Home Gaming AMD confirms High-Bandwidth Memory, first sizes revealed

AMD confirms High-Bandwidth Memory, first sizes revealed

3 min read
44

f2b685d2_y6esX8i.jpg

It feels like we’ve been talking about High-bandwidth memory (or HBM) for ages now, and there’s a good reason for that. HBM is going to change the face of desktop and laptop GPU computing – and AMD is desperately eager to be the first out the gate with their new Fiji series of cards. But just what is HBM? And, more importantly, how is it going to affect you?

The roadmap to HBM is long and winding, but it essentially comes down to this. DDR memory has been living off exploits from the late 90’s, slicing down the internal clock to increase clock signals and transfer rates. Even the new, DDR4 RAM that Intel will support with Skylake is dependant on this technology – and unsurprisingly the suspected performance gains are rather minimal.

DDR in general is about to hit a memory wall, where the power required to achieve minimal speed differences will hold back computational power that actually puts it to use. This is even evident in SDRAM, or what’s traditionally known as GDDR5 memory on GPUs. With texture sizes increasing, especially as we near 4K resolutions, the aim to to increase the width of the memory bus, and really drive the speeds sky-high. Since GDDR5 is, again, dependant on old technology, achieving this is drawing away power from what your GPU can really do in terms of computing – hence making more advances mostly moot.

gddr5.jpg

HBM changes this entirely, primarily because of how it’s structured on the GPU die itself. In a traditional setup, DRAM chips are placed side by side on a processing unit, similar to the way Intel has done so with the implementation of their Crystalwell Iris Pro graphics. HBM, instead, stack these chips on top of one another, which achieves two things. It decreases the overall footprint the chips have on the die itself, while also reducing power consumption due to the proximity of them all. Lower power, obviously, means far more juice for actual computing on the die itself.

What this means for you is interesting as well. Memory buses can now be thousands of bits wide, rather than hundreds. In fact, the first card AMD is prepping with HBM will sport a 1024-bit bus width, with 100GB/sec of bandwidth per memory stack. With four stacks and 1000MHz effective clock speed, that essentially translates to a total memory bandwidth of 512GB/sec – the types of numbers that are going to allow for faster streaming of massive data packages in future games.

It also means AMD cards will be more power efficient, with the 1.5V draw dropping to 1.3V per stack. The architecture also allows for much smaller form factors – which means Fiji might actually require far less space in your chassis than you expect. The only downside is that AMDs very first implementation of HBM, which should debut in the next few months, will be restrained to 4GB of memory – which really doesn’t take full advantage of the technology it’s going to be paired with.

Nvidia is suspected to debut their own version of HBM with their Pascal cards next year, and I suspect AMD will follow that with much larger cards at around the same time. That’s when we’re really going to see some single-card, 4K capable computing – although the first look at HBM for desktops is something to get really excited about. It’s also something that AMD really needs to nail, given the market crisis they find themselves in currently.

Last Updated: May 20, 2015

44 Comments

  1. Nice

    Reply

  2. Blood Emperor Trevor

    May 20, 2015 at 14:05

    GO GO RED SQUADRON!

    Reply

    • Admiral Chief's Adventure

      May 20, 2015 at 14:05

      ^This

      Reply

  3. Kromas,powered by windows 10.

    May 20, 2015 at 14:07

    Pondering if I should comment on this or not.

    Screw it!

    AMD WILL RULE THEM ALL!

    Reply

    • th3SiCn3ss

      May 20, 2015 at 14:15

      if “memeory” serves me well……. you’ve said that before…..

      Reply

  4. Hammersteyn

    May 20, 2015 at 14:15

    When AMD says the “GameWorks”, it works

    Reply

    • Admiral Chief's Adventure

      May 20, 2015 at 14:16

      I see what you did there…

      Reply

    • Deceased

      May 21, 2015 at 11:55

      😀

      Reply

  5. Pariah

    May 20, 2015 at 14:18

    AMD – Takes 7 years to invent new tech. Includes it on only its top range card.
    Nvidia – Takes 1 year to surpass that tech. Includes it on its entire range.

    I’m calling it.

    Reply

    • Hammersteyn

      May 20, 2015 at 14:19

      Filthy spies

      Reply

    • Kromas,powered by windows 10.

      May 20, 2015 at 14:20

      “Takes 1 year to steal tech and force other to use “theirs” version”

      Fixed it for you.

      😛

      Reply

      • Admiral Chief's Adventure

        May 20, 2015 at 14:21

        ^This

        Reply

      • Pariah

        May 20, 2015 at 14:27

        If they can do it better, let them. It’s all in the name of progress.

        Reply

        • Kromas,powered by windows 10.

          May 20, 2015 at 14:29

          “In the name of progress” was one of Hitlers rally speeches.

          #JustSaying. 🙂

          Reply

          • Admiral Chief's Adventure

            May 20, 2015 at 14:29

            Yoh yoh yoh, you are spot on!

          • Blood Emperor Trevor

            May 20, 2015 at 14:30

            What colour is envy again? 😀

          • Admiral Chief's Adventure

            May 20, 2015 at 14:32

            OOOH OOOOH OOOH I KNOW

          • Pariah

            May 20, 2015 at 14:33

            No idea. But what colour is rage?

          • Admiral Chief's Adventure

            May 20, 2015 at 14:33

            Rage >>> envy

          • Blood Emperor Trevor

            May 20, 2015 at 14:34

            It’s black.

          • Pariah

            May 20, 2015 at 14:36

            That’s racist. 😛

          • Deceased

            May 21, 2015 at 11:57

            This O.O

            You… I like you…

          • Skoobaz

            May 20, 2015 at 14:39

            What if you’re color blind?

          • Pariah

            May 20, 2015 at 14:41

            Which type of colour blind? There are 3.

          • Skoobaz

            May 20, 2015 at 14:41

            Oh I don’t know smarty pants. I was just saying for the sake of saying damnit!

          • th3SiCn3ss

            May 20, 2015 at 14:35

            envy se die LazyGamer forums is boring….. ai julle…..

          • Pariah

            May 20, 2015 at 14:33

            So you’re calling me Hitler. Nice.

          • Kromas,powered by windows 10.

            May 20, 2015 at 14:34

            Calling Nvidia Hitler.

          • Pariah

            May 20, 2015 at 14:35

            Nvidia didn’t say that. I did.

          • Kromas,powered by windows 10.

            May 20, 2015 at 14:37

            Nvidia actually says that a lot.

      • Ranting Raptor

        May 20, 2015 at 14:33

        forced? What exactly do they force?

        Reply

        • Kromas,powered by windows 10.

          May 20, 2015 at 14:35

          They forced certain manufacturers to go green or go away. A while back but I am sure I can get the post for you.

          Reply

          • Ranting Raptor

            May 20, 2015 at 15:12

            Please do. Would be interesting to read about that. Missed those ones

          • Kromas,powered by windows 10.

            May 20, 2015 at 18:03

            I can’t find anything on it cause every time I try to find it on google I get to either the samsung vs Nvidia court battles or the 970 blunder. I’ll search harder this weekend and find the info.

          • Ranting Raptor

            May 20, 2015 at 18:11

            sweet

      • Joe Black

        May 20, 2015 at 16:24

        As the article eludes to everybody has been working on stacked memory for quite a while. AMD in partnership with Hynix is just first to the finish line. I think NVidia calls their’s 3D memory.

        It’s a very nice milestone though. No doubt about it. One which they will benefit from just as much as they will benefit from their timely adoption of closer to the metal graphics APIs like Mantle which will now see them supporting DX12 in everything back to the 7000 series I think.

        Reply

        • Kromas,powered by windows 10.

          May 20, 2015 at 18:01

          Did not know that. Thanks for the info. Sad that it is not proprietary cause AMD needs the boost so the market stays competitive.

          Reply

          • Joe Black

            May 20, 2015 at 23:52

            It was the next logical thing to do. Die sizes were getting too big and with increased efficiency and decreased heat generated/current leakage it was becoming viable. The perfect RAM”chip” will actually be a RAM cube with as much vertical real estate as horizontal.

            Same for CPUs, but I suspect that is going to be much, much tougher to do.

            If you want my opinion AMD has won a very significant first with this one btw.

  6. Krabby Paddy

    May 20, 2015 at 14:50

    Hopefully this gives the Red team lots to cheer about when it is released.

    Reply

  7. Joe Black

    May 20, 2015 at 16:22

    Old news

    Reply

  8. HairyEwok

    May 20, 2015 at 19:10

    You know what happens to things that stack….. They all fall down.

    Reply

  9. Lothy

    May 21, 2015 at 09:23

    what I don’t understand is how they work this stuff out and how suddenly both companies are designing it? Do the both sit down at a meeting and decide what technology will be released when and race to the finish?

    Reply

Leave a Reply

Your email address will not be published.

Check Also

Turns out Microsoft will require a TPM chip for you to install Windows 11

Turns out the much hyped low-specs for Microsoft's new operating system might be more rest…