Analysis: "Are we on the edge of another PC revolution ?"


First, lets focus on why the history of apple as a company put them in the position they're in today where they build everything in-house and it seems to work so well for them. Apple has the upper hand here when it comes to optimizing the software and hardware in a way that Google can never have, because Apple is calling all the shots when it comes to OS, CPU design, and device design. Google doesn't have that luxury.

Google builds one piece of the handset (OS) and have to make it work in tandem with many other companies like Samsung, Qualcomm and Intel (for the radio). This is a very difficult task and is why OEMs like Samsung often have to also contribute a lot on the software side when building something like the S8.

The reason Apple is in this position (where it can control the entire hardware/software creation of the device) is twofold. On the one hand Steve Jobs always wanted to control the software and hardware aspects of the Macintosh because he saw that it made it easier to provide users with better UX this way, and also the more control he could exert over the users the better.

The other fascinating and often overlooked but incredibly important reason why Apple can do what they do with the iPhone has to do with IBM, PowerPCs and a little known company called P.A. Semi. You see, up until around 2006 Apple used PowerPC CPUs (by IBM) instead of x86 (by Intel). It is believed by most that Apple switched to Intel because Intel made more powerful chips that consumed less power. This isn't actually completely true. IBM is who made PowerPC design/chips and by the time 2006 rolled around IBM had sold off thinkpad, OS/2 had failed and they were almost fully out of the consumer space. IBM was completely focused on making large power hungry server class CPUs and here was Apple demanding small power efficient PowerPC CPUs. IBM had no incentive towards making such a CPU and it got so bad with Apple waiting on IBM that they ended up skipping an entire generation of PowerBooks (G5).

Enter P.A. Semi. A "startup for CPU design" if there ever was one. This team seemingly came out of nowhere and created a series of chips called PWRficient. As IBM dragged its feet, this startup took the PowerPC specification and designed a beautifully fast, small and energy efficient PowerPC chip. In many cases it was far better than what Intel had going for them and it was wildly successful to the point where the US military still uses them in some places today. Anyway, their PowerPC processor was exactly what Apple was looking for, which came at a time when IBM had basically abandoned them, and Apple NEEDED this very bad.

So what did Apple do? they bought P.A. Semi. They bought the company. So at this point if you're still reading my giant block of text you're probably wondering but if Apple bought the company who could solve their PowerPC problem, why did they still switch to Intel? And that's where the story goes from just interesting to fascinating: Apple immediately put the team they had just bought in charge of creating the CPUs for the iphone. See, people always ask when is Apple going to abandon the Mac?well the real answer is that they abandoned the Mac when they switched to Intel, because this was the exact time when they not only gave up but abandoned a perfect solution to the Mac's CPU problem, and where they instead re-purposed that solution to make sure that they never have a CPU problem with the iPhone.

So what lessons did Apple learn here? That if a critical component to your device (i.e. CPU) is dependent on another company then it can throw your entire timeline off track and cost you millions in revenue lost (the powerbook g5 that never happened). Apple was smart enough to know that if this was a problem for the Mac it could also be a problem for the iPhone. When a solution arrived for the Mac they instead applied it to the iPhone instead, to make sure there was never a problem.

And that team from P.A. Semi has designed Apples ARM CPUs for the iPhone ever since, and they're at least two generations ahead of the chips Android devices generally use, because they were first to market with a 64bit architecture, and first to allow the use of "big" and "little" cores simultaneously.

And as for Mac users? Well, the switch to Intel allowed the Mac to keep living, but MacOS now comes second to iOS development, and new Mac hardware is quite rare. Apple has announced plans for app development that is cross compatible with iOS and MacOS. Apple has started shipping new Macs along with a second ARM CPU. The iPad Pro continues to gain MacOS like features such as the dock, file manager, multi-window/split support. All signs point to MacOS being on life support. When Steve Jobs introduced MacOS he said it was the OS we would all be using for the next 20 years, and guess what? Time's almost up.

And the irony of it all is that history has now repeated: Apple now has the same problem they had with IBM, but now with Intel. Intel is now failing to produce chips that are small enough and that run cool enough. Apple will have to redesign the internals of the MacBook to support 8th gen chips due to changes intel made. Even the spectre/meltdown bug. The Mac is yet again dependent on a CPU manufacture in a way that harms Apple.

So yes, the iPhone is something to marvel at in terms of its performance. You might be thinking Android is the big loser here, but really it's the Mac and it's Intel. I believe we at the cusp of an event that will make the IBM/PowerPC drama seem small. In five years from now we likely wont even recognize what MacOS and Windows are anymore, and Intel will either exit from the portable consumer space, or they will have to go through an entire micro-architectural re-design and rescue themselves as they did in '93 with the Pentium.

In '93 Intel almost got destroyed because their CISC chips weren't as powerful as RISC chips such as PowerPC. Intel then released Pentium, which is essentially a RISC chip (think PowerPC or ARM) but with a heavy duty translation layer bolted on top to support CISC instructions that every Windows PC required. This rescued Intel up until right now but the industry has evolved and Intel's "fix" in '93 is now their biggest problem for two reasons: 1) they physically can't compete speed/heat/size with ARM now because they have to drag along this CISC translation layer that ARM doesn't need; and 2) Windows is about to introduce native ARM support with a software translation layer. Remember, Microsoft has the same CPU dependency problem that Apple has. And Microsoft's software solution allows them to throw away Intel for something better. Users won't notice the switch to ARM because it's transparent, but they will notice the 20 hours of battery life and thinner devices they get in the future once Intel is gone.
Analysis: "Are we on the edge of another PC revolution ?" Analysis: "Are we on the edge of another PC revolution ?" Reviewed by Kanthala Raghu on January 30, 2018 Rating: 5

No comments:

Song of the Week: Charli XCX - Boom Clap

Powered by Blogger.