AMD has unveiled 2 Polaris GPUs and will deliver "The most revolutionary jump in performance so far"

[Humbug]

It would certainly make fiscal sense to rebrand most of the line-up of a generation if you were planning on spending a lot of money on a general overhaul.

There was a rumour also that the Fury and Fury X were stopgap solutions that AMD had to rush out because Global Foundries couldn't deliver on 14nm finfet in time thus delaying the real next gen (of both Nvidia and AMD).

[Bob Jim]

It depends on what Nvidia is doing also. In the last couple of years Nvidia was unable to pull out anything revolutionary and AMD was ok to rebrand and say here we have the same performance...

But with Nvidia pascal coming this year (and AMD knew this for a long time) I have a feeling that AMD will go with an all new top-bottom 14nm / 16nm finfet linuep.

They almost certainly will. Looking at this article, https://en.wikipedia.org/wiki/Radeon, every time AMD shrinks die size they release an entirely new series of cards, excluding iGPUs used in their APUs.

[Tiber1337]

Darn this marketing tactic, advertising something that will be released more than half a year from now. We all remmeber how much we waited for the last gpu series.

[ivan134]

Darn this marketing tactic, advertising something that will be released more than half a year from now. We all remmeber how much we waited for the last gpu series.

It means AMD's marketing dept is finally doing its job properly. Marketing and software have been AMD's biggest problem. AMD has a better handle on drivers now and it will be beter going forward by the default of dx12 and other low level APIs requiring more work from the devs themselves than GPU vendors. This is a good 1st step and we'll have to wait and see if their marketing dept see it through to the end.

[Bob Jim]

It means AMD's marketing dept is finally doing its job properly. Marketing and software have been AMD's biggest problem. AMD has a better handle on drivers now and it will be beter going forward by the default of dx12 and other low level APIs requiring more work from the devs themselves than GPU vendors. This is a good 1st step and we'll have to wait and see if their marketing dept see it through to the end.

Better handle on drivers that refuse to let me play Star Wars Battlefront II without issues ;-;

[Trixanity]

Also keep in mind that AMD had to trash all their 20nm designs and due to their budget constraints, it was forced to re-brand a lot.

Nvidia on the other hand just went ahead and threw all their money behind Maxwell on 28nm because while it was an undesired scenario, it was merely a bump in the road for them and an economic burden they could easily bear. A costly affair, no doubt, but I believe it paid off for them - big time.

AMD had to bide their time and put out stop gap solutions and push their current offerings to the limit.

I'm quite certain they put the majority of their graphics budget towards Polaris and 14nm.

I really think we should appreciate this new direction RTG is going. The openness is a step in the right direction. No doubt.

[marldorthegreat]

Also keep in mind that AMD had to trash all their 20nm designs and due to their budget constraints, it was forced to re-brand a lot.

I really think we should appreciate this new direction RTG is going. The openness is a step in the right direction. No doubt. 

Actually no, they didnt. The 300 series was meant to be rebranded on 20nm, but couldnt so they opted for a better 28nm instead

 

And you do realise that polaris probably took 5 or more years to develop, meaning rtg has only had control since September.

[Humbug]

Darn this marketing tactic, advertising something that will be released more than half a year from now.

Investors need to be reassured sometimes. And the public likes to know that the company's roadmap is solid.

e.g. on the CPU side if we didn't know about Zen would there be any hope for the future of AMD?

 

Nvidia has been talking about pascal for years.

[alexyy]

Can't wait for some new GPUs let's hope it's not as disappointing as the fury x.

[Trixanity]

Actually no, they didnt. The 300 series was meant to be rebranded on 20nm, but couldnt so they opted for a better 28nm instead

And you do realise that polaris probably took 5 or more years to develop, meaning rtg has only had control since September.

Their 20nm designs, among other things, were supposed to have HDMI 2.0. So the display block was known to be updated. I doubt it was all re-brands for 20nm, unless you can provide a source that says otherwise.

Not once did I say RTG was the mastermind behind Polaris. Read the entire paragraph. I said that their openness is a step in the right direction, not Polaris. AMD has generally kept quiet about what they're working on. Getting this much information is fantastic.

[samcool55]

 

As usual the competition between Nvidia and AMD will give us decent prices. Nvidia will release Pascal to fight Polaris.

Prices are normally high in the interim period when one company launches the new gen and the other is still stuck on old gen. Doesn't look like that will happen this year.

 

Yea i'm hoping that, but i have checked prices of the R9 290, 970, R9 390 , R9 fury X and 980ti and they stayed exacly the same so sadly it's not always the case...

Let's hope nvidia and amd release their products around the same time so prices can come down quickly.

[marldorthegreat]

Their 20nm designs, among other things, were supposed to have HDMI 2.0. So the display block was known to be updated. I doubt it was all re-brands for 20nm, unless you can provide a source that says otherwise.

Not once did I say RTG was the mastermind behind Polaris. Read the entire paragraph. I said that their openness is a step in the right direction, not Polaris. AMD has generally kept quiet about what they're working on. Getting this much information is fantastic.

AMD could not afford to throw out designs because of delayed processes shrink, they didnt ever confirm 20nm. And gpus take more than a year to develop, rtg did not mastermind it, its probably been in development since before the 7970 came out.

[marldorthegreat]

Its possible but NO this cannot go on, its a new series which means all their lineup needs to be made from new 14nm polaris cores no rebrands and other idiotic stuff.

I dont think this is the case, it doesnt make sense to announce new core less power at 1080p vs competition and then throw rebrands in in non top end gpu, which will consume more than they advertise its not right.

 

This time around there shall be a whole new series mid-high end of 14nm polaris gpu's similar for nvidia.Low end is irrelevant nowdays.

I want to upgrade from my GTX 670 to an amd card and i will not accept a mid-high end card thats rebranded, fuck em ill go nvidia again if they do this.

But they wont be rebrands if they are based on 14m, with decreased power and increased performance, broadwell is just a rebranded haswell, on 14nm

[Trixanity]

AMD could not afford to throw out designs because of delayed processes shrink, they didnt ever confirm 20nm. And gpus take more than a year to develop, rtg did not mastermind it, its probably been in development since before the 7970 came out.

Ouch. I clarify my post and you still do not understand what I'm saying. I don't know how to make it any clearer. But let me try again.

Step in right direction = openness.

Openness means not keeping everything hidden, talking about things. In other words transparency.

RTG changes their policy towards openness. This is good.

Polaris != openness

While Polaris is nice and arguably a step in the right direction, I never referenced it in my post, so why you continue doing so is beyond me.

Obviously Polaris was not designed, produced and then shown off in benchmarks in less than four months. That would probably have been an unprecedented achievement.

Oh and Anandtech said this about the 20nm products: "The failure of 20nm has essentially stalled GPU manufacturing improvements, and in RTG’s case resulted in GPUs being canceled and features delayed to accommodate the unexpected stall at 28nm".

Now does this mean any planned product for 20nm was a waste of time and completely trashed? Depends on point of view. It was as far as I know a loss of money. Stuff was planned and designed for this and it never came to fruition. Any design that wasn't made was probably going back to the drawing board in the sense that now things were moving to 14nm and the existing design could probably be improved upon. I doubt it's merely a die shrink of whatever was planned for 20nm.

Just to reiterate: a lot of AMDs plans and designs relied on 20nm but as we all know that never happened. So AMD had to go with the backup plan, which was to wait for 14nm for most of their planned stuff although they have managed a few new chips such as Fiji and Tonga while waiting to pull the trigger on what was in the works.

Polaris in whatever form we get has no doubt been a long time in the making but I never mentioned that as a byproduct of the forming of RTG. You speak as if I did, though I still don't know how this happened. That's a pretty big achievement I'd dare say.

[Tomsen]

Also keep in mind that AMD had to trash all their 20nm designs and due to their budget constraints, it was forced to re-brand a lot.

Nvidia on the other hand just went ahead and threw all their money behind Maxwell on 28nm because while it was an undesired scenario, it was merely a bump in the road for them and an economic burden they could easily bear. A costly affair, no doubt, but I believe it paid off for them - big time.

AMD had to bide their time and put out stop gap solutions and push their current offerings to the limit.

I'm quite certain they put the majority of their graphics budget towards Polaris and 14nm.

I really think we should appreciate this new direction RTG is going. The openness is a step in the right direction. No doubt.

If we look back, at how AMD waited it out for 14/16nm meanwhile Nvidia rushed maxwell at 28nm, it might tell alot regards to both the new generation GPUs.

Nvidia invested in the current, AMD invested in the future. If Nvidia investement in Maxwell at 28nm is affecting pascal at 16nm, they might be in for some trouble, until Volta ofc.

 

 

Actually no, they didnt. The 300 series was meant to be rebranded on 20nm, but couldnt so they opted for a better 28nm instead

 

And you do realise that polaris probably took 5 or more years to develop, meaning rtg has only had control since September.

The 300 series didn't undergo any actual architecture changes, but did see some powergating improvement, better PCB and memory component if I remember correctly.

Also a lot of the 20nm stuff wasn't even GPUs. But SoC and APUs.

[Trixanity]

If we look back, at how AMD waited it out for 14/16nm meanwhile Nvidia rushed maxwell at 28nm, it might tell alot regards to both the new generation GPUs.

Nvidia invested in the current, AMD invested in the future. If Nvidia investement in Maxwell at 28nm is affecting pascal at 16nm, they might be in for some trouble, until Volta ofc.

While Maxwell was probably in many ways the culmination of what Nvidia has been working on over the past couple of years (particularly energy efficiency), I don't think anyone should discount Pascal.

I don't think architecturally speaking, that we should expect big changes compared to Maxwell but there will still be improvements and compute will probably be improved a fair bit and then there's of course a die shrink.

Polaris will be a bigger change for AMD than Pascal will be for Nvidia is what I'm trying to say but that doesn't mean that Pascal can't compete with Polaris. Not at all. I expect neck and neck benchmarks with the usual game biases.

There is still the unknown quantity which is DX12 and Vulkan. By unknown I mean how the new APIs, new graphics cards, drivers and game developers will coexist and behave.

So far we've seen heavy reliance on drivers and then there's stuff like game works. It will be interesting to see the playing field in two years time.

[Tomsen]

While Maxwell was probably in many ways the culmination of what Nvidia has been working on over the past couple of years (particularly energy efficiency), I don't think anyone should discount Pascal.

I don't think architecturally speaking, that we should expect big changes compared to Maxwell but there will still be improvements and compute will probably be improved a fair bit and then there's of course a die shrink.

Polaris will be a bigger change for AMD than Pascal will be for Nvidia is what I'm trying to say but that doesn't mean that Pascal can't compete with Polaris. Not at all. I expect neck and neck benchmarks with the usual game biases.

There is still the unknown quantity which is DX12 and Vulkan. By unknown I mean how the new APIs, new graphics cards, drivers and game developers will coexist and behave.

So far we've seen heavy reliance on drivers and then there's stuff like game works. It will be interesting to see the playing field in two years time.

Well, I was thinking more in the line of when these GPU might actually be available.

I doubt we would see any big chances going for a more compute-oriented GPU architecture. That is what I believe would be Volta.

GPU fans like to exaggerate the difference between the top end card from AMD and Nvidia. They are so fucking close, but most people act like one is truly "owning" the other, meanwhile we often see single digit FPS differences. I'd expect this phenomenal to continue.

Sure, both Nvidia and AMD will have influence in game titles, and how it will perform. If it isn't through drivers, it might be in the game itself. ^^

[Coaxialgamer]

Dammit ; i want to upgrade my gpu , but if i buy polaris when it comes out , maybe pascal will be better . Also , they might release bigger chips after ...

 

WHY IS IT SO HARD TO SPEND MONEY ???

[marldorthegreat]

Dammit ; i want to upgrade my gpu , but if i buy polaris when it comes out , maybe pascal will be better . Also , they might release bigger chips after ...

 

WHY IS IT SO HARD TO SPEND MONEY ???

I convert The great british pound to Euros and use the euros as toilet paper. 

[Shoob]

Dammit ; i want to upgrade my gpu , but if i buy polaris when it comes out , maybe pascal will be better . Also , they might release bigger chips after ...

 

WHY IS IT SO HARD TO SPEND MONEY ???

Just buy whatever fits your budget at the time of purchase. No need to wait months for something slightly better or slightly cheaper.

I remember some people were holding on their purchases of 980 TIs, because the Fury X might have been better. 

 

You'll also never get a Titan Z.

[Fetzie]

Dammit ; i want to upgrade my gpu , but if i buy polaris when it comes out , maybe pascal will be better . Also , they might release bigger chips after ...

 

WHY IS IT SO HARD TO SPEND MONEY ???

If the past is anything to go by, nVIdia and AMD will somehow manage to get their flagship cards approximately the same

[Trixanity]

Well, I was thinking more in the line of when these GPU might actually be available.

I doubt we would see any big chances going for a more compute-oriented GPU architecture. That is what I believe would be Volta.

GPU fans like to exaggerate the difference between the top end card from AMD and Nvidia. They are so fucking close, but most people act like one is truly "owning" the other, meanwhile we often see single digit FPS differences. I'd expect this phenomenal to continue.

Sure, both Nvidia and AMD will have influence in game titles, and how it will perform. If it isn't through drivers, it might be in the game itself. ^^

I misunderstood you then.

I think AMD might get a product out first, but I don't think Nvidia will be far behind. I'm not sure how much that'll affect the enthusiast market. That's where you'll find fanboys. I'm certain a lot of people will avoid AMD products and wait on the competition if that's the case.

The problem right now as we speak is how the market has shifted towards energy efficiency and AMD can't compete there. There's also the issue of some titles completely tanking AMD performance, especially on cards below Fiji. These things will make people jump to Nvidia and make people recommend purchasing Nvidia cards. Then there's the fact that very few people go for the top SKU. GTX 970 have very often been recommended as the best purchase and it's (despite issues) is a well deserved spot. Even if a 390 and all related cards can match it in raw performance.

Of course there's also the alleged driver disparity. This is of course mostly an urban myth that just won't die. "mostly" being the keyword. AMD has on occasion been slow to act in regards to drivers for new game releases. But other than that, it's pretty much the same. I mean if AMD drivers suck, then Nvidia drivers at the very least blow.

But that's an entirely different discussion.

[Coaxialgamer]

Just buy whatever fits your budget at the time of purchase. No need to wait months for something slightly better or slightly cheaper.

I remember some people were holding on their purchases of 980 TIs, because the Fury X might have been better. 

 

You'll also never get a Titan Z.

i'll probably get one when it's really old , just out of curiosity .

[JebKerman]

"The most revolutionary jump in performance so far"

 

dangerous statement AMD.

 

I mean, i hope it lives up to it, but if it doesn't...hoo boy.

[kelvindeschutter]

I misunderstood you then.

I think AMD might get a product out first, but I don't think Nvidia will be far behind. I'm not sure how much that'll affect the enthusiast market. That's where you'll find fanboys. I'm certain a lot of people will avoid AMD products and wait on the competition if that's the case.

The problem right now as we speak is how the market has shifted towards energy efficiency and AMD can't compete there. There's also the issue of some titles completely tanking AMD performance, especially on cards below Fiji. These things will make people jump to Nvidia and make people recommend purchasing Nvidia cards. Then there's the fact that very few people go for the top SKU. GTX 970 have very often been recommended as the best purchase and it's (despite issues) is a well deserved spot. Even if a 390 and all related cards can match it in raw performance.

Of course there's also the alleged driver disparity. This is of course mostly an urban myth that just won't die. "mostly" being the keyword. AMD has on occasion been slow to act in regards to drivers for new game releases. But other than that, it's pretty much the same. I mean if AMD drivers suck, then Nvidia drivers at the very least blow.

But that's an entirely different discussion.

how much they will sell also depend on how the pre-build market will adopt just look in the store almost all NVIDIA gpus