HBM2 is now official, supports up to 32GB VRAM and 1TB/s bandwidth

[AlTech]

The High Bandwidth Memory 2.0 Spec has been published and here are the details:

 

So basically:

2x the density of HBM1 VRAM and 2x higher amount of vram supported and because it uses 2x the density it has 2x the max throughput.

 

It has a max throughput of 1TB/s (yes 1 Terabyte per second!)

 

and can come in these variations:

8.16,24 and 32GB.

 

8Gb graphics cards should have 256Gb/s bandwidth,

16Gb graphics cards should have 512Gb/s bandwidth

24GB graphics cards should have 768Gb/s bandwidth

32GB graphics cards should have 1TB/s bandwidth

 

Depending on the amount of DRAM dies used.

 

 

 

HBM2 will basically double the bandwidth offered by HBM1 – which is quite an impressive feat considering that HBM1 is already around 4 times faster than GDDR5. Not only that but power consumption will be reduced by another 8% – once again over an existing reduction of 48% over GDDR5 (of HBM1). But perhaps one of the most significant developments is that it will allow GPU manufacturers to seamlessly scale vRAM from 2GB to 32GB – which covers pretty much all the bases. As our readers are no doubt aware, HBM is 2.5D stacked DRAM (on an inter-poser). This means that the punch offered by any HBM memory is directly related to its stack (layers).

 

Where HBM1 could go as high as a 4-Hi stack (4 layers), HBM2 can go up to 8-Hi (8 layers). The 4-Hi HBM stack present on AMD Fury series is basically a combination of 4x 4-Hi stacks – each contributing 1GB to the 4GB grand total. In comparison, HBM2’s 4-Hi stack will offer 4GB on a single stack – so the Fury X combination repeated with HBM2 would actually net 16GB HBM2 with 1TB/s bandwidth. Needless to say, this is a very nice number, both in terms of real estate utilization and raw bandwidth offered by the medium.

 

As far as use-case confirmations go, Nvidia at-least, speaking at the Japanese version of the GTC confirmed that it will be utilizing HBM2 technology in its upcoming Pascal GPUs. Interestingly however, the amount of vRAM revealed was 16GB at 1 TB/s and not 32 GB. The 1 TB/s number shows that Nvidia is going to be using 4 stacks of HBM – and the amount of vRAM tells us that its going to be 4-Hi HBM2. They did mention however, that as the memory standard matures they might eventually start rolling out 32GB HBM2 graphic cards.

 

 

http://wccftech.com/jedec-publishes-hbm2-specifications-scale-8hi-32gb-stacks-1-tbs-bandwidth/

 

Leave a comment below!

 

Personally this is really exciting! I'm guessing AMD is gonna use loads of this new VRAM and use HBM2 on polaris.

Edited January 15, 2016 by AluminiumTech

[don_svetlio]

Less voltage than DDR3? nice

[Guest]

Great job! Now make it non-volatile and put it on a CPU to let me store my OS on it.

[AlTech]

Great job! Now make it non-volatile to let me store my OS on it.

That'll cost you $4.99 Billion dollars and 99 cents sir.  /jk

[FloRolf]

Hopefully now that high end cards get hbm2, all low end cards will be fed eith gddr5. I hate to see gddr3 cards around in 2016...

[desertcomputer]

Top of the line would cost around 1K. 

[zMeul]

-

already covered here: http://linustechtips.com/main/topic/526547-jedec-updates-hbm-standard-press-release/

 

also... WCCFTech 

[sof006]

Shweeeet    :wub:

[Doobeedoo]

Quite nice jump! Can't believe how much improved HBM2 is over HBM since before, at first, to me it looked like that it only came down to have more memory later on when tech can mature and all. Can't wait to see Polaris flagship, 16GB seems that it would come with

[AlTech]

Quite nice jump! Can't believe how much improved HBM2 is over HBM since before, at first, to me it looked like that it only came down to have more memory later on when tech can mature and all. Can't wait to see Polaris flagship, 16GB seems that it would come with

Nope it's Nvidia who said 16Gb so maybe AMD might come out with 32GB to even out the playing field.