AMD publishes Polaris Architecture microsite

[HEXUS]

Offers a tech overview - and some teasing images of the first Polaris GPUs.

Read more.

[darcotech]

Can't wait to see what we will get. Hope for low power/high perfomance chips. especially for notebooks.
I put all my purchases on hold till the new AMD/NVidia chips come.

What about you guys?

[SlickR]

Looks promising, I'm personally expecting about 20% improvement in graphics on the lower level for cheaper prices. For example a 270x replacement to be 20% faster and come for $140

[chj]

I'm expecting a nice boost in performance per watt. But with vega supposedly bringing another significant boost in such a short time frame i'm holding out for that.

[idris]

Can't wait to see the new AMD products! POLARIS/ZEN/VEGA along with the nest gen of consoles all with AMD inside.

[kalniel]

I'm expecting a nice boost in performance per watt. But with vega supposedly bringing another significant boost in such a short time frame i'm holding out for that.

Fairly different markets I think. Vega is high end, Polaris is mid-range volume parts. You'll already know which market you're in

[idris]

I'm expecting a nice boost in performance per watt. But with vega supposedly bringing another significant boost in such a short time frame i'm holding out for that.

Fairly different markets I think. Vega is high end, Polaris is mid-range volume parts. You'll already know which market you're in

Depends on what your take is regarding "mid-range" / "high-end" with the new 14nm process...

"According to the information we have about the 14nm LPP process, and based on transistor density increase, a 232mm² GPU would be roughly equivalent to a 464mm² 28nm processor – at the same TDP levels. Since we already know that AMD is going to be focusing not just on performance but power efficiency as well – this number could be be much higher, in fact we will discuss the number AMD is using below. We can however safely say that this die is more than capable of meeting the ‘minimum VR spec that AMD promises. The more astute of our readers would notice one inconsistency with this information. A 232mm² does not fit with the positioning of the small Polaris die, which according to our estimates should have been around 150mm² and doesn’t fit well with our image of big Polaris which should have been near the 300mm² mark. If you were to take a look at the AMD slide given above and calculate the die shrinks associated with the 14nm jump – the number would be exactly 2.5. If you remember, Raja Koduri has previously tweeted about Polaris being “2.5 times brighter” in the past – so the number 2.5 times is one AMD is throwing around alot.
A 232mm² die multiplied by 2.5 times gives us the performance range of a 28nm 580mm² chip (at the same TDP level) which would be pretty damn impressive. We do not know what kind of power efficiency mix AMD is playing with here so it would be pointless to speculate further about the exact performance levels at this point. It would however, be prudent to keep in mind the difference between a 28nm and 14nm die.

[CAT-THE-FIFTH]

I hope AMD is releasing these soon. It looks like Nvidia is going to unveil the GTX1080 and GTX1070 at the BF5 event this week,and one website has said they are getting review samples soon.

The latest leak says Polaris 10 is around R9 390X level performance,but even then get the chips out AMD!!

[jigger]

Depends on what your take is regarding "mid-range" / "high-end" with the new 14nm process...

"According to the information we have about the 14nm LPP process, and based on transistor density increase, a 232mm² GPU would be roughly equivalent to a 464mm² 28nm processor – at the same TDP levels. Since we already know that AMD is going to be focusing not just on performance but power efficiency as well – this number could be be much higher, in fact we will discuss the number AMD is using below. We can however safely say that this die is more than capable of meeting the ‘minimum VR spec that AMD promises. The more astute of our readers would notice one inconsistency with this information. A 232mm² does not fit with the positioning of the small Polaris die, which according to our estimates should have been around 150mm² and doesn’t fit well with our image of big Polaris which should have been near the 300mm² mark. If you were to take a look at the AMD slide given above and calculate the die shrinks associated with the 14nm jump – the number would be exactly 2.5. If you remember, Raja Koduri has previously tweeted about Polaris being “2.5 times brighter” in the past – so the number 2.5 times is one AMD is throwing around alot.
A 232mm² die multiplied by 2.5 times gives us the performance range of a 28nm 580mm² chip (at the same TDP level) which would be pretty damn impressive. We do not know what kind of power efficiency mix AMD is playing with here so it would be pointless to speculate further about the exact performance levels at this point. It would however, be prudent to keep in mind the difference between a 28nm and 14nm die.

Possibly, but AMD's power gating could hamper the chips performance. I hope we don't don't suffer hard TDP limits like whats on Nvidia cards.

[kalniel]

Not sure what the quoted information has to do with it - AMD have already revised the perf/W down to 2x, and die size between different node processes are really quite inaccurate measures of performance.

However everything AMD have been saying about affordable parts and the console rumours suggest they're going for mid-range with Polaris, saving Vega for high-end (it's also very unlikely that AMD would cannibalise their own sales by announcing a replacement product before they've even released the first one.)

Performance predictions are another matter - my 'mid-range' and 'high-end' comments related to market segment, not performance compared to current products. It's quite likely that the new mid-range equals or even perhaps beat the previous generation's high-end - it would hardly be the first time that had happened

[CAT-THE-FIFTH]

Possibly, but AMD's power gating could hamper the chips performance. I hope we don't don't suffer hard TDP limits like whats on Nvidia cards.

I am more worried if it is true,Nvidia will be launching Pascal in June with review samples already going out now.

If that is the case they have beaten AMD to the punch,at least with the DIY market with a larger part,and it would be plain embarassing especially after they demoed parts 4 months ago and in March. AMD has always been first launching cards on a new node for the last decade.

I hope it is not another classic Global Foundries feck up - they fecked up Llano,Bulldozer and Kaveri launches by between six months to a year,and none of the chips could hit the predicted clockspeeds at launch,IIRC.

Edit!!

I really hope at least they have got decent OEM traction.

[jigger]

I think the OEM market will lap up Polaris cards. Nvidia's Pascal chips have a lot of questions hanging over them and I've got a feeling the new chips might still be very DX11 focussed rehash of the failed 20nm architecture, while this will be AMD third or forth generation low level API design. I think the June time frame for a massive Nvidia chips runs a real risk of AMD crushing Nvidia in Vulkan and DX12 and on price-performance.

[kalniel]

I think the June time frame for a massive Nvidia chips runs a real risk of AMD crushing Nvidia in Vulkan and DX12 and on price-performance.

Price, absolutely. Performance.. hard to see how - it looks like AMD will be close to, but not clearly beating, nVidia's current top products in DX12/Vulkan, if nVidia improve on their current offering as they're expected to then they should have an advantage, possibly quite sizeable. But you'll pay for it.

[jigger]

I'm not sure TBH. The Mantle and DX12 announcements seemed to take Nvidia completely by surprise and I remember Nvidia talking about adding CPU processing to graphics cards to deal with API overheads. Nvidia's answer to Mantle was to very hasitly push out adaptive sync hardware way before it was ready. That said Nvidia probably have enough money to throw at most problems and make them go away, but June just seems 6 months too early for hardware to deal with those problems.

[lumireleon]

One thing is for sure the GTX981Ti and Fury-X are beasts that can take any current workload like a cup of tea but the problem is efficiency, this time round I will not be looking at extra FPS or OpenCL benchmarks than the above cards but less wattage per fps. Its time for GTX960 level performance to be under 100w at full load.

[kalniel]

One thing is for sure the GTX981Ti and Fury-X are beasts that can take any current workload like a cup of tea but the problem is efficiency

I think you're wrong on both statements There are current workloads that they don't deal with - low latency 4k gaming and VR for example. That's here now, not something in the future. Likewise efficiency isn't a problem for the market they are aimed at - both cards run within perfectly reasonable high end thermal and power envelopes.