AMD: Pirate Islands (R* 3** series) Speculation/Rumor Thread
- Thread starter iMacmatician
- Start date
-
- Tags
- amd
I think there's 99% of luck, nothig will be in 20nm this year... . so i can imagine, we will only see 28nm (SHP from Samsung Glofo ? ) for the start of 2015. then i can imagine a die shrink is possible.. 16nm FF from TSMC seems again delayed to late late 2016 ( at best ), so maybe, but im really not optimist for that, a die shrink to 14nm FF from Samsung / GlobalFundries at the end of 2015.. early 2016...
Everyone, include Qualcomm have now leave the 16nm FF TSMC boat ( we could rename it Titanic at this point ), it only stay Nvidia on theory for the 16nm FF in TSMC ( by lack of information, we can only conclude their TSMC contract is still alive for 16nm, or maybe not ), all contract have been closed with TSMC and re allocated to Samsung..
Last edited:
iMacmatician
Regular
From Videocardz: "Radeon Rx 400 series are codenamed Arctic Islands."
This is the first time I've heard of "Arctic Islands."
It originates from SweClockers, they've apparently aquired some new roadmaps from AMD somehow. They've also posted new AMD CPU & APU stuff & schedules
If it's real, those future GPUs had better generate less heat than Hawaii, lest they become the butts of many jokes.
A1xLLcqAgt0qc2RyMz0y
Veteran
@Alexko
Well, heat is the same as power draw, which, all factors being the same, is the same as performance. So if you want a fast chip, it's going to draw power...
It's just a code name, assuming it's even accurate to begin with. Trolls gonna troll, you know? Not much that can be done about that, heh.
Well, heat is the same as power draw, which, all factors being the same, is the same as performance. So if you want a fast chip, it's going to draw power...
It's just a code name, assuming it's even accurate to begin with. Trolls gonna troll, you know? Not much that can be done about that, heh.
@Alexko
Well, heat is the same as power draw, which, all factors being the same, is the same as performance. So if you want a fast chip, it's going to draw power...
It's just a code name, assuming it's even accurate to begin with. Trolls gonna troll, you know? Not much that can be done about that, heh.
All factors being the same, yes, but the key factor is power-efficiency. I said heat to make the point about Arctic Islands clearer.
Not to mention that "Islands" family names don't seem to correlate to graphics IP levels very well.
Considering the power ceiling is the same for all GPU add-in boards, efficiency tends to be roughly similar between manufacturers. Also, high-end board gamers don't really care all that much about efficiency anyhow, they want raw power, and if your product delivers that then they'll buy it. No GPU can be hugely more inefficient than its competitors due to the power ceiling anyway (unless your name's Geforce GTX 4/580 that is... Then you can secretly pull 500W without telling anyone, lol.)
For current products, AMD consumes 60% more power for similar performance? (Say, GTX960 vs R9 285). That's not 'roughly similar.' And I'm sure AMD is very hard at work to make that go away. It makes them an impossible choice for laptops.
(http://www.hardware.fr/articles/932-8/consommation-efficacite-energetique.html)
AMD probably pushed thing back once they realized they would be on 28nm for another cycle. They probably didn't have the resources nvidia did to do nearly as much on 28nm again. I expect things to even out on efficiency when both reach finfets. Even Tonga was not a significant design change from their previous gcn chips besides the colour compression. GCN updates have been very small compared to what nvidia did going from kepler to maxwell.
That's a different topic. The point is that power efficiency matter a great deal more that you suggested.
I was hoping that they would go global foundries 20nm or something. However if its faster than the gtx 980 and not too loud i might get it anyway as my 7850 is getting olx
No, it most certainly is not. It's a fact of reality. Different generations have different properties, including power efficiency and so on, comparing a newer and an older gen GPU is not going to be entirely accurate. What I said is true, in broad terms, which obviously means taking generational differences into consideration.
If that's the case, it's a point which you've yet to prove.
For example, is nobody buying AMD R290 cards at all, because their power efficiency is worse than NV 9x0? If that's the case it would be news to me at least.