Macs may need ARM processors to survive

The heyday of personal computers is past. Those happy years up to 2014 when every shopping centre had outlets wanting you to leave the store with a new Mac or PC have sadly gone. According to Statista, desktop PC shipments are expected to decline from a worldwide total of over 150 million in 2011 to less than 80 million by 2023. As a successful vendor of personal computers, Apple faces a stark choice: innovate, or watch its sales slide inexorably into the sunset.

Almost inevitable requirements for innovation in personal computing are faster processors, both general CPUs and specialised GPUs. Apple has been placing great emphasis on improving software performance – it was the overriding theme at WWDC last year – but its newer technologies such as machine learning and augmented reality need hardware which has more cores and runs code more quickly than ever before. Ultimately, in some domains, the answer may eventually come in quantum computing, but that is still many years away, and may never miniaturise sufficiently to run inside a desktop system.

Take a look at recent hardware benchmarks for Apple’s latest models and one thing is obvious: iOS devices are now delivering as good and better performance than bulkier and more expensive Macs.

Using the Geekbench suite, for example, this £/$/€ 5000 iMac Pro with its eight cores of Intel Xeon and costly Radeon Pro Vega graphics card isn’t that much faster than one of the current iPad Pro models costing from £/$/€ 769 with its Apple A12X Bionic system-on-a-chip. The iPad’s four high performance cores return a multicore benchmark of just under 18K, compared with the iMac Pro’s eight cores with a multicore benchmark of over 33K – twice the number of cores for 180% of the performance.

Compare their graphic performance using Metal 2, and the iPad is slightly faster, with a benchmark of just under 18K compared with the iMac Pro’s 16K. The embarrassing fact now is that the hardware in Apple’s latest iPad Pro has CPU performance very similar to that of a MacBook Pro mid 2018 with a four-core i7 processor, and GPU performance which is already superior to almost any Mac.

Intel’s prices are also far from cheap. About a fifth of the cost of this iMac Pro is its Intel Xeon W CPU and supporting chipset. Supply of those is in the hands of Intel: delays in production of new processors result in delays to new Mac models which are designed to use them. Then there’s the problem of Spectre and Meltdown vulnerabilities, and the latest, appropriately dubbed Spoiler.

Apple’s own systems-on-a-chip could be the answer to most if not all of these issues.

Without them, Apple can only deliver better performance for cost by cutting its margins, which in a shrinking market might increase its share, but would risk reducing its revenues from the sale of Macs. With less money coming in, there’s less justification for continuing investment in macOS, and distinctive features such as the AppKit frameworks would wither even faster than they have.

For macOS to retain the support and investment that we, it, and the Mac needs, Apple has to innovate again, from the hardware up. It has to bring developers back with products which they know can’t run on lesser PCs because only Macs have the hardware to do it.

The cost is that antithesis to innovation, compatibility. The most compatible system with X is always X itself. When Apple launched its classic Macs, it didn’t adopt the Intel 8088/86 processor family, but Motorola 68K. Their page description language was PostScript, which was almost unknown in desktop printing at the time, and its PostScript printers didn’t connect via the then-standard parallel port but through Apple’s own network system, LocalTalk. Its interface wasn’t based on Windows 1.0, but peculiar to Apple. When Apple introduced its first external optical drive in 1988, everyone wondered what you might use it for.

If Apple were only to makes Macs which were compatible with everything else, they and macOS would quickly vanish. I much prefer the prospect of changing processor architecture to oblivion.

Postscript (appropriately!):

I am greatly enjoying comments and discussions about this. I should stress that I don’t know whether this will happen. However, I’d like to remind you that from 1997 to 2002, Apple promoted itself (under Steve Jobs) with the slogan Think Different, often incorrectly attributed as a quote by Steve Jobs. I don’t ever remember Jobs or Apple or anyone else saying Think Compatible.