First Ring Daily 1166: ARM before the Turkey Leg

Posted on November 23, 2021 by Brad Sams in Podcasts, First Ring Daily with 11 Comments

On this episode of First Ring Daily, ARM doesn’t deliver, Apple fights Samsung, and Turkey is nearly here.

RSS | SoundCloud | YouTube iTunes | Google Play

Tagged with

Join the discussion!


Don't have a login but want to join the conversation? Become a Thurrott Premium or Basic User to participate

Comments (11)

11 responses to “First Ring Daily 1166: ARM before the Turkey Leg”

  1. johnnych

    How has Paul already determined that ARM isn't the future of computing? Has Paul seen how many watts and cooling is needed to keep Alderlake running? Intel is going in the opposite direction here...

    The performance-per-watt metric is crucially important for the future of mobile computing, for tablets and laptops.

    I personally love the beautiful, clean, simpler design of the ARM architecture and customers definitely notice when their Intel laptop heats up, drains the battery, and spins the fans!

    Sent from my Macbook-Pro-M1-Max-14-inch-ARM-based laptop (Intel free since pre 2023!)

    • F4IL

      I'd argue that PerfPerWatt is crucially important to servers and desktops too. There is a power and thermal limit to how many cores can be clustered together on a single die and how high those can be clocked.

      The fact that msft have failed to effectively transition their platform to ARM is probably irrelevant.

      • johnnych

        That's a very good point as well - we rarely think about it but data centers are where a huge amount money gets spent (and soon to be saved) with these more efficient, scalable, and performative ARM SoCs...

        • F4IL

          It is also important to note that the simplicity and elegance of the architecture (e.g ARM or x86) is not all that important especially when it comes to end-users and benchmarks. The bottom line is usually speed. Besides what's a few more watts for an added perf boost in the long run?

          The biggest advantage ARM has and will undoubtedly continue to have over x86 is competition. x86 has been a nominal duopoly with an effective monopoly closely guarded by Intel and AMD for at least 2 consecutive decades. The incentive for healthy competition was and still is very small and as a result consumers are paying a high price for inefficient security riddled chips. Eventually things will change but for now, the PC will have to get used to being second class.

        • Paul Thurrott

          I have made this point many, many times.

    • Paul Thurrott

      lol ah man.

      Paul hasn't "determined" anything. Paul is of the opinion that Intel (and maybe AMD) will adapt their chipsets with ARM-like characteristics that will make the shift to ARM unnecessary.

      I wrote about this opinion here:

      • bkkcanuck

        It might be able to be improved, it has happened before... but I think the x86 hardware decoder that translates the x86 ISA into RISC based micro-ops which can then run on an underlying RISCish architecture (parallel pipelines and branch prediction is hard to optimize if one instruction takes 1 clock cycle, and anther takes 50+ clock cycles) -- adds a level of inefficiency into the chip design. My gut is saying this or something like this is inherently why Intel has never (now trying for a couple decades) been able to really implement a reasonably performant low power chip. As long as Intel had the lead in fab technology, this was mitigated -- but Intel has lost that lead. I would bet that Intel will eventually straighten out their fab situation and become competitive, but they likely will never get back to be industry leading ahead of TSMC (at least for the foreseeable future). Basically, if it was doable to compete in efficiency using the legacy x86 instruction set, Intel (which have some great engineers) would have already been successful.

  2. bkkcanuck

    You mentioned DEC Alpha... The result of DEC being acquired by Compaq/HP (I think) was that the chip guys left and that was the beginning of PA Semi - which became part of Apple and the core of their chip business.

  3. nbplopes

    Hi Thurrot, I don’t agree with your Industry wide assessments. Also with your conclusions in many reviews regarding PC laptops when cross referencing with mine or with people familiar with. For instance in an M1 start the day fully charged and connect it in the evening. With an equivalent powerful Intel laptop I start fully charged and at 12:00 to 13:00 need connect it to charge … doing exactly the same kind of work … yet in noisier surface.

    On this subject is no different.

    Having said this, don’t think Windows is going anywhere soon even if Microsoft does not really want to adopt and integrate more “advanced” hardware architectures offering better performance for watt, as well higher processing power in general.

    Windows ecosystem is totally entrenched in the Enterprise in so many way so that the bottom line Windows on ARM does not matter as long as Intel comes out with ok solutions … there is no financial incentive for Microsoft to adopt anything else … they will keep on coming up with “experiments” and feeding the show with them. This not like smartphones back in day were Windows (CE) was not entrenched anyway whatsoever in the Enterprise, was niche, users could device for whatever, so Microsoft could not rely on anything but their technical capacity … and well everyone knows how that went.

    The only way Microsoft might move forward in this regard is if eventually it starts feeling the bite in the Enterprise. That takes much more than powerful devices given the context.

    Hey, but it does not matter .. “Qui ser sera”.

    PS: The single point of failure of all Windows PC and Mobile experiments outside the Intel and AMD realm across multiple vendors is one … Windows and Microsof. Because these multiple hardware vendors and OEM seam to have success with other OSs using the same kind of metal solutions. The reason is simple, there is no clear incentive for Microsoft to consider this more than experiments and to put on a show around it and be able to say “Hey we also do it”.

    • Paul Thurrott

      It's spelled Thurrott. Also, my name is Paul.
      • nbplopes

        Hi Paul. Thank you for the typo correction. I agree that names are important, especially people.

        If I offended you … my apologies. Paul it will be from now on.