Rumor: Polaris validation Failed Might Launch in October Now
Okay before I start on this news-item, I really need to state that this is based on a vague rumor, nothing has been confirmed or denied otherwise. Here's the story, some reports say Polaris 10 can't hit 850 MHz reliably and that availability will be pushed back to October. I sincerely hope the rumor is not true.
The news arrived through Nordic Hardware saying that AMD's partners "won't have any new cards to display at Computex and the only Polaris cards promoted to them from AMD are R9 390/390X performance class but for a mid-range price. Reports say Polaris 10 can't hit 850 MHz reliably, whereas Nvidia already demoed the 2000 MHz marker with their GPUs.
A board partner response seem to be "as of now, no information".
Rumor indicates that this info is backed by a rather large AMD partner, that there will be no Polaris-based video cards from AMD partners at Computex. That means no Polaris on Computex and no availability of the cards shortly thereafter (if this is true).
If true, man can AMD please catch a break ...
AMD Shows Polaris 10 and Polaris 11 ASIC Images - 05/02/2016 07:53 AM
Well .. a little bit as it isn't much. On a new webpage from AMD that are teasing the new Polaris architecture. It's highlighting features as 4th gen. Graphics CoreNext, 4K H.265 60 Hz game-streami...
AMD Polaris 10 GPU To Offer Near 980 Ti Performance For 299 USD? - 04/26/2016 10:03 AM
An interesting news item appeared on game debate yesterday, they claim that AMD held an event in Taiwan and has been showcasing it's upcoming Polaris GPUs. The interesting bit is that Polaris 10 woul...
New AMD roadmap gives more insight in polaris 10 and 11 - 04/21/2016 09:07 AM
A new roadmap appeared online from AMD, in this roadmap you can spot some plans for the future graphics cards based on the Polaris 10 and 11 GPU design. The roadmap also makes it pretty clear that the...
PlayStation 4K SoC Could Feature 2304 Stream processors based AMD Polaris GPU - 04/19/2016 04:57 PM
So the story goes like this, the upcoming upgraded PlayStation 4 is tagged under Codename NEO and is based upon an upgraded CPU, GPU, RAM. The custom SoC chip that empowers it all will get 8 CPU core...
AMD Polaris 11 in shows CompuBench has 1024 Shader processors - 04/11/2016 10:06 AM
This is a pretty intersting find, in the CompuBench database some device entries of AMD Polaris 11 (the entry level to mainstream SKU) has appeared. It's all a bit dodgy to dig out, but each graphi...
Posts: 31495
Joined: 2005-01-08
I'm on the pre-order waiting list at ocuk, but, I really doubt 1070 is gonna be £350. I'm thinking more like £400-450...
My guess is £349.
Posts: 870
Joined: 2016-05-12
No, I don't think that. What I said is that we don't have any concrete information from ANY of them yet. Hell, even the Fermi/Maxwell comparison is huge and they were in the same 28nm node, same for Tahiti/Fiji. You can do stuff by both compromising in some areas, AND get a better design, it's not one or the other. They have BOTH done it in the past.
Fermi and Maxwell weren't on the same node, as usual, not a clue what you're on about now.
You're not making any sense at all, Maxwell doesn't have to play catch up with Fiji (GCN1.2)... It's the other way around. All Pascal had to do was incremental improvements over Maxwell and scale it down thanks to new process.
GCN 1.3 needs to match Maxwell for geometry performance (doubling), match pascal for power efficiency AND die area/complexity AND be priced competitively (despite economies of scale being on NV's side).
Yes it would be naive to think that. If you didn't have such a hard-on for JHH you would probably be pushing this view, I certainly never said anything to that effect. I said Polaris cannot catch-up on all fronts while also maintaining advantages that directly correlate to reduced efficiency, increase chip complexity. Something's gotta give.
Maxwell indeed did make concessions to achieve what it achieved vs Kepler, similarly Kepler made concessions to achieve what it achieved vs Fermi.
I expect 1070 to be held back by bandwidth, 1080 to a lesser extent but can't really know until we gauge efficiency of the new IMC and compare to maxwell data
GloFo '14nm' LPP is not the same node as TSMC '16nm' FF+, the transistors have different characteristics; GloFo has that 7% advantage in gate feature size, TSMC has an advantage in leakage/transistor performance.
Either way, in order to clock as high as Maxwell/Pascal there would have to be architectural changes made; concessions made to shorten paths to allow for higher clocks. It will not, EVER, come for free.
don't know why this upsets you, the same would apply for nvidia, which I'm sure you'll have no trouble agreeing with. Just like my AotS results were meaningless until AMD posted the same damn thing

it's not like nvidia can burden Pascal with all the complexity and redundancy of GCN's control logic while STILL having smaller dies that are more power efficient, less dense and clocked higher than GCN. That would be a free lunch, and no such thing exists
Senior Member
Posts: 1126
Joined: 2009-04-29
https://twitter.com/AMD/status/731158531664171008

Senior Member
Posts: 6414
Joined: 2004-09-30
Polaris could undercut Pascal if it launches 18 May and have same or better performance.
Posts: 31495
Joined: 2005-01-08
No, I don't think that. What I said is that we don't have any concrete information from ANY of them yet. Hell, even the Fermi/Maxwell comparison is huge and they were in the same 28nm node, same for Tahiti/Fiji. You can do stuff by both compromising in some areas, AND get a better design, it's not one or the other. They have BOTH done it in the past.
It's also quite naïve to believe that Pascal only has "benefits", as if it is a magical design or something. Look at the sentence you wrote, and then remember that they are both on practically the same node, with their current products in all segments being around the 5-10% difference on average.
Won't it be memory bandwidth starved compared to the Ti? The Ti in general seems like a much more "balanced" design.
You could say the same about 780ti vs 970. Yet 970 did perfectly fine against 780ti. Mind, I'm talking about up to 1440p. Neither 980ti or the 1070 will be powerful enough for 4k.