JEDEC publishes GDDR5X graphics memory standard

[zMeul]

source: https://www.jedec.org/standards-documents/docs/jesd232
via: http://www.businesswire.com/news/home/20160121006320/en/JEDEC-Announces-Publication-GDDR5X-Graphics-Memory-Standard

press release:

ARLINGTON, Va.--(BUSINESS WIRE)--JEDEC® Solid State Technology Association, the global leader in the development of standards for the microelectronics industry, today announced the publication of JESD232 Graphics Double Data Rate (GDDR5X) SGRAM. Available for free download from the JEDEC website, the new memory standard is designed to satisfy the increasing need for more memory bandwidth in graphics, gaming, compute, and networking applications.

Derived from the widely adopted GDDR5 SGRAM JEDEC standard, GDDR5X specifies key elements related to the design and operability of memory chips for applications requiring very high memory bandwidth. With the intent to address the needs of high-performance applications demanding ever higher data rates, GDDR5X is targeting data rates of 10 to 14 Gb/s, a 2X increase over GDDR5. In order to allow a smooth transition from GDDR5, GDDR5X utilizes the same, proven pseudo open drain (POD) signaling as GDDR5.

“GDDR5X represents a significant leap forward for high end GPU design,” said Mian Quddus, JEDEC Board of Directors Chairman. “Its performance improvements over the prior standard will help enable the next generation of graphics and other high-performance applications.”

 

http://www.anandtech.com/show/9883/gddr5x-standard-jedec-new-gpu-memory-14-gbps

 

 

The GDDR5X-compliant memory chips will come in 190-ball grid array packaging (as compared to 170-ball packaging used for current GDDR5), thus, they will not be pin-to-pin compatible with existing GDDR5 ICs or PCBs for modern graphics cards. But while the GDDR5X will require development of new PCBs and upgrades to memory controllers, everything else works exactly like in case of the GDDR5: the interface signal training features and sequences are the same, error detection is similar, protocols have a lot of resemblances, even existing GDDR5 low and high speed modes are supported to enable mainstream and low-power applications. BGA packages are inexpensive, and they do not need silicon interposers nor use die-stacking techniques which HBM requires.

 

Implementation of GDDR5X should not be too expensive both from R&D and production perspectives; at least, this is something that Micron implied several months ago when it revealed the first details about the technology.

 

---

GDDR5X is a new type of DRAM specifically designed for video cards, by Micron, and will offer 10 - 12 Gbps data rates with future revisions hoping to surpass 14Gbps
architecturally, it's not different from current GDDR5 standard, but offers double the pre-fetch from 8 data words per memory access to 16 data words per data access - in theory giving the memory double the theoretical bandwidth
Micron is expected to release to market this new type of VRAM sometime in the 1/2 half of this year

there are rumors floating around that both AMD and nVidia will use GDDR5X for their low tier offerings

 

from the JEDEC document, it appears that Samsung and SK Hynix have also shown interest in manufacturing GDDR5X

[don_svetlio]

source: https://www.jedec.org/standards-documents/docs/jesd232

via: http://www.businesswire.com/news/home/20160121006320/en/JEDEC-Announces-Publication-GDDR5X-Graphics-Memory-Standard

press release:

---

GDDR5X is a new type of DRAM specifically designed for video cards, by Micron, and will offer 10 - 12 Gbps data rates with future revisions hoping to surpass 14Gbps

architecturally, it's not different from current GDDR5 standard, but offers double the pre-fetch from 8 data words per memory access to 16 data words per data access - in theory giving the memory double the theoretical bandwidth

Micron is expected to release to market this new type of VRAM sometime in the 1/2 half of this year

there are rumors floating around that both AMD and nVidia will use GDDR5X for their low tier offerings

GP104 was spotted shipping with GDDR5X so yeah.

[Andrew Storlie]

All i know is Nvidia better ship their high end cards with HBM or amd is gonna wipe the floor with them.

[TrigrH]

I present to you:  GDDR5X

 

"cos X sounded better than 5.5"

[givingtnt]

I present to you:  GDDR5X

 

"cos X sounded better than 5.5"

they wanted association with titanx. the god

not lolipop, the (now) outdated

[TrigrH]

they wanted association with titanx. the god

not lolipop, the (now) outdated

2050: android runs out of sweets, better call it Android X guys..

[givingtnt]

2050: android runs out of sweets, better call it Android X guys..

lol

can't wait for android maplesirup

with a big ass canadian flag in the bcg

[Diederik_Gr]

I present to you: GDDR5X

"cos X sounded better than 5.5"

GDDR5s+

[deviant88]

The future is HBM2 or nothing. GDDR is done, only for lower end they should use GDDR5x.

[AresKrieger]

GDDR5s+

No game focused product wants to be associated with apple, that's essentially a death sentence.

 

Well I never really expected either company to use HBM on the low tier stuff, they would run into major supply issues if they tried anyway.

[azurezeed]

And who is manufacturing that memory right now?

No news or rumors about it?

[zMeul]

And who is manufacturing that memory right now?

No news or rumors about it?

read the OP, Micron for now

[NuclearPeace]

All i know is Nvidia better ship their high end cards with HBM or amd is gonna wipe the floor with them.

 

Just like the Fury X wiped the floor with the 980 Ti?

 

HBM isn't a magic bullet and it doesn't automatically make GDDR5 useless. HBM is only going to be seen in flagships where the mid range (where most people actually buy cards) is going to be GDDR5X.

[SurvivorNVL]

The future is HBM2 or nothing. GDDR is done, only for lower end they should use GDDR5x.

The future is HBM2, but it still has more latency overall than GDDR.  I'd be content with upper mid-range being GDDR5x.   There is plenty of bandwidth already and the latency is low.  HBM2 is insane in bandwidth and form-factor, but the fact it's soldered on to the chips/board makes for some interesting issues for the future - they'll have to throw away the entire board if there're any defective modules, where-as GDDR you just swap 'em out.  Not -that- easy, but the point stands.

[zacRupnow]

All i know is Nvidia better ship their high end cards with HBM or amd is gonna wipe the floor with them.

Hahaha, they should, but even if they don't AMD is incapable of wiping the floor with anyone as long as the company remains in it's unprofitable state.

[TheRandomness]

Hahaha, they should, but even if they don't AMD is incapable of wiping the floor with anyone as long as the company remains in it's unprofitable state.

They currently are with the 380 and 390 and Fury, all of which perform better than their counterparts.

[zacRupnow]

They currently are with the 380 and 390 and Fury, all of which perform better than their counterparts.

Nope, Nvidia has far more users, the Fury and 390 have better price/performance than the 980 and 970 but the Nvidia cards have more users, probably cause their extra features are actually useful.

[TheRandomness]

Nope, Nvidia has far more users, the Fury and 390 have better price/performance than the 980 and 970 but the Nvidia cards have more users, probably cause their extra features are actually useful.

Or because all laptops and most prebuilts have Nvidia cards.

[Doobeedoo]

Makes sense to use GDDR5X for GPUs that are lower tier and HBM2 for flagship and dual cards.

[Coaxialgamer]

Why not GDDR6? GDDR5 is SO 2010.

[SurvivorNVL]

Why not GDDR6? GDDR5 is SO 2010.

They almost called it GDDR6, but they went with the x, because it's still the same technology but extended - hence the X.

[MonkeyFace]

Nope, Nvidia has far more users, the Fury and 390 have better price/performance than the 980 and 970 but the Nvidia cards have more users, probably cause their extra features are actually useful.

 

What are the extra features over AMD?

[patrickjp93]

All i know is Nvidia better ship their high end cards with HBM or amd is gonna wipe the floor with them.

No they won't. Take the bandwidth of the 980TI and double it. There is plenty of headroom for the performance expected of the next generation. That said, big Pascal is getting HBM 2.

[patrickjp93]

What are the extra features over AMD?

Shadowplay is the big one.

[Eroda]

Shadowplay is the big one.

 

yep and it works good