Sign in with
Sign up | Sign in

Samsung Develops 30nm-class DDR4 Module

By - Source: Tom's Hardware US | B 29 comments

Samsung has developed the industry's first DDR4 DRAM module.

Tuesday Samsung Electronics said that it has successfully developed the industry's first DDR4 DRAM module using 30nm class process technology.

The new DDR4 module can achieve data transfer rates of 2.133 Gbps at 1.2V and also makes use of a new technology called Pseudo Open Drain (POD), which allows the module to consume just half the electric current of DDR3 when reading and writing data.

By comparison, 1.35V and 1.5V DDR3 DRAM manufactured at an equivalent 30nm-class process technology offers speeds up to 1.6 Gbps. Samsung added that--when applied to a notebook--the new DDR4 module reduces overall power consumption by 40-percent when compared to a 1.5V DDR3 module.

"By employing new circuit architecture, Samsung's DDR4 will be able to run from 1.6 up to 3.2 Gbps, compared to today's typical speeds of 1.6 Gbps for DDR3 and 800 Mbps for DDR2," Samsung said in a statement.

Samsung said that late last month it provided 1.2V 2 GB DDR4 unbuffered dual in-line memory modules (UDIMM) to a controller maker for testing. There are now plans to work closely with a number of server makers to insure the completion of JEDEC standardization of DDR4 technologies in the second half of 2011.

"The new DDR4 DRAM will build even greater confidence in our cutting-edge green memory, particularly when we introduce 4 Gb DDR4-based products using next generation process technology for mainstream application," said Dong Soo Jun, president, memory division, Samsung Electronics.

Ask a Category Expert

Create a new thread in the News comments forum about this subject

Example: Notebook, Android, SSD hard drive

This thread is closed for comments
Top Comments
  • 13 Hide
    qhoa1385 , January 5, 2011 1:36 AM
    I'm not ready to upgrade my motherboard yet...

    damn you technology!
Other Comments
  • 13 Hide
    qhoa1385 , January 5, 2011 1:36 AM
    I'm not ready to upgrade my motherboard yet...

    damn you technology!
  • 2 Hide
    Travis Beane , January 5, 2011 1:43 AM
    Care to tell us what kind of latencies to expect Samsung?

    I hope I can pop in 6x 4GB DDR4 2400 into a X78 motherboard rocking a octocore Sandy. :) 
  • Display all 29 comments.
  • 1 Hide
    falchard , January 5, 2011 1:48 AM
    The speed worries me. Shouldn't it be at 3200?
  • 0 Hide
    joytech22 , January 5, 2011 2:14 AM
    Aww and i just got 16GB of DDR3.. (accidentally ordered 2x8GB)

    I have to buy a new motherboard this year just for Bulldozer, now i have to buy another one next year as well because i won't be able to handle not having the best..
  • 5 Hide
    sideshowbob32 , January 5, 2011 2:35 AM
    I am still on my ddr2 1100 maybe i will just wait to ddr4 ha. =)
  • 0 Hide
    sideshowbob32 , January 5, 2011 2:36 AM
    SideShowBob32Second thought I wonder how big the difference will be from ddrr3 to ddr4?.
  • 0 Hide
    dogman_1234 , January 5, 2011 3:08 AM
    Now if we can make native DDR4 Quad channle a standard for 2016.
  • 0 Hide
    Caffeinecarl , January 5, 2011 3:28 AM
    SideShowBob32I am still on my ddr2 1100 maybe i will just wait to ddr4 ha. =)

    I second that. I might even wait for DDR5!
  • 1 Hide
    derek2006 , January 5, 2011 3:53 AM
    I'm pissed that I can't even find a good replacement mobo for my core 2 q6600 with ddr2 anymore. I don't wanna buy DDR3 yet.
  • -1 Hide
    Hellbound , January 5, 2011 4:11 AM
    Perhaps Intels x68 will be ddr4
  • 0 Hide
    mirako347 , January 5, 2011 5:09 AM
    Holy O - o
  • 0 Hide
    f-14 , January 5, 2011 5:18 AM
    falchardThe speed worries me. Shouldn't it be at 3200?

    nah they aren't always able to take steady jumps. sometimes they milk a technology first with small speed incremental bumps. there's already ddr3-2100 but ddr3-2000 is more common.
    33~66~100~133~166~200~266~333~400~533~667~800~1066~1200~1333~1600~2000~2100~2133->2333->2400->2666, ?DDR5? 2800->3000->3200->3466
    just go by the 266,333,400 and add your next multiplier #
    ram speeds are pretty predictable for every generation unless they start doubling DDR2 speeds at DDR4/5 Gen which i think more then likely since they're approaching the 22nm barrier by then with DDR5? i know i fouled something up some where but i'm too tired to think this late at night.
  • 1 Hide
    archange , January 5, 2011 5:57 AM
    I'm puzzled: why not forgo DDR4 completely, just like the GPU makers did and skip right to DDR5 instead?
  • 0 Hide
    stingstang , January 5, 2011 6:03 AM
    It took long enough. I remember an article saying it should only be 2-3 years between RAM updates, and that DDR3 was late. This is...what? 3 years late?
    But yea, I still have my DDR2 1150 cards, and they do just fine. I'll have them another year or longer.
  • 1 Hide
    Scott2010au , January 5, 2011 7:10 AM
    Making quad channel standard would require 'yet another' CPU socket, with a significant increase in pin count. So much so power consumption would increase.

    When new generations of DDR are released they typically start +25% to +33% above the clock speed of the last generation, with latencies that defeat the gain in performance. They're geared towards the notebook segment of the market - as that is the largest growing segment and has been for years.

    As for GDDR4 and GDDR5 there are patent issues, ATI made some very intelligent moves in this part of the market a long time ago. Sadly their marketing department is too conservative compared to NVIDIA. They do have a double chipset support advantage though, since Intel isn't licensing tech to NVIDIA any more - NVIDIA thought they'd mention they want to try and make their own CPUs, and may side with ARM when Windows 8 is mainstream. This will just cost them a huge share of the high profit margin market segments.

  • 1 Hide
    Scott2010au , January 5, 2011 7:23 AM
    I'm pretty sure it'll be Z68, and X78 for the unreleased Intel chipsets later in the year btw. Any site talking about X68 is probably making stuff up.
  • 2 Hide
    razzb3d , January 5, 2011 9:03 AM
    Travis BeaneCare to tell us what kind of latencies to expect Samsung?I hope I can pop in 6x 4GB DDR4 2400 into a X78 motherboard rocking a octocore Sandy.

    Why not 8x4gb? or any capacity? Why not go Quad-Channel? You saw what performance enhancemnt the Nahnlem offered over dual-channel architecture.

    Having a 256-bit memory bus seems logical, especially now that CPU manufacturers seem very keen on making on-die GPUs. A 256bit bus should increase the on board video performance significantly. If you are curios, just crack open you desktop or laptop with on-board GPU, and remove one of the memory modules:

    AMD Radeon 4200 with 64 bit bus (1x2GB DDR2 667MHz) - 1670 3dmarks
    AMD Radeon 4200 with 128 bit bus (2x2GB DDR2 667MHz) - 2118 3dmarks
    AMD Radeon 4200 with 128 bit bus (2x2GB DDR2 800MHz) - 2271 3dmarks

    AMD Radeon 5200 with 256 bit bus (4x1GB DDR3 1333MHz) - ???? (hopefully ~ 4500 3dmarks)

    I used 3dmark 2006 and the laptop is a Compaq CQ61.
  • 1 Hide
    JOSHSKORN , January 5, 2011 9:03 AM
    Screw this, I'm gunna now wait for PCI-e 3.0, an affordable 1 TB SSD, DDR4, (8-core) Ivy Bridge, Windows 8 and DirectX12 to come out before I upgrade. When all these pieces fall into place, I'll upgrade. Not until then.
  • 1 Hide
    Scott2010au , January 5, 2011 9:35 AM
    Because it is much cheaper to have 256-bit traces on a daughter board that is optional than forcing it onto much larger and much more costly motherboards - that's why.

    Additionally it would not be viable to use Unbuffered, Non-Registered, memory modules in a configuration beyond 6 DIMMs, let alone 4.

    If people want such architectures they can pay for it by building a Xeon or Opteron Server/Workstation Hybrid. The motherboards will set you back at least $500.

    Also, doubling the memory throughput in the x86 or x64 architecture does not double the performance, since a typical cache has ~ 80% hit rate, and there are several cache tiers in the hierarchy.

    For those that want greater performance in 3D then they can use a PCIe 2.0 x16 slot, or multiple of them, and pay for a GPU with far more transistors, etc (It is the same with doubling the interface throughput of PCIe x16 slots, it is only a very small part of the solution, typically only used to load compressed textures into video memory).

    You could cost it out, and it would be easily beaten by a modern, yet far more affordable, GPU created using the same fabrication tech (i.e. 32nm).

    Or basically, going from very poor graphics performance to 'slightly less than very poor' graphics performance by tripling the cost of the motherboard and adding another 380 pins to the CPU, which will cause it's cost to skyrocket, isn't something that is going to appeal to consumers versus just adding in a daughter-board style video card at 1/10th the cost.
  • 2 Hide
    hardcore_gamer , January 5, 2011 9:51 AM
    I was expecting DDR5 as the next gen maim memory
Display more comments