Ashraf Eassa: Apple’s silicon will overtake Intel’s in 4 months

Easy-to-follow analysis from a guy who knows his chips.

From How Apple Dethroned Intel As the World’s Most Innovative Chipmaker, posted Sunday on The Motley Fool:

Back in 2013, Apple introduced the A7 system on a chip (SoC) as part of its then-flagship smartphone, the iPhone 5s. The A7 was impressive for several reasons. Firstly, it was the very first 64-bit ARM processor to ever hit the market, which gave Apple some performance and efficiency advantages over its fellow mobile competitors (who, rather comically, tried to downplay the need for 64-bit chips as they furiously worked on their own).

What caught my attention at the time, though, was that the A7 delivered CPU performance at a frequency of 1.3 gigahertz that was very similar to chip giant Intel’s very best processor, known as Haswell, at the same frequency. Now, Intel’s chips, at the time, ran at much higher frequencies (in excess of 3 gigahertz), but what the strong per-gigahertz performance of the A7 chip signaled to me was that Apple had built a very impressive base from which to build up in future smartphone chips.

Fast forward to today, and Apple’s best iPhone and iPad processors deliver performance for CPU tasks — Intel’s specialty — that’s competitive with Intel’s best notebook computer processors but in sleeker, lower-power devices than what Intel’s chips can fit into.

I believe that when Apple introduces its next iPhone in about four months, it will deliver equal or better CPU performance to Intel’s best notebook processors designed to consume 15 watts but at a fraction of the power consumption. [More details follow.]

My take: If I didn’t already suspect it, Eassa would have convinced me. Thanks to friend-of-the-blog John Garner for the pointer.

See also: Ben Bajarin on Apple’s secret sauce

8 Comments

  1. David Emery said:
    Well, chip/computer architecture design has really stagnated in the last 25 years, with the utter dominance of first WinTel, and then ARM. About the only advances were made in low power support.

    0
    May 28, 2018
    • Steven Noyes said:
      I would never call the substantial (and they are substantial) gains in performance per Watt stagnated.

      1
      May 29, 2018
      • David Emery said:
        I see your point. But I think that’s much more of silicon implementation than it is of instruction set architecture. Those performance gains are made with the same instruction set architecture we’ve been kicking around for a long time.

        0
        May 30, 2018
  2. David Drinkwater said:
    I’m not an electrical engineer or circuit designer. I’m a semiconductor (sensor) manufacturing engineer. Respectfully, Eassa’s work has always smelled off to me. Apple may exceed Intel in very specific applications, but it would be amazingly difficult to win out over so many years (decades) of development. That said, there were people in the cellphone industry who said “Apple’s not just gonna walk in here…”

    1
    May 28, 2018
  3. Richard Wanderman said:
    It seems to me it’s less a matter of Apple or anyone else building a competitive processor to Intel for the Mac, it’s a matter of whether Apple can make MacOS run on an ARM chip and whether application developers will be able to follow along (or will want to).

    0
    May 28, 2018
    • David Emery said:
      Getting MacOS to run on ARM is not a big deal. I think I’ve seen reports there is an active product doing the work within Apple. Microsoft has Windows running on an ARM (and I’d bet the Windows code base is a lot more closely coupled to Intel architectures than MacOS is.)

      If the software is well-written, it -should be- just “recompile and test.” That being said, there are always “gotchas” (usually implicit assumptions you didn’t realize you made.) The ‘test’ part isn’t cheap, depending on how much the developer has invested in automated (regression) testing. The core question, I think, for a software product manager is whether s/he will have to support both Intel and ARM versions concurrently. That doubles the testing burden.

      Finally, I wonder how much dependence business users have on being able to run Windows under Parallels/VMWare. This was a life-saver for me a couple of times when I was employed, but my sense is the number of Windows-only business applications has declined (mostly being replaced by web hosted applications.)

      1
      May 29, 2018
      • Richard Wanderman said:
        I hope you’re right. Having gone through all of Apple’s processor upgrades on the Mac, I can personally attest that they weren’t as easy as Apple thought they’d be. For me, where was about a year of mess each time.

        Then there’s the issue of folks who have both desktop and laptop Macs. How do we buy and license software for both without going broke?

        1
        May 29, 2018
        • David Emery said:
          Well, it wasn’t as easy as Apple promised (I lived through 68k-PPC, and then PPC-Intel), but in my experience (and I did some studies on system portability), Apple still pulled off minor miracles. Other processor transitions have been catastrophes.

          0
          May 30, 2018

Leave a Reply