-unless-its-the-porsche-of-phones/. As I said Intel should have done the same instead of waiting for so long.

 

[tential]

What does it mean "it got kicked out"? I wasn't aware there was a supernational mobile SoC club that kicks out its members

NVIDIA didn't have much success in that market and progressively exited it. For instance this story, that highlights the beginning of a change of strategy, is 2 years old http://www.droid-life.com/2014/05/2...smartphones-unless-its-the-porsche-of-phones/. As I said Intel should have done the same instead of waiting for so long.

You said they left because it was razor thin margins, when he was correcting you in telling you that Nvidia didn't have a competitive chip.
You know what he meant, they didn't have much success because they couldn't deliver a competitive product...
Just like intel.

 

[Adored]

Margins are even thinner in automotive compared to smartphones (for Nvidia).

 

[renderstate]

You said they left because it was razor thin margins, when he was correcting you in telling you that Nvidia didn't have a competitive chip.

You know what he meant, they didn't have much success because they couldn't deliver a competitive product...

Just like intel.

The lack of success AND the razor this margins did the trick, one does not exclude the other and I certainly didn't imply so. I was also talking about the console market, that has razor thin margins too. Apart from discrete gfx NVIDIA is clearly after (potentially) highly remunerative and fast growing markets (deep learning and automotive).

 

[tential]

The lack of success AND the razor this margins did the trick, one does not exclude the other and I certainly didn't imply so. I was also talking about the console market, that has razor thin margins too. Apart from discrete gfx NVIDIA is clearly after (potentially) highly remunerative and fast growing markets (deep learning and automotive).

That doesn't address Adored's post above about the margins for cars....

 

[renderstate]

Margins are even thinner in automotive compared to smartphones (for Nvidia).

For now yes, but cars are evolving like never before and cost structure could change. I believe few doubt cars will become more and more like compute hubs.

 

[parvadomus]

I can see the chip being as fast or faster than stock 980TI. But not with 256-bit GDDR5.
I expect a higher clocked part, maybe ~1300Mhz with GDDR5X, that would do it for sure.

EDIT: expect this chip to be released later, and of course at a much higher price. (now we will get the 4850, later the 4890 )

 

[airfathaaaaa]

What does it mean "it got kicked out"? I wasn't aware there was a supernational mobile SoC club that kicks out its members

NVIDIA didn't have much success in that market and progressively exited it. For instance this story, that highlights the beginning of a change of strategy, is 2 years old http://www.droid-life.com/2014/05/2...smartphones-unless-its-the-porsche-of-phones/. As I said Intel should have done the same instead of waiting for so long.

it means that google and lg openly blamed tegra for being a poor perfomer and a battery eater

simply put they didnt deliver at anything besides graphics

also yeah nvidia is use to make claims that they dont care and bla bla

but you remember nvidia clash with intel back in 2009 till 2011? when nvidia wanted to emulate x86 on denver? remind me what was their statement after that? didnt they said that their arm+gpgpu solution is superior to x86?
where is denver now? where is their chipset department after intel remove their licence because they wanted to lock down intel cpus to be used only with certain gpus?

Interestingly the agreement also classifies an “Intel Architecture Emulator” as being a proprietary product. At first glance this would seem to disallow NVIDIA from making an x86 emulator for any of their products, be it their GPU holdings or the newly announced Project Denver ARM CPU. . Being officially prohibited from emulating x86 could be a huge deal for Denver down the road depending on where NVIDIA goes with it."


1.8. “Intel Architecture Emulator” shall mean software, firmware, or hardware that, through emulation, simulation or any other process, allows a computer or other device that does not contain an Intel Compatible Processor, or a processor that is not an Intel Compatible Processor, to execute binary code that is capable of being executed on an Intel Compatible Processor.

1.12. “Intel Compatible Processor” shall mean any Processor that (a) can perform substantially the same functions as an Intel Processor by compatibly executing or otherwise processing (i) a substantial portion of the instruction set of an Intel Processor or (ii) object code versions of applications or other software targeted to run on an Intel Processor, in order to achieve substantially the same result as an Intel Processor; or (b) is substantially compatible with an Intel Processor Bus.