AMD are 'actively working on the GDDR 6 standard for their upcoming graphics cards'

[Master Disaster]

Usual caveat, WCCF so take with salt although this time they claim it's been confirmed with AMD staff...

 

So apparently some AMD engineers have been posting info on their LinkedIn profiles stating they are currently designing a GDDR 6 memory controller.

Quote

A rumor recently surfaced involving the LinkedIn page of an AMD technical engineer that listed the company working on a GDDR6 memory controller. I reached out to some people and can confirm that AMD is indeed working on the GDDR6 standard and will be actively using it in future graphics cards. As for the question that I am sure everyone will ask next, no AMD will still be sticking to HBM2 for its high end next generation graphics cards in 2018 (aka Navi).

 

The leak originally stemmed from a picture of a LinkedIn profile showing an AMD engineer listing the GDDR6 memory controller technology in his portfolio. This is usually a fairly obvious way of confirming a leak but the profile in question was nowhere to be found. This is why I decided to reach out to sources familiar with the matter myself and can confirm that AMD is indeed working on GDDR6 memory technology and will be adopting it.

Ok so I suppose you're all wondering if that means Navi won't be getting HBM2, well rest assured AMD will be sticking with HBM for it's flagship line of cards. Sources state manufacturers will have access to GDDR 6 as early as Q1 2018.

Quote

The next obvious question becomes when and where will we see it used, to that, the only reply I got was that AMD will still be sticking with HBM technology in high end graphics cards in 2018. Samsung, Micron and SK Hynix all have roadmaps that show their GPU SKUs rolling out by the end of 2017 or early 2018. In either case it looks like video card manufacturers will have access to the incredibly fast memory standard from Q1 2018. Since pricing will almost certainly be expensive in the beginning and it remains to be seen just how it fares in comparison with HBM.

 

Now AMD has previously teased that Navi will next-gen HBM memory, so it remains to be seen whether they are talking about HBM2 or the elusive HBM3 standard (which should be still faster than the GDDR6 standard) Edit: they are talking about HBM2. In any case, we do know for sure the company will be sticking with HBM technology for all its high end graphics cards indicating that either they will not be rolling out any GDDR6 based cards in 2018, or keeping it limited to mid-end or the professional side of things.

Micron, Samsung & SK Hynix have confirmed they are working on next gen GDDR modules, expected to come in 16GB form with a transfer rate of 16Gbps, bandwidth of 64Gbps per chip and will run at 1.35v

Quote

Samsung, Micron and SK Hynix have all officially stated that they will be producing the fastest and lowest-power DRAM for next generation products.  Samsung has currently listed a 16Gb GDDR6 DRAM in their portfolio but that can be expanded upon in the future when production hits full swing. With a transfer rate of 16Gbps, the DRAM will be able to pump out 64 GB/s bandwidth (per chip). The memory operates at just 1.35V.

 

Compared to current generation GDDR5 DRAM, we are looking at both, increased bandwidth and transfer speeds (8Gbps vs 16 Gbps) at lower power consumption (1.5V vs 1.35V). The specifications can easily be compared to current DRAM standards. We know that GDDR5 can reach up to 9 Gbps as NVIDIA launched their GTX 1060 in faster memory variants. That’s the practical limit of where the memory architecture can go. Then we have GDDR5X, a refined implementation of GDDR5 that is better in every aspect.

 

The GDDR5X memory operates at much faster speeds and has practically hit 16 Gbps as confirmed by Micron themselves. While GDDR5X can hit same speeds as GDDR6, the latter comes with better optimizations and higher densities. – and Samsung it claiming it to be an upgrade over G5X. We are looking at speeds of 12-14 Gbps becoming standard in the graphics industry while 16 Gbps will ship out in the high performance sector. There’s also up to 32 Gb density support while GDDR5/X max out at 16 Gb.

Quote

From Samsung's Website:

Samsung 16Gb GDDR6 Memory – The fastest and lowest-power DRAM for next generation, graphics-intensive applications. It processes images and video at 16Gbps with 64GB/s data I/O bandwidth, which is equivalent to transferring approximately 12 full-HD DVDs (5GB equivalent) per second. The new DRAM can operate at 1.35 volts, offering further advantages over today’s graphics memory that uses 1.5V at only 8Gbps

https://wccftech.com/amd-gddr6-memory-upcoming-graphics-cards/

 

Seems pretty legit, it's no secret GDDR 6 was coming but now we have 'proof' from a second in the chain source that they are also getting ready for it.

 

It's the usual doubling from previous gen stuff (ignoring GDDR 5X) and should be a pretty nice upgrade when we get hands on.

[WereCat]

The part of reason for AMD using HBM is the significantly lower power requirements compared to GDDR5 / GDDR5X.

They would have to improve the GPU efficiency significantly for Navi to ditch HBM which will not happen.

So it makes only sense to use GDDR6 for the lower cost Polaris equivalent of the RX 6xx lineup as I dont see HBM pricing and availability getting much better in 2018 for it to make sense to use on a lower-end graphics cards.

[samcool55]

Good, whatever the replacement for Polaris is, it's likely GDDR6 will cause the biggest performance bump.

RX 580's are basically bottlenecked by ram so unless they improve their compression stuff a lot, faster vram is the best option to improve performance.

[lishde]

I want hbm 2 on a 1050 ti. Not cost effective but it would be tiny and use even less power, 60 watt passive cooled.

 

[captain_to_fire]

35 minutes ago, Master Disaster said:

Usual caveat, WCCF so take with salt although this time they claim it's been confirmed with AMD staff...

I thought this forum has came into conclusion that WCCF is "fake tech news" just like Bloomberg?

Edited December 4, 2017 by hey_yo_

[Master Disaster]

Just now, hey_yo_ said:

I thought this forum has came into conclusion that WCCF is "fake tech news"? 

They've had more successes than failures this year for sure

[lishde]

Its not that there cnn, they just don't vet everything leaked to them as much as they should

 

[Taf the Ghost]

15 minutes ago, lishde said:

Its not that there cnn, they just don't vet everything leaked to them as much as they should

 

I wouldn't insult WCCF by putting them on the low-level of CNN.

[Ryujin2003]

1 hour ago, hey_yo_ said:

I thought this forum has came into conclusion that WCCF is "fake tech news" just like Bloomberg?

It's not like there were and Geekbench scores listed... So it could be worse.

[Hugsy Malone]

2 hours ago, WereCat said:

So it makes only sense to use GDDR6 for the lower cost Polaris equivalent of the RX 6xx lineup as I dont see HBM pricing and availability getting much better in 2018 for it to make sense to use on a lower-end graphics cards.

Well Samsung started ramping up their HBM2 production towards the end of the summer, so we should see an increase in supply.
Who knows if ddr6 will be used as a lower end alternative, it will be interesting to see!

[Tech Enthusiast]

2 hours ago, WereCat said:

The part of reason for AMD using HBM is the significantly lower power requirements compared to GDDR5 / GDDR5X.

Yah, would be fun to see a Vega with GDDR5X. Finally a reason for those 1000+ watt power supplies.

 

2 hours ago, WereCat said:

They would have to improve the GPU efficiency significantly for Navi to ditch HBM which will not happen.

Actually they need to do this either way. Even if they stick to HBM, they need to find approx 35% power savings at least. 

[Trixanity]

Obviously they're working on getting GDDR6 for mainstream cards. Memory bandwidth is pretty much always an issue so moving onto GDDR6 is only natural. What will be interesting is where HBM and/or flagship cards will be heading. Will it be HBM2? HBM3? Something else? Nextgen memory would hint that it's not HBM2 but it could just be marketing for faster HBM2.

 

There is no doubt AMD has a memory problem; both in bandwidth and in price/abundance. So it will be interesting in 2018 and 2019 where we'll see where AMD is heading.

[ItsMitch]

2 hours ago, hey_yo_ said:

I thought this forum has came into conclusion that WCCF is "fake tech news" just like Bloomberg?

cnn*

[Doobeedoo]

Yeah, saw this coming. Expected GDDR6 used for future GPUs and keeping HBM3 for high end. Makes sense really. 

[Matu20]

HBM on high end and GDDR on mid to low end cards, makes complete sense to me.

[bcredeur97]

i bet after some refinements, we'll see  20Gbps+ GDDR6 in a few years

 

a big big problem right now though is the world's memory shortage... memory is so expensive right now. i hope the higher density here helps with that?

[cj09beira]

1 hour ago, bcredeur97 said:

i bet after some refinements, we'll see  20Gbps+ GDDR6 in a few years

 

a big big problem right now though is the world's memory shortage... memory is so expensive right now. i hope the higher density here helps with that?

dont know if it will, the move to new nodes certainly will, as smaller dies means better costs, (one of the reasons hbm2 is more expensive as it has a larger die than hbm1)

[Misanthrope]

Yeah well, I guess HBM is not the future then? Too soon to rely on it? You've wasted time across 3 generations of products for fucking nothing?

 

GOOD JOB.

[Bit_Guardian]

6 hours ago, samcool55 said:

Good, whatever the replacement for Polaris is, it's likely GDDR6 will cause the biggest performance bump.

RX 580's are basically bottlenecked by ram so unless they improve their compression stuff a lot, faster vram is the best option to improve performance.

Except GDDR6 in a 128-bit bus is already giving you 256GB/s speeds w/8GB of VRAM. The balancing of frame buffer to bandwidth is getting ridiculous.

[samcool55]

2 minutes ago, Bit_Guardian said:

Except GDDR6 in a 128-bit bus is already giving you 256GB/s speeds w/8GB of VRAM. The balancing of frame buffer to bandwidth is getting ridiculous.

Tbh it can't be high enough. I mean smaller busses = lower costs and higher bandwidth = more potential performance.

It can only help price/performance and overall performance in a good way.

[cj09beira]

19 minutes ago, Misanthrope said:

Yeah well, I guess HBM is not the future then? Too soon to rely on it? You've wasted time across 3 generations of products for fucking nothing?

 

GOOD JOB.

saying they will use gddr6 and saying that hbm isnt the future are two completely different things, they are saying though that it isn't cheap enough for low end cards, for that better supply is needed

[cj09beira]

4 minutes ago, Bit_Guardian said:

Except GDDR6 in a 128-bit bus is already giving you 256GB/s speeds w/8GB of VRAM. The balancing of frame buffer to bandwidth is getting ridiculous.

dont forget that they will probably make lower density chips just like what happened with gddr5 at the start, so it wont be a problem, but it might enable a aib to release a version of the card with twice the vram

[DanielMDA]

6 hours ago, samcool55 said:

Good, whatever the replacement for Polaris is, it's likely GDDR6 will cause the biggest performance bump.

RX 580's are basically bottlenecked by ram so unless they improve their compression stuff a lot, faster vram is the best option to improve performance.

You can OC the GDDR 5 to 9GBPS (9216MBps). I have one. I can tell. (Saphire Nitro+ Special Edition RX 580 8GB @1450MHz Core and 9216MBps VRAM)

[cj09beira]

1 minute ago, DanielMDA said:

You can OC the GDDR 5 to 9GBPS (9216MBps)

and they can improve the timings, a significant amount

[Bit_Guardian]

21 minutes ago, samcool55 said:

Tbh it can't be high enough. I mean smaller busses = lower costs and higher bandwidth = more potential performance.

It can only help price/performance and overall performance in a good way.

But now look at how much you'll be packing into an 8GB frame buffer. This is nowhere near enough bandwidth to be accessing it at full tilt. And a 256-bit bus now defaults to 16GB of VRAM? That's crazy.