Apple and ARM - A Year Later

3 minute read

So…now we have the M1 Pro and M1 Max – two very powerful processors and some of the first really excellent ARM-based end-based professional systems. I’ve had an ARM laptop for a while, a Pinebook Pro, and while I find it a nice platform on which to work (especially with elementary OS installed) it is not in the same league. Not at all. I seriously covet the new MacBook Pros.

One of the promises of ARM-based systems has been good power consumption, and we saw that delivered with the original M1-based Macs, most notably with the fanless MacBook Air. I think it truly delivers on the vision that Steve Jobs presented on the Air from the beginning – an “all day” laptop that has performance enough to not break a sweat doing what most people do. My Pinebook Pro also delvers good battery life (not as nice as the Air, but still very good) and with further integration into the RockChip SoC could turn out to be even better over time. However, where ARM processors have always excelled is when the SoC is developed very closely with the software. Mobile phones proved this, both iOS and Android platforms provide incredibly fast computers that last for days. However, bringing this to the “generic” PC experience has always been challenging. This is because the platform has to be more open and more flexible, which means software isn’t tied as close. You have only two viable ways to solve this:

  1. Limit what runs on the machine, making it work more like mobile phones do (this is the approach Microsoft took with Windows S)
  2. TIGHTLY integrate the OS and the SoC and design them together so that they can manage everything working on them (the Apple approach with M1)

I think we have seen that Option 2 is the superior option at delivering results. This is not meant to be a dig on Microsoft – in fact, Windows 10 and Windows 11 are excellent OS’s and if MS decided to go the CPU-design route (either in-house, or to work on a bespoke design with another firm) then they could have these same gains. Every other attempt at making ARM-based “generic” computing devices, from my Pinebook Pro, Windows S machines, all the way back to the ARM Netwinders that has not taken this approach has not been able to succeed. Where ARM has succeeded – the original Acorn RISC boxes, Raspberry Pi, AWS Gravitron, Apple M1, etc. – is where there is either an open source OS that can be continuously and deeply be tailored to the SoC, or the combination of HW and SW being engineered together, which is almost an embedded systems approach (i.e. the approach used in mobile phones.)

Bring them on, I say! Let’s see more like this!

Categories: ,

Updated: