Lots of "coulds" in that article, but it's nice to hear that there's something interesting on the horizon.

btarunrImagine R9 390-like performance at half its power-draw.

I think most people are hoping for more performance, not necessarily more efficiency. Unless they price the r9 480 on par with the current 390 offerings and replace them I don't see a point in such an SKU.

xenocideI think most people are hoping for more performance, not necessarily more efficiency. Unless they price the r9 480 on par with the current 390 offerings and replace them I don't see a point in such an SKU.

If the nomenclature holds, 480 should be aimed at the price of the current 380 segment, not the 390 one. Meaning upon release it should be priced above 380, but under 390, so I'd say around 300 euro?

Polaris 10 GPU “67DF:C4” Specifications

• 14nm FinFET, GlobalFoundries
• Diffused in USA (New York state)
• Assembled in Taiwan
• 2304 Cores (silicon: 2560)
• 36 Enabled Core Clusters (silicon: 40)
• New GPU Architecture (not GCN “1.5”)
• 256-bit Memory Controller
• 8GB GDDR5/GDDR5X (when available)

That's interesting. Mid-tier with 8GB. If true that's going to be 8GB in the $300-$400 range. It would also mean no high-end replacement until HBM2 Vega.

They could also offer the GDDR5X as the "X" variant 480X.

The final nail in Fiji's coffin?

If the 480 has 390 performance then that's disappointing. But then again, they plugged Polaris on "performance per watt." The same thing they did with Fiji. The moment AMD started that "per watt" talk I knew not to be excited with Polaris. Code word for negligible performance boost but better power consumption.

If that is a selling point for you, great. But consider this. 390 performance was obtainable by the 290/X which is how many years old now?

Will that be enough to take on Pascal? Ehh...not expecting much at this point. Although if rumors are true about Pascal still not having decent compute, it may not matter.

I'd say R9 490 and 490X will come later then, as semi big GPU (about 350-450mm²) and replace Fury / Fury X in performance, + maybe 20-30%. After that Vega (maximum size 14nm GPU) will be delivered which will be Fury X2, with HBM2 memory (the direct successor to Fury X), and a bigger boost in performance. Just some thoughts.

eagerly waiting for the cards, hopefully R9-480 will make a decent 1440p gpu.

NC37If the 480 has 390 performance then that's disappointing.

If an estimated ~130w card has the same performance as a 275w card, I wouldn't call that disappointing. If we speculate that the 490 is just a 480 scaled up to the ~250w mark, that'd be some really decent performance.

They are actually making entire lineup of brand new GPU's? Has hell frozen over? I bet you entire R9-400 series will be just a die shrink of a R9-300 series and the R9 Fury 2 will be a die shrink of a Fiji. They are doing the rebranding shit for too long to all of a sudden entirely drop it now.

AMD could keep the clock speeds relatively low, with 800-1050 MHz GPU clocks. Imagine R9 390-like performance at half its power-draw.

I'd rather have double the performance at the power draw of a R9-390, no the other way around... People buy graphic cards to play games at high framerate, not to "save electricity". What a weird backwards logic.

wccftech.com/amd-radeon-r9-480-470-polaris-10-polaris-11/

AMD will finally have some competitive laptop GPUs, nice.

RejZoRI'd rather have double the performance at the power draw of a R9-390, no the other way around.

And you will get it, but you will have to pay $$$.

xenocideI think most people are hoping for more performance, not necessarily more efficiency. Unless they price the r9 480 on par with the current 390 offerings and replace them I don't see a point in such an SKU.

And they're getting more performance. R9 480 succeeds the R9 380, not the R9 390. Logically, they'll price it on-par with R9 380's launch price.

FluffmeisterThe final nail in Fiji's coffin?

In Hawaii's, yes.

KananI'd say R9 490 and 490X will come later then, as semi big GPU (about 350-450mm²) and replace Fury / Fury X in performance...

AMD are on record saying that the next (Polaris) series will consist of two GPUs. By the looks of it, both the lower Polaris 11 (128-bit) and Polaris 10 (256-bit) are launching as salvage parts first which provides 5 SKUs ( two from Polaris 11, three from Polaris 10). The Fiji replacement being Vega in Q1 2017. Looking at it logically, a 480 should have around the performance of a 390, or you'd be looking at a degree of stagnation considering these parts aren't renamed silicon.

RejZoRThey are actually making entire lineup of brand new GPU's? Has hell frozen over? I bet you entire R9-400 series will be just a die shrink of a R9-300 series and the R9 Fury 2 will be a die shrink of a Fiji.

Unlikely. Even if the architecture is just tweaked, AMD will certainly update the silicon logic blocks (VCE, UVD etc) and there is no way in hell that AMD die shrink Pitcairn/Curacao/Trinidad (R9 370/370X)- the GPU is already in its fifth year of service and simply lacks too many features.

RejZoRThey are doing the rebranding shit for too long to all of a sudden entirely drop it now.

That would be death for AMD. I'd expect some low end rebrands - mainly for OEM and mobile parts, simply because AMD hasn't announced successors to Oland and Cape Verde, but otherwise the very least you could expect is for AMD and Synopsys to lay out the logic from Tonga and update the I/O for the new GPUs.

xviIf an estimated ~130w card has the same performance as a 275w card, I wouldn't call that disappointing. If we speculate that the 490 is just a 480 scaled up to the ~250w mark, that'd be some really decent performance.

Sounds about right based on previous history. The core of Fiji was effectively Tonga doubled in size, and no reason to think that AMD and Synopsys wouldn't continue the mix-and-match logic to keep R&D, development cost, and time-to-market manageable.

RejZoRI'd rather have double the performance at the power draw of a R9-390, no the other way around... People buy graphic cards to play games at high framerate, not to "save electricity". What a weird backwards logic.

That is definitely true for high end cards, but I'm more interested in what happens at the low end of things, say in low profile cards using only socket power. With that kind of effeciency you could theoretically build some very nice budget gaming machines from SFF i5's.

Hmm, interesting either way with them launching some lower parts first and keeping the bigger brothers in the shadows. Curious if they will actually use GDDR5X on these or upper models of the same chip.

Let's just hope that these lead to lower prices. There hasn't been that 200€ sweet spot in a long time like HD4850 when it released. Some serious gaming performance for a smart price.

Except HD5850, HD6950 and HD7850/7950, there really weren't any great priced graphic cards from AMD...

Agree with @Xzibit. Looks like no new cards at the top end until 2017. Interesting to see some summer benchmarks of the new cards.

Why can't AMD tick tock like Intel and die shrink Fiji?

siluro818If the nomenclature holds, 480 should be aimed at the price of the current 380 segment, not the 390 one. Meaning upon release it should be priced above 380, but under 390, so I'd say around 300 euro?

The thing is, proper 4k gaming needs about 50% more computing power than Fury X. If we don't get that, shaving 30-40W off power consumption isn't going to do much for serious gamers. And serious gamers are those that buy 380/480 or better cards.

Power consumption and HDMI Adaptive-Sync and HDR readiness might be the big marketing push for Polaris. AMD will bank on those as well as more games/dev taking advantage of async in DX12 for that performance push.

the54thvoidWhy can't AMD tick tock like Intel and die shrink Fiji?

Probably too much invested at this stage. R&D would have been prodigious, and I'm guessing that AMD signed up for a certain quantity of components. Since HBM(1) hasn't exactly has widespread adoption, it maybe a case of having to commit to a minimum contract prior to Hynix commencing production - the same might also apply to Amkor's assembly contract.
I'd think that if AMD could have die shrunk Fiji and had access to HBM2 early it might have been enticing. Fiji undoubtedly - after R&D, assembly, and component cost - definitely wouldn't aid AMD's average selling price which presently sits just under $29 per unit, but without access to discrete mobile, the only option was hitch higher price tags to the Duo and S9300 X2, and hope it gains some appeal against a pretty well balanced Firepro W9100 (whose own life is will probably be extended with the 32GB version).

bugThe thing is, proper 4k gaming needs about 50% more computing power than Fury X. If we don't get that, shaving 30-40W off power consumption isn't going to do much for serious gamers. And serious gamers are those that buy 380/480 or better cards.

one 290X handles 1440P, the 295x2 decimates everything, even at 4K if the drivers work.

The 480 looks to be on par with a 290X, so good up to 1440P, IF the 490 is a double 480 it will be in the vicinity of the 295x2, and able to handle 4K without the added complexity of CF

Brusfantometone 290X handles 1440P, the 295x2 decimates everything, even at 4K if the drivers work.

The 480 looks to be on par with a 290X, so good up to 1440P, IF the 490 is a double 480 it will be in the vicinity of the 295x2, and able to handle 4K without the added complexity of CF

If the 490 is a double 480, it will still have all the driver complexity issues of CF. Unless you mean, 490 as being double the power in a single GPU.

And what 490?

FluffmeisterThe final nail in Fiji's coffin?

Fiji replacement, Vega, is expected in Q1 2017.
Anyway, what's wrong with Fury Nano, that costs roughly as 980 yet is faster, smaller?

RejZoRExcept HD5850, HD6950 and HD7850/7950, there really weren't any great priced graphic cards from AMD...

4850 was.

the54thvoid...driver complexity issues of CF..

AMD is pushing for "multi-gpu aware" game development. As far as I understand, it would mean devs to keep in mind "multi-gpu"-ness, which would also simplify drivers (no need to present 2 cards as single one)