GDDR6 Internal Qualification Completed. Mass production to commence in H12018

[AlTech]

Micron has finished internal qualification for GDDR6 and they expect mass production to commence within the first Half of 2018.

 

Quote

With rumors swirling pretty widely that the next generation of video cards from both vendors will be based around GDDR6 memory, there’s understandably a lot of interest in the development and release status for the new memory. While GDDR6 is not set to ship until 2018, 2017 is now just weeks from coming to a close, meaning “later” is quickly becoming “sooner” and vendors are releasing increasing amounts of information about their GDDR6 plans.

No exception to that rule is Micron, who published a blog post this morning recapping their 2017 graphics memory efforts and outlining their production initiatives for 2018. The company already has a foot in the door on the next-generation memory game with GDDR5X, and as previously announced by the company, they’re looking to pivot that experience into producing GDDR6.

Getting to the meat of Micron’s blog post then, the company has revealed today that they have finished the design and internal qualification of their 12Gbps and 14Gbps GDDR6 chips.  This means that at least internally, they have completed what’s arguably the hardest parts of GDDR6 development from the memory side of matters, and now they can focus on mass production and the development of even higher speed bins. Interestingly, according to the company they actually came in a bit ahead of schedule here, as they weren’t originally planning to be done with internal qualification quite this soon.

 

One of the benefits of GDDR6 should be lower power due to reduced voltages from 1.5v to 1.35v .

 

A table of GDDR5X compared to GDDR6

 

It is expected for GDDR6 to reach between 12 and 14 Gbps effective throughput.

 

Quote

As far as mass production goes then, Micron has previously discussed reaching mass production of GDDR6 in the first half of 2018, and as of their latest update that’s still the goal. In fact the company is already sampling the chips. Notably here, they’ve said that they’re “pushing” for H1 on mass prodcution, but with internal qualification slightly ahead of schedule, it puts them in a better position to hit their goals.

Meanwhile the company’s chip designers are now buckling down to work on the next set of GDDR6 speed bins, including 16Gbps GDDR6. This is an area in particular that the company is looking to leverage their GDDR5X experience, as the biggest challenge at this point is the signaling rather than the memory cells themselves. Successive GDDR standards have generally only made modest changes to the internal clockspeed of the memory cells, while the interface speeds themselves have continued to ratchet up. And while GDDR5X and GDDR6 have some notable differences between them, Micron’s existing experience with 12Gbps+ GDDR5X is generally applicable to GDDR6 as well

 

So yeah, cool.

 

GDDR6 shares a lot in common with GDDR5X. And it is anticipated that GDDR6 will be commanding a price premium over standard GDDR5 for some time after launch.

 

Source:

https://www.anandtech.com/show/12186/micron-finishes-gddr6-internal-qualification

[bleedblue]

Is this better than HBM2?

[TrigrH]

4 minutes ago, bleedblue said:

Is this better than HBM2?

no and it uses more power, but its significantly cheaper.

[Nicnac]

yay more expensive vram

[Ordinarily_Greater]

no GDDR6Z?

[AlTech]

4 hours ago, bleedblue said:

Is this better than HBM2?

No. Unfortunately not.

 

3 hours ago, Nicnac said:

yay more expensive vram

After launch the prices could potentially come down but part of the increases in price of G5, G5X and G6 is the rising cost of NAND flash due to companies artificially raising the price of it.

 

3 hours ago, TrigrH said:

no and it uses more power, but its significantly cheaper.

Pretty much this. Although with that said. the difference won't be as huge as it is now between HBM2 and GDDR5.

[AlTech]

46 minutes ago, Ordinarily_Greater said:

no GDDR6Z?

GDDR5X was a temporary solution designed to provide more bandwidth to bandwidth limited GPUs no matter the cost. G5X uses more Power and is more expensive than G5 but provides enough bandwidth for GPUs such as the GTX 1080 and 1080Ti.

[Ordinarily_Greater]

47 minutes ago, AluminiumTech said:

GDDR5X was a temporary solution designed to provide more bandwidth to bandwidth limited GPUs no matter the cost. G5X uses more Power and is more expensive than G5 but provides enough bandwidth for GPUs such as the GTX 1080 and 1080Ti.

i guess this GDDR6 will be taken to a higher level by adding Z to the last character

[Bit_Guardian]

1 hour ago, AluminiumTech said:

No. Unfortauntely not.

 

After launch the prices could potentially come down but part of the increases in price of G5, G5X and G6 is the rising cost of NAND flash due to companies artificially raising the price of it.

 

Pretty much this. Although with that said. the difference won't be as huge as it is now between HBM2 and GDDR5.

Companies are not artificially increasing the price. They're converting over all of their manufacturing capacity to 3D NAND, which uses more foundry floor space compared to planar, and demand is meanwhile continuing to go up.

 

And because the big DRAM makers are also the big NAND makers (Hynix, Samsung, and Micron), the foundry space shortage affects both product lines. Now, Intel could return to its roots and start making DRAM in mass, and that would solve the DRAM pricing issue, but that's unlikely to happen.

[Doobeedoo]

So we will see it in like mid to high range of GPUs using it at first due to price, while HBM(3)? for top tier GPUs that will come?

[AlTech]

1 hour ago, Ordinarily_Greater said:

i guess this GDDR6 will be taken to a higher level by adding Z to the last character

I'm guessing no, based off of the original article.

[AlTech]

1 hour ago, Doobeedoo said:

So we will see it in like mid to high range of GPUs using it at first due to price, while HBM(3)? for top tier GPUs that will come?

Yeah, budget GPUs probably will still use GDDR5 (so think something like an RX 650 and RX 660), meanwhile possibly GDDR6 for mid range (RX 670 and RX 680) and potentially HBM2 for High End.

 

HBM3 hasn't been developed yet and HBM2 is still in its infancy. HBM2 costs need to come down though for it to be used on Nvidia's High End GPUs.

[AlTech]

1 hour ago, Bit_Guardian said:

Companies are not artificially increasing the price. They're converting over all of their manufacturing capacity to 3D NAND, which uses more foundry floor space compared to planar, and demand is meanwhile continuing to go up.

Can't remember which exact article but I swear, a few Japanese companies decided to jack up the prices of NAND flash and artificially limit supply to increase profit margins or something. Also, there's NAND shortages from some companies switching the purpose of their factories.

[Matu20]

10-20% more bandwidth, but a whole new arch? Doesn't seem too good.

[Doobeedoo]

37 minutes ago, AluminiumTech said:

Yeah, budget GPUs probably will still use GDDR5 (so think something like an RX 650 and RX 660), meanwhile possibly GDDR6 for mid range (RX 670 and RX 680) and potentially HBM2 for High End.

 

HBM3 hasn't been developed yet and HBM2 is still in its infancy. HBM2 costs need to come down though for it to be used on Nvidia's High End GPUs.

Right, HBM3 still far off. Though for Navi stated "Next Gen Memory" so it crossed my mind.

[Trixanity]

42 minutes ago, AluminiumTech said:

Can't remember which exact article but I swear, a few Japanese companies decided to jack up the prices of NAND flash and artificially limit supply to increase profit margins or something. Also, there's NAND shortages from some companies switching the purpose of their factories.

Because they are. They know unit prices would go down if they attempted to meet demand. They've been there before where profit was very low. Why would they wish to go there again? It's the same with the HDD industry. They had a flood years ago and have somehow never recovered meaning market prices would never reach the previous record low.

[Bit_Guardian]

7 minutes ago, Trixanity said:

Because they are. They know unit prices would go down if they attempted to meet demand. They've been there before where profit was very low. Why would they wish to go there again? It's the same with the HDD industry. They had a flood years ago and have somehow never recovered meaning market prices would never reach the previous record low.

Sigh, again, no. These are economies of scale. The price of production per unit drops the more units you produce. Datacenter demand for drives is still increasing faster than manufacturing can keep up. It's not artificial at all.

[Trixanity]

26 minutes ago, Bit_Guardian said:

Sigh, again, no. These are economies of scale. The price of production per unit drops the more units you produce. Datacenter demand for drives is still increasing faster than manufacturing can keep up. It's not artificial at all.

Sure if you want to believe that.

The industry has even hinted that they could meet demand but they've decided not to. Why would they do this? Because the unit price gets too low and it suddenly reduces profit.

 

It's pure cartel from here on out. It's better business. It only takes one to spoil it though but I don't think either party is willing to do that as it'll eventually hurt business.

[Bit_Guardian]

44 minutes ago, Trixanity said:

Sure if you want to believe that.

The industry has even hinted that they could meet demand but they've decided not to. Why would they do this? Because the unit price gets too low and it suddenly reduces profit.

 

It's pure cartel from here on out. It's better business. It only takes one to spoil it though but I don't think either party is willing to do that as it'll eventually hurt business.

Are you joking? Semiconductor manufacturing economics is one of the most well-documented things in this industry!

 

Where are these supposed hints? If anyone could capture all of that market by offering something significantly cheaper (I got a 1TB Sandisk Ultra II for $200 last November vs. the $250 it is now), they would. That's a ton of money to get.

 

And there are too many players for it to be cartel. Micron (Crucial), Intel, Samsung, Hynix, Toshiba/WD, and a couple smaller ones.

[mrchow19910319]

6 hours ago, TrigrH said:

no and it uses more power, but its significantly cheaper.

wait.. i dont get it, HBM2 is superior but why AMD couldn't sell as many cards as nvidia? 

 

[Guest]

5 hours ago, Ordinarily_Greater said:

no GDDR6Z?

GDDR6Xp

[Bit_Guardian]

18 minutes ago, mrchow19910319 said:

wait.. i dont get it, HBM2 is superior but why AMD couldn't sell as many cards as nvidia? 

 

Because AMD doesn't have the high end and doesn't have enough mindshare for consumers to really look at them when buying graphics cards.

[mrchow19910319]

2 minutes ago, Bit_Guardian said:

Because AMD doesn't have the high end and doesn't have enough mindshare for consumers to really look at them when buying graphics cards.

i owned both RX480 and GTX1060. And from my own experience, GTX1060 is MUCH MUCH cooler, quieter, and the driver did not fuck up my system that much. And i think a lot of time their product are out of stock is another problem...

[Trik'Stari]

On 12/26/2017 at 12:39 PM, mrchow19910319 said:

wait.. i dont get it, HBM2 is superior but why AMD couldn't sell as many cards as nvidia? 

 

1. HBM2 is harder to produce and naturally more expensive to make requiring more resources and is therefore more expensive to offset the loss in profit margin.

 

2. HBM2 is harder to produce and therefore has lower yields which garner higher prices yet again (And less availability naturally)

 

[Blackie Sheen]

On 26/12/2017 at 12:39 PM, mrchow19910319 said:

wait.. i dont get it, HBM2 is superior but why AMD couldn't sell as many cards as nvidia? 

 

because AMD's cards with HBM2 are still slower than Nvidia's flagship. A better dram config is nothing without the appropriate computing power to back it up.