AMD Polaris architecture – GCN 4.0

[Mtom]

People want lower power consumption AND power. Or efficiency, in other words. The argument about that was never in total isolation like you're making it sound. With Maxwell, Nvidia were able to provide better-than-Titan performance while drastically cutting power consumption at the same time. That's what the 'excitement' about that was. Maxwell was simply a more efficient architecture than what AMD had(has, technically).

If AMD can get on equal footing here with Polaris, then awesome. I dont think anybody will deny that is a good thing. However, if AMD cards use less power but are proportionally less powerful, then no, nobody is going to consider that a 'win' for AMD(though notebook and small form factor users will be happy to have even more options).

The cards need to be powerful in their own bracket. The R9-480 don't have to compete with the 1080Ti (or even the 1080). It should be a good upgrade to those running a 380/960 card. They get a lot more power with the same power consumption they had.

 

[Mtom]

It's a good deal, but it's not that magical unless it's at least clearly 'better than 390' level.

I think if AMD could do a 490 that is very close to a 390X for $200 and then a 490X that beats a 390X for $250, it would be hard for Nvidia to match.

I expect the GTX 1060 to be around GTX 970 performance at least, and that will likely be a $200 card, so that's kind of what AMD should be thinking about.

They clearly said, they will bring the minimum VR spec to more affordable level...which is now at $330-350 with the 390/970.
So yeah i bet around $250...just the 480 will beat the 390(x), not the 490.

 

[Seanspeed]

The cards need to be powerful in their own bracket. The R9-480 don't have to compete with the 1080Ti (or even the 1080). It should be a good upgrade to those running a 380/960 card. They get a lot more power with the same power consumption they had.

You're really just saying the same thing I am, just in a different way. So yea, I agree.

Just pointing out that the argument about power consumption was never solely about power consumption and nothing else. It was about overall efficiency.

They clearly said, they will bring the minimum VR spec to more affordable level...which is now at $330-350 with the 390/970.
So yeah i bet around $250...just the 480 will beat the 390(x), not the 490.

I'm not sure what you think the 480 will be specifically. Do you think there will only be a 480 and a 490 and no 'X' versions? I'm just not sure how they're going to turn Polaris 10 into 4 different GPU's, ya know?

It seems you're arguing naming schemes more than anything here.

 

[AlamoX]

personally i see any GPU 30-40% faster than a 390 as being useless especialy for it's price, it's overkill for 1080/1440p, and it's nowhere near what you need for 120fps or 4k.
so unless the next flagship comes with about 100% performance increase over a 390, then meh, not for me!

 

[flopper]

tech and beauty in one deal
The duo might be something that foretell the Vega duo.
Binary stars looks awesome.

http://videocardz.com/59344/xfx-radeon-pro-duo-pictured-up-close

 

[CAT-THE-FIFTH]

personally i see any GPU 30-40% faster than a 390 as being useless especialy for it's price, it's overkill for 1080/1440p, and it's nowhere near what you need for 120fps or 4k.
so unless the next flagship comes with about 100% performance increase over a 390, then meh, not for me!

You do realise we have not had those sorts of increases for like a decade??

The GTX280 was around 40% faster than an 8800GTX and probably less than that over an 8800 Ultra:

http://tpucdn.com/reviews/Zotac/GeForce_GTX_280_Amp_Edition/images/perfrel.gif

The HD4870 was around 50% faster than an HD3870 but the HD3870 was slower than an 8800GT:

https://tpucdn.com/reviews/Diamond/HD_4870/images/perfrel.gif

The HD5870 was 30% to 40% faster than an HD4870 and 20% faster than a GTX285:

https://tpucdn.com/reviews/ATI/Radeon_HD_5870/images/perfrel.gif

The GTX480 was 10% faster than an HD5870:

https://tpucdn.com/reviews/NVIDIA/GeForce_GTX_480_Fermi/images/perfrel.gif

The HD7970 was 30% to 40% faster than an HD6970 and 10% faster than a GTX580:

https://tpucdn.com/reviews/AMD/HD_7970/images/perfrel.gif

The GTX680 was 5% faster than an HD7970 and 20% faster than a GTX580:

https://tpucdn.com/reviews/NVIDIA/GeForce_GTX_680/images/perfrel.gif

The Geforce Titan was 30% faster than a GTX680,40% faster than a GTX580 and 60% faster than a GTX580:

https://tpucdn.com/reviews/NVIDIA/GeForce_GTX_Titan/images/perfrel.gif

Double R9 390 performance would be at least a 50% increase in performance over a GTX980TI:

https://tpucdn.com/reviews/Gigabyte/GTX_980_Ti_XtremeGaming/images/perfrel_2560_1440.png

I am not sure how you are expecting a more mainstream GPU to achieve something which has not been done for like a decade.

 

[Diagro]

Tbh, Im wanting a 50% performance increase from my 290x to what ever the next flagship is.

 

[CAT-THE-FIFTH]

The flagship card is Vega - Polaris is mainstream.

If you want big performance increases you need to wait for the bigger Nvidia and AMD chips being released later in the year or next year.

 

[SimBy]

You do realise we have not had those sorts of increases for like a decade??

The GTX280 was around 40% faster than an 8800GTX:

http://tpucdn.com/reviews/Zotac/GeForce_GTX_280_Amp_Edition/images/perfrel.gif

The HD4870 was around 50% faster than an HD3870:

https://tpucdn.com/reviews/Diamond/HD_4870/images/perfrel.gif

The HD5870 was 30% to 40% faster than an HD4870 and 20% faster than a GTX285:

https://tpucdn.com/reviews/ATI/Radeon_HD_5870/images/perfrel.gif

The GTX480 was 10% faster than an HD5870:

https://tpucdn.com/reviews/NVIDIA/GeForce_GTX_480_Fermi/images/perfrel.gif

The HD7970 was 30% to 40% faster than an HD6970 and 10% faster than a GTX580:

https://tpucdn.com/reviews/AMD/HD_7970/images/perfrel.gif

The GTX680 was 5% faster than an HD7970 and 20% faster than a GTX580:

https://tpucdn.com/reviews/NVIDIA/GeForce_GTX_680/images/perfrel.gif

The Geforce Titan was 30% faster than a GTX680,40% faster than a GTX580 and 60% faster than a GTX580:

https://tpucdn.com/reviews/NVIDIA/GeForce_GTX_Titan/images/perfrel.gif

Well, not really. HD4870 was in some cases more than 100% faster than previous gen. And was very affordable.
http://images.anandtech.com/graphs/atiradeonhd4870_062408145208/17130.png

 

[SimBy]

It's also worth noting how far behind AMD was performance wise yet still had 50% market share.

It really tells you how incredibly skewed marketshare is right now.

 

[Diagro]

The flagship card is Vega - Polaris is mainstream.

If you want big performance increases you need to wait for the bigger Nvidia and AMD chips being released later in the year or next year.

Yea that's not a problem.

the 290x handles itself well in 4k for the time being.

And when AMD / game devs fix crossfire i can just turn on the second one for a boost

 

[CAT-THE-FIFTH]

Well, not really. HD4870 was in some cases more than 100% faster than previous gen. And was very affordable.
http://images.anandtech.com/graphs/atiradeonhd4870_062408145208/17130.png

Some cases - the average was not as much and remember the HD3870 was a £150 card and hard significant weakness(AA being one of them) and was not the fastest - I had one and a 8800GTS 512MB.

People are just making grandiose performance claims that Polaris must be double the speed of an R9 390 when even the Titan was like £700 for that 60% average increase over a £400 GTX580,and launched 9 months after the GTX680.

AMD has stated its a mainstream desktop and laptop series of chips,targetting price/performance.

Vega is the enthusiast chip.

We knew this for a long time.

If people expect the £250 chip to be 50% faster than a GTX980TI,then what do they expect Vega to be?? Double the speed??

That would make Polaris and Vega the greatest GPUs probably made in a decade.

I doubt it.

Yea that's not a problem.

the 290x handles itself well in 4k for the time being.

And when AMD / game devs fix crossfire i can just turn on the second one for a boost

Plenty of us don't have R9 290 or GTX970 level cards - if AMD can distil the performance of a £350 to £450 Fury into something costing £175 to £250,that would be a reasonable increase for the segment.

Plus the R9 290/390 despite being solid cards are hard to fit into smaller SFF rigs,which makes the GTX970 the only real choice in that segment(Not had an ATX main rig for a long time).

If you even look at the last few generations even the much regarded HD5870 was only really 20% faster on average than a GTX285 but was cheaper,consumed less power and had DX11(which no game really used at the time).

It sold like hot cakes at £300 and it wasn't a massive increase in performance either.

 

[SimBy]

Some cases - the average was not as much and remember the HD3870 was a £150 card and hard significant weakness(AA being one of them) and was not the fastest - I had one and a 8800GTS 512MB.

People are just making grandiose performance claims that Polaris must be double the speed of an R9 390 when even the Titan was like £700 for that 60% average increase over a £400 GTX580,and launched 9 months after the GTX680.

AMD has stated its a mainstream desktop and laptop series of chips,targetting price/performance.

Vega is the enthusiast chip.

We knew this for a long time.

If people expect the £250 chip to be 50% faster than a GTX980TI,then what do they expect Vega to be?? Double the speed??

That would make Polaris and Vega the greatest GPUs probably made in a decade.

I doubt it.

Oh don't get me wrong I don't expect anything faster than Fury non X from Polaris. Was just saying we had large leaps in performance before.

 

[CAT-THE-FIFTH]

Oh don't get me wrong I don't expect anything faster than Fury non X from Polaris. Was just saying we had large leaps in performance before.

I don't disagree but generally over weak cards,or at a massively increased cost since massive chips are used. If people expect these be mega performing GTX980TI slaying cards,they are setting themselves up for disappointment.

If AMD did manage somehow get close to a GTX980TI at half the price I think AMD would be onto a winner.

The GTX1070 might get there,but is probably another segment up,ie, £300 to £400.

This is why that even though I have a relatively weak stop gap card,I am going to try and wait until Vega is released if possible,just to put some downwards pressure on the AMD and Nvidia cards under £300.

People forget that the first batch of 28NM cards had very disappointing pricing for the performance they had. It took like nearly a year,for things to really start improving.

The only reason the 40NM cards were decent value from the start was because AMD were very aggressive on pricing. Even then cards like the HD5770 were not that well priced at launch IMHO.

 

[Seanspeed]

personally i see any GPU 30-40% faster than a 390 as being useless especialy for it's price, it's overkill for 1080/1440p, and it's nowhere near what you need for 120fps or 4k.
so unless the next flagship comes with about 100% performance increase over a 390, then meh, not for me!

390 is a good 1080p card, but not great for 1440p.

You really want Fury/980Ti performance for 1440p unless you are not that picky about hitting 60fps.

 

[Mtom]

You're really just saying the same thing I am, just in a different way. So yea, I agree.

Just pointing out that the argument about power consumption was never solely about power consumption and nothing else. It was about overall efficiency.


I'm not sure what you think the 480 will be specifically. Do you think there will only be a 480 and a 490 and no 'X' versions? I'm just not sure how they're going to turn Polaris 10 into 4 different GPU's, ya know?

It seems you're arguing naming schemes more than anything here.

Well so far we only heard that the P10 will be the 480, and if its a bit faster than a 390 but not beating the Fury this is the usual step up from a new generation. So i'll go with tah.

 

[Relentless81]

I'm starting to get the feeling that Fury range will stay top end for the next 12 Months. I'm starting to wonder if its worth waiting as I want an upgrade to my 290x as its struggling to hold 60fps in bf4 ultra settings 1440p

Its getting frustrating all this waiting, release the specs and names of the cards already!

 

[Seanspeed]

You do realise we have not had those sorts of increases for like a decade??

The GTX280 was around 40% faster than an 8800GTX and probably less than that over an 8800 Ultra:

http://tpucdn.com/reviews/Zotac/GeForce_GTX_280_Amp_Edition/images/perfrel.gif

The HD4870 was around 50% faster than an HD3870 but the HD3870 was slower than an 8800GT:

https://tpucdn.com/reviews/Diamond/HD_4870/images/perfrel.gif

The HD5870 was 30% to 40% faster than an HD4870 and 20% faster than a GTX285:

https://tpucdn.com/reviews/ATI/Radeon_HD_5870/images/perfrel.gif

The GTX480 was 10% faster than an HD5870:

https://tpucdn.com/reviews/NVIDIA/GeForce_GTX_480_Fermi/images/perfrel.gif

The HD7970 was 30% to 40% faster than an HD6970 and 10% faster than a GTX580:

https://tpucdn.com/reviews/AMD/HD_7970/images/perfrel.gif

The GTX680 was 5% faster than an HD7970 and 20% faster than a GTX580:

https://tpucdn.com/reviews/NVIDIA/GeForce_GTX_680/images/perfrel.gif

The Geforce Titan was 30% faster than a GTX680,40% faster than a GTX580 and 60% faster than a GTX580:

https://tpucdn.com/reviews/NVIDIA/GeForce_GTX_Titan/images/perfrel.gif

Double R9 390 performance would be at least a 50% increase in performance over a GTX980TI:

https://tpucdn.com/reviews/Gigabyte/GTX_980_Ti_XtremeGaming/images/perfrel_2560_1440.png

I am not sure how you are expecting a more mainstream GPU to achieve something which has not been done for like a decade.

Kepler was actually a pretty big jump over Fermi, especially in the newer games at the time. 660Ti was often 50% faster than 560Ti. 670 often 50% faster than 570. The 680 was less so over the 580, but when you remember that the 680 was only the 'midrange' Kepler on a 300mm die compared to the 570mm die of the 570/580, it's still quite an impressive leap.

All that changed there was that they renamed the x70/80 series to be midrange for a generation in order to sell the bigger die top end cards later as yields improved. But the leap was there and it was big.

 

[TheRealDeal]

Kepler was actually a pretty big jump over Fermi, especially in the newer games at the time. 660Ti was often 50% faster than 560Ti. 670 often 50% faster than 570. The 680 was less so over the 580, but when you remember that the 680 was only the 'midrange' Kepler on a 300mm die compared to the 570mm die of the 570/580, it's still quite an impressive leap.

All that changed there was that they renamed the x70/80 series to be midrange for a generation in order to sell the bigger die top end cards later as yields improved. But the leap was there and it was big.

Tpu make the gtx670 35% faster at 1200p on release comparing to the gtx570.

https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_670/28.html

This was around the average for all the cards you mentioned over the other. It's more about average and not often as if it was often 50% faster then it had to be often around 20% faster to average out around 35%.

The 660ti had the biggest lead over the 560ti which was round about 40-45% which is decent.

 

[Freddie1980]

Oh don't get me wrong I don't expect anything faster than Fury non X from Polaris. Was just saying we had large leaps in performance before.

The benchmark to look out for is the Ashes of the Singularity a game built on DX12 from the ground up running on a video card designed for DX12 from the ground up. This will give us a indication of what the card is capable of, so I'm not expecting fantastic leaps for DX11 games over then the natural gains you get from an improved designed and more cores etc as DX12 is the future.