Leaked slide reveals next AMD FX platform

AMD’s Bulldozer-based FX processors have yet to arrive, but already, Turkish site Donanim Haber has a leaked slide detailing what to expect from their successors. Dubbed FX Next, these next-next-gen CPUs will feature the same reworked “Piledriver” CPU cores as Trinity, AMD’s upcoming replacement for Llano.

What’s new in Piledriver? According to the slide, a 10% increase in x86 performance. Version 3.0 of Turbo Core is also on the menu, as are a couple of new instructions. FX Next will reportedly use the same AM3+ socket as the first wave of Bulldozer CPUs, so you should be able to pop it into existing motherboards.

The AMD slide suggests that FX Next will use the same 9-series chipsets that have been around since May. Those chipsets are actually just rebranded versions of AMD’s old 8-series gear, which has been gracing motherboards for a year and a half now. That means no native USB 3.0 support and a slower 6Gbps SATA controller than competing Intel platforms. The 990FX does offer gobs of PCI Express 2.0 connectivity, but that’s gonna look dated if PCIe 3.0 hardware starts popping up next year.

We don’t know exactly when AMD’s next FX platform will arrive, of course. We’re still waiting for the first one, whose launch may be hampered by poor 32-nm yields at GlobalFoundries.

Comments closed
    • ronch
    • 8 years ago

    So is Zambezi [u<]really finally[/u<] going to be unveiled next week, Oct. 12?

    • link626
    • 8 years ago

    10% increase in multicore performance is not something to brag about.

    10% increase in single core performance is. AMD is getting its ass kicked hard, where a 3ghz dual core intel can beat a 2.4ghz quad core amd.

      • forumics
      • 8 years ago

      for all you know a BD core can perform 500% faster than a SB core thus 10% increase puts a trinity core at 550% faster than SB. we won’t know for sure until BD is released.
      so lets keep pure speculation to ourselves shall we?

        • NeelyCam
        • 8 years ago

        No we shall not. Speculation is all we have, since AMD can’t get the damn thing out the door.

        And speculation is fun. It all sadly ends when the awful BD performance numbers finally become official.

          • Joe Miller
          • 8 years ago

          This time you deserve the -1

            • NeelyCam
            • 8 years ago

            Yes I did.

          • sschaem
          • 8 years ago

          The numbers will be mixed… The price/performance will be ok in most cases, and great for some.

          I think for gaming, specially with a Radeon, the FX-8150 will be a great choice.
          I can only guess that AMD optimized the pants off their Radeon drivers for the FX/Radeon combo.

          The hope is AMD can show that for all general computer task is doesn’t matter anymore.
          Its great that SB might be 20% faster at some java benchmark, but that translate to a fraction of a second advantage in real life.
          That a web page load in 1.1 seconds or 1.08 second is of no relevance to anyone.
          What matter is when you want heavy computing, BD offer similar performance at lower prices.
          And for gaming, once you are GPU limited, spending more for a faster CPU doesn’t make sense.
          The hope is that a 6970+i7-2600k or 6970+FX-8150 will deliver the exact same gaming FPS… but one is $50 cheaper.

          ‘Speculation’ is fun… The big one is : Is october 12th the date the FX is announced ?

            • NeelyCam
            • 8 years ago

            [quote<]I think for gaming, specially with a Radeon, the FX-8150 will be a great choice.[/quote<] The translation from that french magazine said that BD is well behind SB in gaming. I don't know why, but it seemed like they were sure about this... It sounded like they have already benched the chip and are just waiting for the NDA expiration (but somehow screwed up and the magazine leaked out). So, my guess would be that BD will lose to SB in gaming, but I don't dare make it a "was neelycam right" moment

        • sschaem
        • 8 years ago

        We know that a BD core is about half the speed of a SB core in most workload.
        FMA4 optimized code show a lead, but thats it.

        My concern is that AMD has no clue on how to market Zambezi…
        This weakness is actually a strength but I have yet to see ANY AMD marketing monkey express it to anyone.

    • jcw122
    • 8 years ago

    Begin the complaining about the fact that it’s a leak!!! Kids these days…

    • smilingcrow
    • 8 years ago

    Athlon: First Contact.
    Athlon II: The Wrath of the OEMs.
    Phenom III: The Search for Bulldozer.
    Bulldozer IV: The Voyage from Dresden
    PileDriver V: The Final Stepping.
    Vishera VI: The Undiscovered Multiplier
    Zambezi VII: Generations Behind.

      • travbrad
      • 8 years ago

      George Lucas is going to go back and change everything. 🙁

    • bcronce
    • 8 years ago

    AMD7000 series GPUs are suppose to have PCIe3, last I heard. That would mean AMD’s newest platform will not fully benefit their newest GPUs.

    I highly doubt any *real* speed difference, but still.

      • sschaem
      • 8 years ago

      Still kind of a weird situation where Intel would unlock more feature out of a Radeon then AMD own Bulldozer…

        • willmore
        • 8 years ago

        Which Intel chip would that be?

          • sschaem
          • 8 years ago

          Models are announced all over the place.
          [url<]http://www.tomshardware.com/news/z68-PCI-Express-3.0-mobo-pcie,13569.html[/url<] This is coming in a few month, while AMD slide show that even there next gen platform wont support PCI 3.0. Not a huge deal unless you are into HPC ... wait ins't bulldozer Interlagos opeteron supposed to work well with stream computing HW? [url<]http://www.intel.com/content/dam/doc/white-paper/pci-express3-accelerator-white-paper.pdf[/url<]

            • willmore
            • 8 years ago

            I really thought HPC fans would have wanted the memory bandwidth of SNB-E.

            Isn’t IVB headed for laptops at first? Is IVB expected before the SB refresh?

            • sschaem
            • 8 years ago

            Everything point that IB will be out a few month (3 to 6) before piledriver.
            SB-E with PCIe3 support also look to be out way before piledriver…

            Reminder, AMD slide show that piledriver wont have PCIe3.0

            So Intel might have the only PCI3.0 platform for AMD Radeon untill 2013 ?

    • burntham77
    • 8 years ago

    Piledriver? Wasn’t that the name of a Transformer? One of the Constructacons I think.

      • bcronce
      • 8 years ago

      “Piledriver” sounds like something my wife would know.

        • kizzmequik_74
        • 8 years ago

        She watches pro wrestling?

    • Tristan
    • 8 years ago

    AMD will put another 8 cores for 10% performance improvement.
    Moore law is their primary workhorse.
    But CPU cores are not GPU shaders.

    • stmok
    • 8 years ago

    hhh

      • sschaem
      • 8 years ago

      How about we let AMD ship Zambezi before we take all those speculation to heart ?

        • stmok
        • 8 years ago

        hhh

          • sschaem
          • 8 years ago

          Unless the slides are dated post September 28th, I wouldn’t put much weight into any of it.

          Recent disclosure hint that AMD might have to postpone availability of 32nm products by 3 to 6 month.

          • Waco
          • 8 years ago

          Assuming the leaked slides are real…

      • bittermann
      • 8 years ago

      So much guesstimation bullsh*t in your long post I can almost smell it through my internet connection! 🙂

    • smilingcrow
    • 8 years ago

    I’m looking forward to Bulldozer Deep Space 9 Core and Llano Voyager.

    • ronch
    • 8 years ago

    If (first-gen FX CPUs suck) then (10% better performance != enough);

      • sschaem
      • 8 years ago

      10% in a multimedia workload. This is most likely is a tweak to some SSE instructions. The rest look like it will be the same.
      Not sure why even bother investing this effort in such a small upgrade… unless its to fix other undisclosed issues ?

        • Game_boy
        • 8 years ago

        You don’t get 10% from one tweak, or they would have just done it in BD (especially since it’s only coming 3 months after BD launches). It’s probably hundreds of slight changes that add up 10% in the best case (media) and something less than that in other workloads.

          • sschaem
          • 8 years ago

          Yes you can. The benchmark is most likely for code that use FMA3 and thats it no other changes have been done.

          The updated BD is most likely focused on fixing bugs. Possible turbocore in BD doesn’t work as expected on this 32nm process.
          Most likely calibration problems ? Having to release a new version of turbocore ,like you said “it’s only coming 3 months after BD launches”, is highly suspicious.

            • Game_boy
            • 8 years ago

            That makes no sense, any media software out there now or soon would be using FMA4 because the only shipping CPU with FMA (BD) implements FMA4. FMA3 is apparently worse and will only be implemented by Intel in 2013. So any software using FMA would already be sped up by BD.

            I prefer to be more optimistic than you in that they wouldn’t call one specific instruction improvement and some bugfixes a new core name.

            • sschaem
            • 8 years ago

            You have a point that FMA3 support will just help running Intel optimized code and FMA4 could deliver the same with BD.

            So ok, Speculation could go both ways: lots of small changes that aggregate to a 10% speedup in AMD multimedia bench,
            or very focus optimization (like just a faster divide and sqrt implementation)

            My take is that AMD just optimized a couple of SSE instructions.

            I’m pessimist because AMD is giving no clue that ANYTHING is going well…
            beside that Helium2 overclock guiness record, the past 12 month have been all bad news.

            You would think that if AMD had anything positive going on they would use it to boost their image and company moral?

            • travbrad
            • 8 years ago

            [quote<]like you said "it's only coming 3 months after BD launches", is highly suspicious.[/quote<] And you believe AMD launch dates from a leaked slide WHY?

            • Game_boy
            • 8 years ago

            Because they demoed working Trinity hardware openly 6 months ago.

    • dpaus
    • 8 years ago

    Oh, cool – another leak!

      • ronch
      • 8 years ago

      Donanim Haber is leakier than a sinking boat.

      • sschaem
      • 8 years ago

      Next, leaked slides of the successor of the successor of the yet un-released chip…

      It feels like all the engineer and architect left a long time ago and AMD is left with just a bunch of manager making slides.
      “Wouldn;t it be cool if we had a product that was better then the one we thought of yesterday?!”
      “yea make a slide and leak it to Mr D. so everyone know how great we are doing”

        • Joe Miller
        • 8 years ago

        This is good:)

Pin It on Pinterest

Share This