Former Samsung, AMD chip exec defects to Apple

Apple has poached another high-profile chip executive. This time, the new hire is making headlines because he’s defecting straight from Apple’s courtroom foe and smartphone arch-nemesis. According to the Wall Street Journal’s Digits blog, Jim Mergard has left Samsung’s Austin, Texas chip design team to go work for Apple.

The blog says Megard was among the "most prominent" new recruits in that team. Mergard’s career also included a 16-year stint at AMD, where he served Vice President and Chief Engineer and was "known for playing a leading role in the development" of AMD’s low-power E-series APUs, which are part of the Brazos platform.

This hire may be more significant than some realize. Patrick Moorhead, a former AMD VP and Corporate Fellow who now heads his own analyst firm, told the Journal that Mergard offers "deep expertise" in system-on-a-chip devices. The way Moorhead sees it, Mergard would be "very capable of pulling together internal and external resources to do a PC processor for Apple."

Apple is no stranger to custom chip design, of course. The company has featured its own A-series chips in devices dating back to the iPhone 4. Its new A6 SoC, which powers the iPhone 5, even eschews cookie-cutter ARM Cortex cores in favor of Apple’s own, ARM-compatible ones. Building a bigger, faster chip and slapping it in a next-gen MacBook Air wouldn’t necessarily be a huge step up from that—and by the looks of it, Mergard could be the man to make it happen.

Comments closed
    • albundy
    • 7 years ago

    so he is going from companies that fabricate real products to a company that fabricates their logo on their designer chassis. must be really good money!

    “The way Moorhead sees it, Mergard would be “very capable of pulling together internal and external resources to do a PC processor for Apple.”

    if that were the case, why wouldnt Mergard just create his own cpu company and sell rights to apple?

      • Scrotos
      • 7 years ago

      You need money to create your own CPU company. Look at Transmeta. Look at PA Semi. PA Semi was lucky in that they had ARM experience and got bought up. Transmeta got bought up and shut down. PowerVR reinvented themselves for embedded graphics which kept them alive. What other new companies have popped up and did anything? How about these guys? [url<]http://caustic.com/[/url<] Remember them? Where's that raytracing hardware from 2009? Even big companies like Intel can throw a ton of money at CPU designs and have them fail. Look at Itanium. The only reason that's still around is that Intel got everyone else to kill off their own CPUs (PA-RISC, MIPS, Alpha, etc.) and put their lot into Itanium. Now they are even migrating to x86 Xeon hardware instead and HP has to pay MS to keep working on Itanium since they forced all their customers to that platform and haven't ported all their products over to x86. You make it sound so easy when in reality it really, really isn't. Far better to go to an established company with a ton of resources than to go it alone.

      • Beelzebubba9
      • 7 years ago

      Name one other smartphone handset maker that designs their own CPU cores. Go ahead, I’ll wait.

        • Helmore
        • 7 years ago

        Qualcomm?

        Linky: [url<]http://www.anandtech.com/show/5559/qualcomm-snapdragon-s4-krait-performance-preview-msm8960-adreno-225-benchmarks[/url<] I know I know, they don't really build any smartphones for the consumer market, but they do make their own smartphones :P.

          • BobbinThreadbare
          • 7 years ago

          Samsung.

    • LocalCitizen
    • 7 years ago

    it can’t be that simple. there must be some sort of non-compete clause in his Samsung contract.

      • dpaus
      • 7 years ago

      Ever seen one of those successfully enforced?

        • LocalCitizen
        • 7 years ago

        not often, but just last Jan, Giovanni Visentin of IBM tried to go to HP. IBM actually requested a injunction. IBM lost, I’m no legal expert, but looks like some of things IBM did could’ve cause the court ruling to go either way.

        • Sam125
        • 7 years ago

        Probably not as often as they should be, but if Apple begins to release much more advanced SOCs then you can bet Samsung will be going over the architecture with a fine tooth comb. Rather than luring Mr Mergard away for specific knowledge, Apple most likely wants him for his years of experience in leading design teams rather than holding any valuable trade secret.

    • tbone8ty
    • 7 years ago

    [url<]http://www.techpowerup.com/173557/AMD-Announces-Preliminary-Third-Quarter-Results.html[/url<] [i<]"AMD today announced that revenue for the third quarter ended September 29, 2012 is expected to [b<]decrease approximately 10 percent sequentially.[/b<] The company previously forecasted third quarter 2012 revenue to decrease 1 percent, plus or minus 3 percent, sequentially. The lower than anticipated preliminary revenue results are primarily due to weaker than expected demand across all product lines caused by the challenging macroeconomic environment."[/i<] OUCH! yah safe to say AMD is Fk'd! this happened last quarter and their stock dropped 20% expect sub $2 levels by the end of October. cue take over rumors in 1, 2, 3...

      • destroy.all.monsters
      • 7 years ago

      The problem being there’s no one that can take them over without them losing the x86 license.

      About the only way it could or would happen is with a sort of reverse merger in the same way that Apple/Next or ATI/ArtX occurred. Only if AMD is the surviving entity in a merger can they keep their license.

      At some point there will be a question of x86s possible value in the market but that time isn’t now. I can’t even see who would bother just taking on the former ATI side except possibly intel which makes the whole concept a non-starter in my view.

        • Geistbar
        • 7 years ago

        That’s only if the current cross-licencing agreement holds up in court. In all likelihood, due to factors such as monopoly concerns, a purchaser would get to keep access to the x86 licence. Right now, the risk and the cost to take that risk (aka buy AMD) is too high. If AMD’s value falls enough then I expect that somebody will buy them up and take their chances.

          • destroy.all.monsters
          • 7 years ago

          If it gets to SCOTUS all bets are off – and more likely than not they’ll consider ARM’s market share a compelling reason to say that there’s enough competition.

          If it never gets there – then possibly but it also depends on who’s in the white house at the time too.

            • Geistbar
            • 7 years ago

            How often do these types of disputes make it to the supreme court? The president won’t have any effect on ongoing court proceedings, and as far as I know there’s no oversight here that is comparable to what the FCC provides over its domain.

            There are other factors that could easily weigh in besides competition as well. Notably, that the agreement appears to be, in practice, rather one sided.

            • destroy.all.monsters
            • 7 years ago

            The president will have a lot to do with proceedings since it’s he that tells DoJ how he wants it handled and it will take DoJ pressure (or the lack of it) in order to force the issue of x86 openness. How all anti-trust cases are handled is due to whoever is in the executive branch at the time.

            Sure it’s largely one-sided – but someone would have to actually sue intel in order for it to even be put on trial. Jen Hsun might want another bite at that apple but frankly with their focus on Tegra I don’t see it.

            Presuming that anyone would want AMD’s corpse for anything other than patents is wishful thinking I’m afraid.

        • sschaem
        • 7 years ago

        Apple could just get all the GPU IP to invest in its SoC, and dump the x86 license altogether.

        AMD paid 6 billion for ATI, Apple can grab ‘ATI’ on the open market for 2 billion today
        (Plus they get AMD and seamicro as a side bonus)

        They would also build a killer iTV gaming platform (to kill sony and microsoft)
        And side bonus of getting PC GPU at cost.

        But the fact that Apple han’t made the move yet, show that they believe they can recreate the technology themselves. (With ‘half’ of AMD designer staff already working there, it might be true)

      • sschaem
      • 7 years ago

      Its would be ironic if by the end of the year if AMD is worth less then what they paid for seamicro.
      They are headed in that direction fast.

      • mnecaise
      • 7 years ago

      Stocks down > 10% in overnight trading. It will open well below $3 this morning.

    • sschaem
    • 7 years ago

    What software will run on this macbook ARM edition?
    And will this chip really be better then Intel next gen 14nm Haswell (edit:<broadwell)?

    I see nothing in his 16 year track record at AMD that prove he can beat Intel next gen.

    And then what, will he use an imageon GPU ?
    Apple got near ZERO gpu experience, how will they build a SoC that spank Intel Haswell.

    If Apple pull this off (Beat Intel Haswell CPU performance/watt and beat AMD GPU design), its going to make the industry shiver to the bone!!!
    Reality ? The guy as some expertise, but to even thinking one man that worked 16 year at AMD is bigger then the entire Intel research team is ludicrous.

      • dmitriylm
      • 7 years ago

      Who is this in response to? Or are you just arguing with yourself?

        • jdaven
        • 7 years ago

        Lol! Yeah, I felt like a vagrant just wandered into the room and started spouting off nonsense.

        • sschaem
        • 7 years ago

        Just read the last paragraph of this article written by C. Kowaliski…

        I’m not addressing this at him personally, just putting forth how little weight this argument has.

      • vaultboy101
      • 7 years ago

      sschaem makes a good point though.

      I am a happy Mac OS X user on my Macbook Pro Retina and I would be pretty annoyed if Apple go for one of their lame low IPC ARM chips.

      That means that there will be a performance regression if they ditch Intel for ARM which is not acceptable to Mac users.

      This was not the case when Apple switched from PowerPC to Intel. The Core arch was far faster than IBM’s at the time.

        • sschaem
        • 7 years ago

        And not to be ignored, people that use macbook and not ipad run x86 software. Might be Acrobat, some web service, xcode IDE, etc..?
        An emulator would KILL any battery advantage a magical AMD SoC CPU (directed by an ex AMD fellow) would have over intel 14nm broadwell.

        • Beelzebubba9
        • 7 years ago

        I think the point, from Apple’s perspective, isn’t as much to shoot for the moon and make an Intel [whatever] killer, but to have other options that keeps them from being a slave to Intel’s roadmap. From the articles and editorials Anand’s written on the topic, it seems like Apple is pretty unhappy with Intel’s inability to deliver decent GPU performance in line with what Apple would like in the segment, and considering Apple’s massive bank account it only makes sense to burn 2% of their cash on hand to have a possible alternative.

        Also, I find it [b<]very[/b<] unlikely that Apple will be able to make an ARM SoC that's better than what Intel can do in absolute performance to watt, but we've long ago passed the point of good enough performance on the desktop for most users and will likely be there soon in the ARM and mobile space. The A5X in my iPad 3 is a joke in terms of benchmark performance compared to any non-Atom CPU from the last 5 years, but I've never once felt the need for more speed from it. The A6 was a solid first step in this direction, and I wouldn't be surprised to see something much more interesting from Apple in a few years.

      • NeelyCam
      • 7 years ago

      They don’t have to beat Intel/AMD, as long as they are close enough – Apple products don’t have to have the best performance to sell well. And replacing Intel’s expensive parts in MBAs and MBPs with in-house chips would cut cost a lot, adding to profit margins.

        • deathBOB
        • 7 years ago

        Nailed it.

        • sschaem
        • 7 years ago

        Its not about beating Intel, its about even coming close.

        “The new macbook.. 2 time slower but $80 cheaper! its magical.. small print: none of your software work”

        Yea, Apple strategic team is dying to make this happen, that they hired a 16 year AMD veteran now working at Samsung…

        Get real. and imagine an emulator running x86 on ARM.. yeah that GREAT for battery life and performance.

        Time to think about the scenario… it makes no sense. its a loose loose situation for Apple.

        Again, you boot you ARM macbook.. whats software do you install first ???

          • NeelyCam
          • 7 years ago

          Ask yourself: does it really matter if the performance is 50% slower? It’s still “Good Enough”. And if that comes with lower power consumption, all the better.

          Software can be rewritten – that’s what they did with the PowerPC->x86 switch. It’s the sales volume that really matters.

            • sschaem
            • 7 years ago

            Do you have any proof that broadwell on 14nm is 50 % less power efficient to run desktop software then a potential ARM SoC ?

            And how long before we seen all the software suite re-written for ARM ?

            Untill then 2 thing happen… no software running on the new macbooks, or they run under emulation.
            Emulation KIlLS performance per watt.

            So even if Apple could build a laptop SoC that consume half has much as Intel 14nm design,
            this would be all lost. End result, apple end-up with a slower platform consuming more power.

            The only gain is loosing not paying Intel for the markup, but instead send that directly to TSMC or samsung to fab its magical PC centric SoC.

            Apple will not replace its x86 macbook with ARM based models. But I can see ipad getting some super ‘docking stations’… taking a page from asus transformer pad.

            • NeelyCam
            • 7 years ago

            I said Apple-designed chips might be “50% slower” – not that Intel chip would be 50% less power efficient. No – I don’t have proof that Broadwell would be 50% less power efficient than an Apple ARM chip, but you don’t have proof either that it’ll be 100% faster. If you do, please link.

            [quote<]Emulation KIlLS performance per watt.[/quote<] Emulation may hurt efficiency, but it's a stop-gap measure while waiting for rewrites (much like what Intel is doing for cell phones). [quote<]So even if Apple could build a laptop SoC that consume half has much as Intel 14nm design, this would be all lost.[/quote<] So, you imply that emulation essentially doubles power consumption. Do you have any proof of that? [quote<]The only gain is loosing not paying Intel for the markup, but instead send that directly to TSMC or samsung to fab its magical PC centric SoC.[/quote<] Intel has notoriously high profit margins, and part of that is coming from the design - not just manufacturing. Apple designing its own chip would save it some money just because of that. Moreover, with Apple's volume (and will/ability to pay in advance), I'm sure they can negotiate a pretty good deal with Samsung or TSMC where Apple is getting silicon cheap, and the volume funds Samsung/TSMC manufacturing R&D. Apple has so much money that they could even join Crolles2 alliance, build their own fab and start making their own chips. [quote<]Apple will not replace its x86 macbook with ARM based models.[/quote<] Famous last words.. You didn't address my key point that "good enough" performance is good enough for an Apple product, and they will still sell well. Agree or disagree? My view is that Apple doesn't have to compete on benchmarks with other PC manufacturers - their customers don't make buying decisions solely on that.

            • Beelzebubba9
            • 7 years ago

            Traditional CPU Benchmarks have been irrelevant for most users for years – pretty much any Core 2 or better PC with a fresh OS install and a SSD feels fast, and that’s what pretty much everyone cares about. My iPad has about 1/20th the CPU power of my overclocked Core i7 desktop, but anything it can do, it can do about as fast as I’ve ever needed it to. That feels far more like the future than using edge usage cases to justify the superiority of one solution over another.

            But we nerds tie our identity to the products we own, so that’ll be our mantra forever I suspect. 🙂

            Apple has the time and money to build an alternative to what Intel is offering, so it makes total sense for them to make sure they have an alternative should their roadmap diverge from Intel’s. It should also be noted that Intel can be a nasty company to work with, and any of us PC veterans should be able to remember the time when Intel would withhold CPU shipments from Vendors selling AMD CPUs to apply pressure and keep them an all-Intel shop. It really only makes sense for Apple to pursue an alternative source of supply, even if they never end up using it. It’s not like research and development into faster ARM CPUs will go to waste for them.

      • destroy.all.monsters
      • 7 years ago

      I don’t know why people are assuming that this is necessarily for a laptop or desktop he may well just end up as project lead on their A-series of ARM chips for mobile devices. They may be looking at product segmentation along the lines of Win 8/RT but I doubt it.

      For a more budget offering that runs IOS perhaps. Regardless it’s well known that Apple has traditionally put engineers on projects just to ensure a second source and bargaining power – “Star Trek” when Jobs was there the first time and hedging his bets on PowerPC with a team devoted to having a version for intel the last.

      Not worth getting worked up over no matter how you slice it.

        • sschaem
        • 7 years ago

        I do see this guy possibly working the A team, maybe in the goal to make a next generation iTV with console class performance.

        Maybe a docking station for iOS device that leverage a local more power hungry version.
        Dock your ipad and now you have a 3ghz quad core ARM with 8meg of L3 cache, etc.. to run your office apps.

        But frankly making ARM based OSX macbooks sound unrealistic.

          • destroy.all.monsters
          • 7 years ago

          “But frankly making ARM based OSX macbooks sound unrealistic.” I agree but with all the money Apple has it wouldn’t surprise me at all if they had something in the labs, show it to intel when it’s time to negotiate to get a better deal; and a lower priced laptop running IOS doesn’t seem like the dumbest idea in the world either.

      • blastdoor
      • 7 years ago

      A few thoughts:

      1. If Apple were to put their own ARM-based SOC in a consumer-oriented Mac, I doubt that it would be in the Haswell timeframe. For one thing, I’m sure they’d want it to be a 64 bit CPU rather than the current 32 bit for ARM SOCs. So if it happens, it will probably be 3-5 years from now.

      2. The marginal practical benefit of faster CPUs in PCs/Macs has been in steep decline. That’s why people don’t upgrade as often and it’s why Apple could get away with a Core2Duo in the MacBook Air for as long as they did. The upshot is that Apple doesn’t need to beat or even match Intel in terms of performance in a consumer-oriented Mac.

      3. Software may not be that big of a deal. The most painful transition for Mac developers in the last decade was not the transition from PowerPC to Intel — it was the transition from Carbon to Cocoa. But the transition to Cocoa is now complete, which means that recompiling from Intel to ARM might not be so hard for most developers. Remember that on the Mac, you develop apps one way and one way only — you develop them Apple’s way, using Apple’s development tools and compilers. This makes CPU transitions much easier for Apple and Apple’s developers.

      4. I could actually imagine Apple putting a 64-bit ARM CPU in the Mac Pro before the more consumer-oriented machines. Three reasons: (1) Xeons are ridiculously expensive, (2) Apple could use the same chips in their data centers as in the Mac Pro, and (3) at the high end per-core performance is often less important than total performance, which means that Apple could throw in more cores to compensate for lower per-core performance.

    • MadManOriginal
    • 7 years ago

    [url=http://i108.photobucket.com/albums/n4/RobertOak/rats_leaving_ship.jpg<]So sad...[/url<]

      • sschaem
      • 7 years ago

      Yea, Samsung is in big trouble !

Pin It on Pinterest

Share This