'

Samsung has started mass producing 2nd Generation 8GB HBM2

A upgrade of 50% over 1st Generation 8GB HBM2 memory

Samsung has started mass producing 2nd Generation 8GB HBM2

Samsung has started mass producing 2nd Generation 8GB HBM2

Reducing power draw is a vital aspect when creating next-generation products, either allowing customers to access the same performance with lower power draw, reducing power bills and increasing battery life or allow the same amount of power to be used to make the product run faster.  

Samsung has announced their second generation of 8GB HBM2 memory, codenamed Aquabolt, has entered mass production. Samsung claims that this is the first 8GB HBM2 chips that can deliver 2.4Gbps of data transfer per pin, a 20% increase over their existing 8GB HBM2 memory. 

More astounding is that Samsung has also been able to reduce the power consumption of their chips significantly, achieving speeds of 2.4Gbps at 1.2V, with last generation HBM2 requiring 1.2V to access speeds of 1.6Gbps or 1.35V to achieve 2.0Gbps. This change represents a bandwidth upgrade of 50% when comparing these 8GB HBM2 chips at 1.2V. 

With these changes, a single 8GB HBM2 memory chip from Samsung can now achieve 307 GB/s of bandwidth, which is significantly faster than most of today's existing HBM2 solutions. Using four of these new HBM2 chips would offer 1.2 TB/s of bandwidth to users with 64GB of capacity, which is an astounding figure.  

  

Samsung has started mass producing 2nd Generation 8GB HBM2

 

The sheer bandwidth offered by these chips opens a lot of doors for future mid-range graphics cards, as a single 8GB chip of 2nd generation HBM2 can deliver almost as much memory performance as a GTX 1080 (307GB/s VS 320GB/s) while utilising only a single memory chip.    

You can join the discussion on Samsung's 2nd generation of HBM2 memory on the OC3D Forums

«Prev 1 2 Next»

Most Recent Comments

11-01-2018, 06:11:09

Emu
What kind of latency are they seeing with these chips? Bandwidth is great but latency is important as well.Quote

11-01-2018, 06:12:45

cj09bruno
This is good news for amd, it means a small vega gpu with a single hbm2 stack that could replace polaris 10.Quote

11-01-2018, 06:39:15

cj09bruno
Quote:
Originally Posted by Emu View Post
What kind of latency are they seeing with these chips? Bandwidth is great but latency is important as well.
good point
why cant just let us have access to the memory timings
(miners would love it )Quote

11-01-2018, 11:36:38

NeverBackDown
Quote:
Originally Posted by Emu View Post
What kind of latency are they seeing with these chips? Bandwidth is great but latency is important as well.
Latency is extremely low.

This is good news to hear. I wonder if any Vega chips will switch over. Though I expect Samsung to price them as a premium product.Quote

11-01-2018, 12:14:19

WYP
Quote:
Originally Posted by Emu View Post
What kind of latency are they seeing with these chips? Bandwidth is great but latency is important as well.
Part of the reasoning behind HBM was to lower latency, bringing memory physically closer to the CPU with highspeed interconnects.

Not sure how this new HBM2 compares to Samsung's 1st generation 8GB HBM2, but it will be much better than an equivalent GDDR5/5X setup.Quote
Reply
x

Register for the OC3D Newsletter

Subscribing to the OC3D newsletter will keep you up-to-date on the latest technology reviews, competitions and goings-on at Overclock3D. We won't share your email address with ANYONE, and we will only email you with updates on site news, reviews, and competitions and you can unsubscribe easily at any time.

Simply enter your name and email address into the box below and be sure to click on the links in the confirmation emails that will arrive in your e-mail shortly after to complete the registration.

If you run into any problems, just drop us a message on the forums.