No HBM2 memory this year from AMD and quite possibly nVidia either.

[Evildeffy]

Hey folks...

A short while ago this came up on the rumour mill (3 days ago really):
Nvidia Pascal GTX 1080 8GB, GTX 1070 Launching This Summer – To Debut In April At GTC 2016

This outlined the rumour that it will be equipped with GDDR5X instead of HBM2 for their first tier of cards.
Which of course was a bit puzzling due to the fact that it was hyped around having HBM2 but we treated it as rumour.

Now the following as surfaced:
AMD: Polaris Is All About Sweet Performance Per Dollar

Which may look like an ordinary marketing text along with technical details but the second-to-last paragraph is very important to read...

Finally, Raja confirmed that indeed HBM2 will be coming next year with Vega but not before then. A decision that was made because HBM2 is simply not ready for 2016 from an economic standpoint Raja explained. It wasn’t going to allow him and his team to make a compelling performance per dollar product in 2016 due to the high costs associated with it in its early ramp. Thankfully however, we’re not going to have to wait until 2017 for Polaris.

In which it shows that AMD will NOT be making use of HBM2 with their cards as the head of Radeon Technologies Group confirms it.
This, unlike nVidia's rumour of not using HBM2, is confirmed from AMD's side but it is not stated what is to be used instead.

It's possible that when AMD is not willing or capable of sourcing HBM2 that nVidia will not be either.. though untill nVidia confirms the same news we cannot ever blindly believe that to be even remotely true.

I don't care which brand of GPU I'd go for as long as it's the most powerful when I upgrade... be that nVidia or AMD.
And that's how you all should approach this from your own perspectives rather than being a blind sheep.

Leave your thoughts and speculation but keep the fanboyism out of this thread... enjoy!

[Bigvizz]

Still reserving my judgement for release, mindless speculation, is mindless. Time will tell.

[Artorius]

Should we expect HBM1 or GDDR5X from AMD?

[Evildeffy]

Still reserving my judgement for release, mindless speculation, is mindless. Time will tell.

Your choice to abstain and that's fine.. but it's the opposite of what I asked and to state that mindless speculation is mindless is ... well USELESS as the word speculation means exactly that, speculation.

I'm stating this because your statement, barring it being poorly stated, is also quite demeaning because you're essentially saying that anyone who speculates is a mindless drone and that is quite offensive actually.

It would honestly have been better if you hadn't replied at all with your current post.

- - - Updated - - -

Should we expect HBM1 or GDDR5X from AMD?

Unknown so far... both are possible, as well as basic GDDR5.

[Deleted]

A bit more info right here.

Curious about the HBM2 inclusion in Vega on the roadmap and what that means for Polaris? Though he didn’t say it outright, it appears that Polaris will be using HBM1, leaving me to wonder about the memory capacity limitations inherent in that. Has AMD found a way to get past the 4GB barrier? We are trying to figure that out for sure.

Why is Polaris going to use HBM1? Raja pointed towards the extreme cost and expense of building the HBM ecosystem prepping the pipeline for the new memory technology as the culprit and AMD obviously wants to recoup some of that cost with another generation of GPU usage.

[Bigvizz]

Your choice to abstain and that's fine.. but it's the opposite of what I asked and to state that mindless speculation is mindless is ... well USELESS as the word speculation means exactly that, speculation.

I'm stating this because your statement, barring it being poorly stated, is also quite demeaning because you're essentially saying that anyone who speculates is a mindless drone and that is quite offensive actually.

It would honestly have been better if you hadn't replied at all with your current post.

Lawls, you get bothered easily don't you, you could've just ignored my comment just as easily. To put it simply it's my opinion and every has one right?

[Evildeffy]

Lawls, you get bothered easily don't you, you could've just ignored my comment just as easily. To put it simply it's my opinion and every has one right?

Everyone is entitled to their opinion .. nothing wrong with that as long as it's not in a demeaning "I AM HOLIER THAN THOU" attitude.
That was my point which flew right above your head apparently.

- - - Updated - - -

A bit more info right here.

Cool beans... But I do indeed wonder about some of the things mentioned because as it stands 4GB will be reaching limits in 1440p if game devs continue the way it goes even though that 4GB limit is generally not reached right now.

[Bigvizz]

Everyone is entitled to their opinion .. nothing wrong with that as long as it's not in a demeaning "I AM HOLIER THAN THOU" attitude.
That was my point which flew right above your head apparently.

I suppose that's your opinion, but you're the one getting your panties in a ruffle over a single comment.

[Nevcairiel]

If they can't overcome the HBM1 4GB limitation, they better use something else. 4GB is not a competitive model in 2016 (wasn't in 2015 either, but thats another story)
I was kinda hoping for HBM2 from NVIDIA in Summer, but it doesn't look like they are going to make that. GDDR5X is already a decent boost over normal GDDR5 (and competitive to HBM1), so assuming that is used, its probably going to be fine, just not perfect.

[Deleted]

Cool beans... But I do indeed wonder about some of the things mentioned because as it stands 4GB will be reaching limits in 1440p if game devs continue the way it goes even though that 4GB limit is generally not reached right now.

If they can't overcome the HBM1 4GB limitation, they better use something else. 4GB is not a competitive model in 2016 (wasn't in 2015 either, but thats another story)

4 GB is enough. The Fury(x) use dynamic Vram, they use the system RAM when the 4 GB Vram is fully used. [H] did some testing on this.

The AMD Radeon R9 Fury X kind of backs that statement up since it was able to allocate dynamic VRAM for extra VRAM past its 4GB of dedicated VRAM capacity. We saw up to a 4GB utilization of dynamic VRAM. That allowed the Fury X to keep its 4GB of dedicated VRAM maxed out and then use system RAM for extra storage. In our testing, this did not appear to negatively impact performance. At least we didn't notice anything in terms of choppy framerates or "micro-stutter." The Fury X seems to be using the dynamic VRAM as a cache rather than a direct pool of instant VRAM. This would make sense since it did not cause a performance drain and obviously system RAM is a lot slower than local HBM on the Fury X. If you remember a good while ago that AMD was making claims to this effect, but this is the first time we have actually been able to show results in real world gaming. It is awesome to see some actual validation of these statements a year later. This is what AMD said about this in June of 2015.

Note that HBM and GDDR5 memory sized can’t be directly compared. Think of it like comparing an SSD’s capacity to a mechanical hard drive’s capacity. As long as both capacities are sufficient to hold local data sets, much higher performance can be achieved with HBM, and AMD is hand tuning games to ensure that 4GB will not hold back Fiji’s performance. Note that the graphics driver controls memory allocation, so its incorrect to assume that Game X needs Memory Y. Memory compression, buffer allocations, and caching architectures all impact a game’s memory footprint, and we are tuning to ensure 4GB will always be sufficient for 4K gaming. Main point being that HBM can be thought of as a giant embedded cache, and is not directly comparable to GDDR5 sizes.

So if HBM 1.0 comes to Polaris it will also use dynamic Vram.

[Artorius]

Using system/CPU RAM for video makes it way slower than it would be using the card's VRAM.

[Vash The Stampede]

HBM is so expensive and new that I didn't expect HBM to be a thing this year.

[Remilia]

More than likely we're going to just get normal GDDR5 from both camps. HBM2 is expensive and supposedly Samsung has started production? Not entirely sure how that's going. AMD has the memory controller for it already worked out in Fury but Polaris 10/11 are meant to be low mid range where price is well... not absurd.

As for GDDR5X, it hasn't even started mass production yet which is slated for Summer this year.

Take it as what you will, but.
http://www.pcper.com/news/Graphics-C...-HBM2-and-more

Curious about the HBM2 inclusion in Vega on the roadmap and what that means for Polaris? Though he didn’t say it outright, it appears that Polaris will be using HBM1, leaving me to wonder about the memory capacity limitations inherent in that. Has AMD found a way to get past the 4GB barrier? We are trying to figure that out for sure.

Won't surprise me if that's true. At 1920x1080 4GB hasn't been shown to be a bottleneck yet in reality.

[DeltrusDisc]

Am I remembering incorrectly? Wasn't HBM(1) on at least a couple cards from AMD's more previous generation?

[Lathais]

Am I remembering incorrectly? Wasn't HBM(1) on at least a couple cards from AMD's more previous generation?

Pretty sure it was just the 390X, Fury, Fury X and Fury Nano, you know, the expensive cards. So the lower end cards will not have it due to expense.

[Evildeffy]

Pretty sure it was just the 390X, Fury, Fury X and Fury Nano, you know, the expensive cards. So the lower end cards will not have it due to expense.

No 390X on that list of HBM cards, just the Furies and Nano cards.

[Lathais]

No 390X on that list of HBM cards, just the Furies and Nano cards.

ok, I remembered reading this:
http://wccftech.com/amd-fiji-feautre...er-than-gddr5/

I guess they never released it though.

[Artorius]

Only the Fiji XT has HBM, so furys and the nano.

[Faldric]

I could kick my ass for waiting to buy a new card. The closer we get to the release of pascal/polaris the worse the news get.

I bet once they are here, they will be a mere 5-10% faster than current gen cards and thats it. I already see myself buying an older card because the performance difference will be so small that it doesn't justify the bigger price.

[Artorius]

They'll still be 14nm FinFet though, so you can at least expect way better performance/power than previous cards.