The AMD Memory Roadmap: DDR3, FBD and G3MX Examined

With today's announcement we can finally talk about AMD's memory roadmap; the two questions we often hear are: when is AMD planning on moving to DDR3 and what about Fully Buffered DIMM? Both are answered with today's disclosure.

AMD will begin the DDR3 transition on the desktop in the second half of 2008 with its Shanghai processor, but AMD won't fully move to DDR3 until 2009 with Bulldozer. The DDR3 transition beginning with Shanghai and completing with Bulldozer is very similar to the cautious approach AMD took to DDR2 adoption. By the time 2009 rolls around, DDR3 should be very cost competitive with DDR2 and the transition should be seamless.

Despite rumors to the contrary, Bulldozer won't support Intel's Fully Buffered DIMM standard, and instead will use what AMD is calling the "G3 Memory Extender" (G3MX).

FBD addresses the problem of not being able to maintain memory frequency while increasing the number of memory sockets on a motherboard, something that impacts the high end server market. The FBD solution is to serialize the memory bus by placing a buffer chip on each memory module that communicates with the memory controller and memory devices. The memory controller only needs to worry about driving data to these buffers, and the buffers deal with getting data in/out of the memory devices.

While FBD supports up to 8 DIMMs per memory channel, there are three major drawbacks: 1) higher cost per module, 2) higher latencies due to serialization and 3) higher power consumption. AMD has said that it evaluates new memory technologies at each generation, and although it won't rule out FBD for future products, it simply doesn't make sense today.

G3MX addresses the same issue of maintaining memory performance while driving up the number of slots per channel in a more economical manner. The technology simply calls for custom, ASIC-class, buffer logic to be placed on the motherboard itself between the memory controller (in this case the CPU) and the memory slots. There is no conversion of memory interface and thus performance/power shouldn't be impacted nearly as much as FBD, and the big upside is that you can use standard DDR3 memory in the sockets since G3MX is implemented at a motherboard level.

The downside is that with G3MX you are still dealing with a parallel memory interface, which becomes difficult to implement at higher speeds and loads. AMD insists that it can work around any issues related to motherboard design and trace routing, and that G3MX is presently a better solution than FBD for its needs.

The first G3MX implementation will arrive with Bulldozer in 2009; like FBD, it will be limited to high end server/workstation platforms.

Codename Mania Bulldozer Performance Expectations
Comments Locked

31 Comments

View All Comments

  • flashbacck - Thursday, July 26, 2007 - link

    Whoever decided those acronyms were necessary should be fired.
  • fzkl - Thursday, July 26, 2007 - link

    Like mentioned, the obvious great benefit of having low power x86 chips on mobile phones is the software aspect. PC applications can now run on phones reducing aspirin needs for developers. However, what does this mean in terms of security? Can we see mobile phones needing frequent patches, antivirus, firewalls like in the case of desktops?

    If this were to be the case we would have successfully made a simple device like the mobile phone(in usage terms) a high maintenance product which a layman might have trouble with.
  • sheh - Thursday, July 26, 2007 - link

    x86 doesn't imply any OS or API. Linux, which is commonly used today on all kinds of devices, can work just as well on x86. Conversely, nothing prevents virus writers from writing viruses for Linux running phones.
  • beyoku - Thursday, July 26, 2007 - link

    what happened? Was this article recelty pulled off or something?
    NDA???
  • Guuts - Thursday, July 26, 2007 - link

    Looks to me like he's trying to get the images working...
  • erwos - Thursday, July 26, 2007 - link

    Fusion looks like it'll be a fantastic chip for UMPCs and laptops. Hopefully they'll manage to squeeze more than CPU one core on there, too. Bobcat looks similarly fun - x86 phones! VIA was also discussing this idea, and I could really go for it.

    That said, AMD is really under-delivering with Barcelona - I suspect the next few years will be pretty rough. Intel has set a low ceiling price for the Barcelonas ($270 - same as the Q6600), and that's not going to be good for AMD's margins.
  • mlau - Thursday, July 26, 2007 - link

    quote:

    x86 Phones!


    Embedded is ruled by ARM, Freescale, mips and sh derivates; amd and intel are going to have a tough time getting a super-ugly system like pc-x86 (with it's
    legacy baggages "bios", "acpi" [you know, the stuff windows requires to run], ...) into that space.
  • Spoelie - Friday, July 27, 2007 - link

    you do not need to have those things to run an x86 cpu

    loot at EFI for example, in use by apple on their x86 based macs = no more bios.
  • qpwoei - Sunday, July 29, 2007 - link

    The real problem with x86 is that it's inherently not power efficient. To get good x86 performance requires lots of transistors and lots of power due to complex decoders and schedulers. A much simpler architecture like ARM requires very few transistors to run efficiently (at the expense of slightly less compact executable code), and is much more suited for battery-powered devices.

    Not to say that there won't be x86-powered devices in the future, just that I don't expect them to really gain much of a foothold in any place where battery life is important (eg: phones).
  • ss284 - Thursday, July 26, 2007 - link

    Return of the Jedi

Log in

Don't have an account? Sign up now