AMD to clarify ”ambidextrous” strategy on Monday

You can fault AMD for a lot of things lately, but being boring and predictable definitely isn’t one of them. Mere days after announcing an extensive round of layoffs, the company has put out a press release saying it will host a news conference next Monday about its "ambidextrous strategy." Here’s the meaty part of release:

AMD (NYSE: AMD) today announced that President and Chief Executive Officer Rory Read, and Senior Vice President and General Manager Global Business Units Lisa Su will host a news conference to provide updates and more detail on the company’s ambidextrous strategy. The news conference will take place in San Francisco on Oct. 29, commencing at 1:30 p.m. PDT.
The news conference is for invited attendees; however a real-time audio and video webcast of the presentation can be accessed on the AMD Investor Relations home page: ir.amd.com. A live Twitter feed can also be followed at both @AMD_Unprocessed and #AMDFeed.

Now, the "ambidextrous" link just points to the Financial Analyst Day page from last February, but we know very well what AMD is talking about. We wrote all about it at the time.

In short, AMD’s new leadership has been dropping some pretty strong hints that they want to integrate non-x86 (potentially ARM) cores in future AMD processors. At the Analyst Day event, AMD CEO Rory Read specifically used the word "ambidextrous" in reference to CPU instruction set architectures, and Senior VP and General Manager Lisa Su added that AMD isn’t going to be "religious" about architectures going forward. That’s about as explicit as the company got at the time, but it sounds like we’ll be hearing some specifics very soon.

Comments closed
    • Voldenuit
    • 7 years ago

    Ambidextrous [adj]: To go from this
    [url<]http://imageshack.us/a/img829/382/picardfacepalmwl.jpg[/url<] to this: [url<]http://imageshack.us/a/img197/6931/picarddoublefacepalm.jpg[/url<] I kid, I kid. Seriously, best of luck to AMD in their future FrankenPC endeavours.

    • Sam125
    • 7 years ago

    Yeah, going “ambidextrous” probably isn’t a bad idea since as of right now AMD doesn’t have anything for the low-power and embedded markets anymore.

    Going back, I think the AMD BoD decision to fire Dirk Meyer was a huge mistake. He saved the company by selling non-performing divisions and although he did sell off AMD’s mobile division to Qualcomm, engineering a new dedicated low power architecture and creating a new business division would’ve been the easy part once their financials were stabilized. The fact that they ditched Meyer during a rather important transition is why AMD was floundering about for the past several years because someone or someones on the BoD wasn’t thinking forward enough.

      • sschaem
      • 7 years ago

      What lies are you trying to spill ? ATI SoC division was successful and profitable.
      And its one of the biggest success story at Qualcomm.

      By selling and divesting AMD of all its mobile expertize D. Meyer set AMD 3 years behind.

      AMD got into the gutter because D. Meyer Baby bulldozer was a disaster at all level.
      It was late, not on target, hot and slow. Remember he was in charge of engineering when the TLB bug crush AMD server business, and he killed it with BD.

      This man should never have been CEO, AMD is where it is today almost solely because of him.

    • link626
    • 7 years ago

    after the news conference, amd stock will probably drop below $2.

    • Silus
    • 7 years ago

    Considering their cash flow problems and this:

    [url<]http://www.fudzilla.com/home/item/29235-amd-faces-cash-crunch[/url<] They need to do something. The only question is: will it save them ?

      • NeelyCam
      • 7 years ago

      Maybe they could sell their fabs…?

        • chuckula
        • 7 years ago

        Or the Brooklyn Bridge?

    • ludi
    • 7 years ago

    Ambidextrous (n): having the ability to go John Woo style on both feet.

    • BIF
    • 7 years ago

    They’re laying off people left and right. With both [s<]hands[/s<]...er, arms. Ambidextrous is right, sheesh!

    • albundy
    • 7 years ago

    boring and predictable means a steady and growing roadmap. i like it!

    • HisDivineOrder
    • 7 years ago

    AMD’s slow march out of the x86 market continues apace. Eventually, they’ll do the foolish thing and try to be yet another ARM maker. They’ll think there’s this huge market waiting for them. Problem is, there are already too many players there and they’ll get lost in the crowd.

    Then they’ll be bankrupt.

      • Arag0n
      • 7 years ago

      Yeah, but AMD has pretty good know how in several key areas of CPU performance. They can reuse plenty of know how to build their custom ARM SoC’s and be more powerful than anything on market, specially if they can design 64 bit server SoC’s.

      • chuckula
      • 7 years ago

      I don’t know if AMD will go bankrupt in the ARM space, but the devil you know may be better than the devil you don’t know.

      With Intel, AMD had one big competitor, but Intel is very predictable, Intel likes to have a price premium on its products, and while Intel certainly wins in several technological areas there are still areas like GPU where AMD retains a strong technological advantage. For example, look at Haswell: Intel has trumpeted lots of information about it and I guarantee that AMD already has engineering samples in house for testing. Haswell will also not be cheap, meaning that AMD has wiggle room on pricing even if it can’t win on performance. Oh, and AMD can always run to court and sue Intel (again) over some real or imagined grievance. Intel will then settle out with AMD to give it a surreptitious cash injection as long as the terms are not too crazy.

      Now look at the ARM world: everything is much much more proprietary (no detailed engineering talks 6 months in advance of product launch), you are dealing with multiple competitors that are all HUGE (Samsung alone is a larger company than Intel), AMD has zero technological advantages in this space (AMD’s mobile GPUs are all owned by Qualcomm), and the competitors have no problems being cut-throat on pricing because they are often getting money from vertical integration that AMD cannot take advantage of. Oh, and AMD will never be able to get away with suing all of its competitors as being “monopolies” in the ARM space unless there is massive consolidation.

      Anyway, for anyone thinking that AMD can just throw out an ARM chip and suddenly be making buckets of money, the situation is far more complex.

        • blastdoor
        • 7 years ago

        All very true.

        I think bankruptcy is the most likely outcome for AMD. At that point, its pieces can be sold off to the highest bidder.

        The (distant) second most likely scenario is that current shareholders accept reality and sell AMD for the price its worth to a company like Samsung, Qualcomm, or Apple. I suspect that shareholders will have a very hard time accepting reality, though.

        • ET3D
        • 7 years ago

        It’s true, but look at NVIDIA. It managed to get into this space and be successful enough without any prior experience. NVIDIA has a big advantage over AMD in marketing (IMO), but I think that AMD shouldn’t have a problem creating something that’s technically competitive in this space. I think it has enough CPU expertise to be able to create good custom cores, and it’s already working on making its GPU architecture as low power as possible for the ultra low power x86 space. So I think that it has a chance to trump NVIDIA.

          • chuckula
          • 7 years ago

          1. It’s unclear how much money Nvidia is really making in this space. Remember that Nvidia is in a very much better position financially than AMD is, and can afford to take some risks in these areas. Tegra 3 gets a lot of attention, but it is unclear how much profit Nvidia has made so far (they are banking on future returns).

          2. Nvidia has been developing Tegra since 2007 at the earliest (could even be 2005-2006 for in-house development). The first version of Tegra didn’t sell commercially at all. Tegra 2 had limited commercial sales. Tegra 3 was the first version that finally started to turn the corner in late 2011 to 2012. That’s a LONG time between the beginning of the project and the first signs of potential commercial success. AMD is at the same stage in 2013 where Nvidia was in 2007…..

            • kalelovil
            • 7 years ago

            The first Tegra did end up in the Microsoft Zune HD and Microsoft Kin series, although those product lines were both failures.

            • NeelyCam
            • 7 years ago

            The second Tegra ended up in Motorola Atrix – the first dual core phone in the US (and one that’s still surprisingly competitive in the marketplace)

      • Jasked1
      • 7 years ago

      Qualcomm’s Adreno was AMD technology that AMD sold to them back in 2008. Many thought this was a mistake at the time, and it probably was. This was likely AMD’s only reasonable chance of getting into the ARM market.
      (http://en.wikipedia.org/wiki/Imageon)

      • BestJinjo
      • 7 years ago

      AMD isn’t stopping making CPUs and GPUs. Instead they are diverting money that was wasted for 5-6 years trying to compete with Intel juggernaut on CPUs above $200 towards faster growing markets. AMD has no chance to beat Intel in the high-end CPU game and just to keep up it costs a tremendous amount of $. Thus, they already decided they will not waste $ on a task that they know cannot be done in a decade. Intel not only has a more advanced CPU architecture but is always 1 node ahead of AMD. That means AMD’s engineers would never beat Intel until Intel makes a mistake because even if AMD miraculously designed a CPU that’s faster in IPC, it’ll still have worse power consumption by virtue of being 1 node behind. Frankly, if it wasn’t for the Netburst debacle, Intel would have been ahead every single generation until now. You could say AMD got lucky that Intel made a miscalculated move with Netburst.

      The alternative then is to focus on areas you are good at (GPUs) and expanding into new growth markets which happen to be smartphones and tablets. You seem to always criticize everything AMD is doing but what alternative ideas have you offered for them to undertake instead? Let me guess continuously pouring money into CPUs which they have done since A64/X2 and failed with Phenom I / II / Bulldozer / Vishera. Particularly as the PC market, which provides 85 percent of AMD’s sales, has been drying up because of the economic melt down. So that can’t be a good strategy to continue wasting $ on.

      • MadManOriginal
      • 7 years ago

      I don’t think they are looking to be another ARM maker, at least not pure ARM chips and certainly not soon. It sounds like what they are trying to do is use ARM instruction IP alongside x86 and/or GPU to make SoC chips that ‘have it all’ – good enough x86 performance, good enough graphcis performance, and ARM on the side for low power draw and computing power situations or dedicated tasks. Picture something like ARM for I/O, dedicated silicon for encoding/decoding, background system tasks, encryption (which I think they’ve specifically mentioned). These are all things ARM already does so it wouldn’t be hard, they might not all make sense though, I am just throwing out examples.

      *I don’t know how combining x86 and ARM might work in software. Drivers perhaps?

    • Hattig
    • 7 years ago

    Heh, they’ll pull out of x86 over the next few years, concentrate on high-end custom ARM and SoC design services, and leave Intel in the monopoly do-do.

    How it’ll play out, who knows. I’m sure AMD could design a lovely high performance 125W ARMv8 based APU though.

      • khands
      • 7 years ago

      I wonder how many cores you could shove into an SOC with that TDP.

        • Scrotos
        • 7 years ago

        Probably 16 ARM cores. Or 8 ARM modules. It really depends. Could be 24 ARM “units” if AMD plays their cards right!

        • willmore
        • 7 years ago

        The limiting factor comes down to I/O, and not how many low power cores you can fit in.

        All HPC ends up being about communication when you scale up.

        • MadManOriginal
        • 7 years ago

        There’s nothing magic about the performance/watt of ARM. Sure, you could shove a lot of weak cores onto a die but it wouldn’t end up hugely different in overall performance from another CPU with the same power draw.

    • Tristan
    • 7 years ago

    AMD is ideal capitalistic company. They do not make profit from our money.

      • MadManOriginal
      • 7 years ago

      Broken English makes this post go from funny to hilarious.

      +1 !

        • BIF
        • 7 years ago

        I agree! +1 !

    • ronch
    • 7 years ago

    AMD has been executing pretty well lately. Here’s hoping they’ll succeed fixing their company for good.

    • chuckula
    • 7 years ago

    So far they’ve dropped a low-end ARM core into some future designs to handle DRM instead of building the circuits themselves. That’s not quite a heterogeneous compute strategy but it has some of the underpinnings of one.

    My guess is that this whole ambidextrous thing will show up in server blades from Seamicro. You’ll see ARM blades being integrated with the same backend power & communication fabric that the existing x86 blades use. If AMD can re-sell Intel parts in Seamicro servers, I don’t see why it couldn’t do so with ARM parts. Oh, and it is very unclear if AMD will be heavily involved in designing the ARM silicon or will just contract with an existing manufacturers and focus more on system integration.

    Here’s the most out-there guess: AMD wants to compete with Nvidia on the “project denver/boulder/aspen/whatever” front. For some reason, AMD decides that ARM cores would work better in a stand-alone video card product than its own x86 cores… I’m not sure why and it’s not really a vote of confidence in AMD’s design abilities if this were true though. Now, an APU that is effectively a video card with just enough CPU to keep the GPU fed with data I could see. I’d just assume that AMD would rely on its own x86 talents to pull it off instead of jumping into the me-too ARM market.

    • Arag0n
    • 7 years ago

    It may mean an ARM core for low power usage and x86 for backward compatibility. I’m wondering how good it can be for a Windows 8 tablet…. I would prefer them to get an ARM license and use their know how in areas like Hyper Transport, I/O connectivity, memory interfaces and GPU to create a kick ass full ARM chip.

    The single next best thing (or first) should be an ARM core with x86 compatibility, that means an ARM with x86 execution capabilities if needed (but slow ones for sure).

      • willmore
      • 7 years ago

      Didn’t they do that in June? The ARM license, that is.

        • Tom Yum
        • 7 years ago

        They did but from memory they licensed the ARM Cortex-A5 for its TrustZone security tech. The Cortex-A5 is the lowest performance Cortex processor ARM licenses, so I don’t think they would select that if they intended to have ARM compatibility.

    • MadManOriginal
    • 7 years ago

    I think it means the left hand won’t know what the right hand is doing. 😀

    But no, really, aside from CPU cores it ought to include integrated CPU/GPU as well…maybe they are too invested in ‘Fusion’ branding already, but the two things – ARM or alternative cores, and CPU+GPU – could be under the same umbrella.

      • sweatshopking
      • 7 years ago

      THEY SHOULD MAKE A MAXIMUM CPU/GPU/MC/ETC. THEY COULD PUT ALL THE THINGS IN THERE! POWERPC! SPARC! X86! X64! ARM! MIPS! THAT CHINESE ONE!
      IT COULD RUN [b<] EVERYTHING!!!!! [/b<] WELL, IF YOU HAD THE SOFTWARE TO DO IT.

        • Arclight
        • 7 years ago

        That sounds cool as sh*t and i want that.

          • khands
          • 7 years ago

          Call it “The Emulator”

          • Scrotos
          • 7 years ago

          Seconded.

        • ronch
        • 7 years ago

        [quote<]IT COULD RUN EVERYTHING[/quote<] More like it could run AMD to the ground.

          • sweatshopking
          • 7 years ago

          meh.

      • Alexko
      • 7 years ago

      I haven’t seen anything with “Fusion” on AMD’s slides in a while. I think they’ve essentially dropped it.

      They just talk about APUs, HSA, and apparently ambidextrous stuff, now.

Pin It on Pinterest

Share This