GTX 1080 unveiled; 9 teraflops

[Pemalite]

JEMC said:

Pemalite said:

The traces aren't really the problem. That is what the interposers are for.
However... Interposers are also built at 65nm and they do cost a fair chunk of coin, you could technically have two stacks of HBM to each interposer for a total of 8 if the Interposer was designed to accomodate it, but they aren't. It's one thing to do HBM version 2, doing an Interposer version 2 on a different fabrication process in order to accomidate multiple stacks is a different game entirely.

But then Costs will also blow out, only so much you can do per market segment, HBM is already expensive.

As for HBM's successor, AMD has just labelled it as "Next Gen memory". I would hazard a guess that as HBM is seen as "wide" AMD/JEDEC/Memory companies might go for a faster approach and take advantage of HBM's inherent advantages and drive up the speed.

As for the APU.
Anandtech at one point asked it's reader base what users wanted to see out of AMD and AMD was actually reading and responding to some of the comments... And without question everyone wanted a beefy APU with built-in GDDR5 or better Ram, AMD had actually done something similar with the 780 chipsets where they would bundle 64Mb-128Mb of DDR3 Ram for the motherboards graphics.
So AMD knows there is demand, at-least in the enthusiast community.

Now the APU in question isn't for your every day consumer, it is for the HPC market, Aka. Server grade stuff that already uses 16+ CPU cores and beefy GPU's for compute, it just happens to be that AMD sells both seperate one as Opteron the other as FireGL.
AMD themselves have already confirmed the existence of such a product, but what kind of hardware still remains to be seen, but keep in mind it is for the HPC market, it's likely to be extremely chunky, they don't typically do low-end stuff. ;)

Thus... Cooling isn't going to be an issue, these will likely come complete as a "compute add-in board" or an all-in-on motherboard with tons of ports for drives.

And even then, you don't need to go liquid cooling to cool 200-300w, a motherboard allows for a larger surface area than a GPU and GPU's have been cooling such wattages like that for years thanks to heat pipes, vapour chamber etc' designs.

Cooling is an issue when all those 200 or 300W of heat come from a single chip the size of an APU. There are no CPU heatsinks designed to tame such a monster, which is why I talked about AMD having to go with a CLC unit to cool such thing.

But it doesn't really matter, because such an APU for the HTC market won't go into retail for us to buy it.

That said, there was a rumor a couple of months ago of AMD working on a Bristol Ridge APU with 16CUs, that would put it almost on the same level as an XboxOne (because with DDR4, the bandwidth would be quite lower). For something more powerful, we'll have to wait for Zen and Pascal, well into next year.

*Edit*

I found the rumor: http://wccftech.com/amd-bristol-ridge-16-cu-apu/

The point is. You *wouldnt* be using a CPU heatsink. CLC cooling wouldn't be required.

This isn't just a "small" and hot chip, it would be a monolithic chip with a relatively large surface area, not unlike a GPU, that kinda removes the potential for concentrated heat.

And you are half right, HPC/Server/Professional gear does trickle down into consumer retail channels, I can jump onto multiple PC retailers here in Australia and buy up Xeon, Opteron, FireGL, Quadro, Tesla etc'. I would assume the same thing would happen over in the USA. - Price is another matter entirely though. :P

JEMC said:

Unfortunately, AMD themselves has said that they aren't targeting the high end market with their upcoming Polaris cards. The 1080 doesn't have competition and the only reason it has that price, it to leave room for the future 1080Ti (whenever it's released).

And no, none of the upcoming Polaris cards will have HBM. The high end Polaris 10 is rumored to have GDDR5X, like the GTX 1080, while the rest will have GDDR5.

 

Sort-of.
The competition to the Geforce 1080Ti is going to continue to be Fury untill Vega drops late this year or early next year.
Fury is unlikely to win on performance merits alone, so AMD may just discount it's price to stupidly low levels, wouldn't be surprised if the 400 series other than Polaris is entirely just rebadged parts from the last 4-5 years... And when Vega launches AMD rebadges all the current 400 series cards and supplements it with Vega in the high end for the 500 series.

AMD's hands are a little tied at the moment though, they need to follow the plan that was already set in motion, AMD's new GPU plans that they started to organize this year by letting the GPU segment become independent again, probably won't come to fruition for another few years.

KaosMike said:

@JEMC

Looks like you are mostly right about those rumors, however, I'm still thinking that HBM1 should fit well into laptop form factors, basically where it makes the most sense.  So perhaps all mobile variants could be using it going forward.

http://wccftech.com/amd-polaris-10-polaris-11-launch-event/

AMD confirmed that Polaris 10 is aimed at both mainstream desktop products and high-end notebook products which means that this chip is quite fast. The Polaris 10 could be similar to NVIDIA’s GM204 which serves both, a high-end mobility lineup and the mainstream desktop lineup while staying competitive in terms of performance and efficiency.

edit: It would be funny if AMD came out and said something like, "Hey, check out our new mid-range cards at 399 bucks a piece, oh and by the way, the absolute perf is 4/5 of a NV1080 and about same if not better perf/watt too"

-M

HBM1 isn't likely to be used in Polaris because from what I can tell... Polaris doesn't have the appropriate memory controller to make use of it if AMD's slides are anything to go by.

Also, the benefits of bandwidth becomes more important the higher in resolution you go, for laptops where the majority are topping out at 1080P and the majority sit at 720P... This isn't likely to be a defining factor, it would be just unnecessary wasted cost.
In devices where size is an important factor, then OEM's will likely continue to use Intel or AMD's APU's due to their inherent advantages.

Polaris isn't going to be AMD's fastest mobile chip either, that will continue to be the Radeon R9 M395X (Tonga), which has almost twice the hardware as mobile Polaris (Aka. Radeon R9 M490)
Majority of AMD's mobile line-up with the exception of Polaris will also be re-badged parts from the last 4-5 years.

What truly interests me at this point though... Is nVidia having it's chips at TSMC 16nm while AMD has it's at Global Foundries 14nm should give AMD a slight advantage as it's Gate pitch and SRAM Cells should have the edge.

JEMC said:

Mainstream is a vague concept for AMD, that has used in the past to talk about very different hardware and price points. For example, they used it with the 270X card, and a year later they used the term to talk about the Tonga chip that powers the 380 and 380X cards. Big difference in power and price points.

But it has never been top end.

The best we should expect from Polaris 10, unless AMD has been lying all this time (which seems unlikely), is a card close to GTX 1070 performance.

Mainstream shifted between the 200 and 300 series because the 300 series introduced another tier of cards.

If we were to consider Fury as Enthusiast, then the 390 would be high-end, that would make the 380 mainstream.
With the 200 series the 290 was Enthusiast, the 280 was high-end and the 270 was mainstream.
It's been like that since the 7000 series, AMD essentially adds a new high-end card which pushes the mainstream targeted card a notch higher.