name99

dewme said:

Apple is like a marathon runner trying to establish a new world record- they are racing to achieve a personal goal that they have set for themselves. The fact that there are other competitors in the race is incidental. 

If Apple was only racing to beat the competition they could get by with only being marginally better than their closest rivals. 

Competition is good, but for companies (and people) who are focused on excellence rather than simply winning, competition isn’t a good enough benchmark to get you where you need to be. 

Let the Samsung’s and tech pundits of the world focus on the excitement of what they perceive as a race, which is in reality a breathless quest to see who’s going to finish in second place. 

This is, I think, a more perceptive comment than most realize. 
I've frequently talked about the difference between Technical and Financial companies. Technical companies don't fear the future, don't fear change, because they are confident that they can make and direct change, that whatever comes next will be to their benefit. Finance companies are terrified of change because all they feel confident in is the ability to keep doing what they're doing today. 

Few companies, even those nominally in "Tech" are Technical companies, companies that aren't scared of change, scared of disrupting themselves. nVidia kinda sorta. AMD right now. Intel back in its glory days, from about the 386 to about Sandy Bridge. 
When engineers at Apple propose "OK, for the A15 our goal is another 25% performance improvement, and we figure let's add SVE2 to extend our wide vectors to integer support", management is saying "Right on! Make it so!". 
When engineers at Intel propose "OK, 7nm is so dense, how about we put AVX512 and 4 cores everywhere", immediately the discussion turns into "well how will that affect our Xeon sales? And what if it reduces the prices we can charge for mid-range? How about you cripple the AVX512 support by making the scatter gather really slow? Or how about we give them lots of cores, which sounds good in the ads, but we don't give them enough DRAM controllers to actually run all those cores optimally?" 

And so it goes. The Apple engineers (at least for now, who knows how long it will last) have carte blanche to make the best product they can, year after year. The Intel engineers are not allowed to do ANYTHING that might make a product at level n of the 29 step marketing segmentation an acceptable, cheaper, replacement for level n+1. 

"In retrospect, it's funny that outside observers were stating at the time that Apple wasn't moving quickly enough, that it should have opened the App Store a year earlier, that it couldn't afford to delay its macOS release by even a few months, or that it really should be scrambling to get Sun Java or Adobe Flash working on iPhone, or working on some other priority such as MMS picture messaging. "

Thank goodness nowadays we have a better class of outside observers, with much more thoughtful insights as to what Apple should and should not be doing...

Thanks for the article. I look forward to the rest of the series. 
My one suggestion would be that I'd like to hear more of the PA Semi side of things. 
Obviously CPU performance is the product of both frequency and IPC (ie a smarter design that can perform many instructions per cycle without wasting much time on dead cycles). Intrinsity seems very much geared to ramping up frequency, but longer term Apple seems to have decided (correctly, IMHO) to concentrate on very wide, very high IPC designs that don't strive for maximum frequency.
You'd figure what's needed for this decision (the mindset/confidence to go down this path, and the necessary tools like simulators) must have come from somewhere. So did they come PA Semi; or were the tools created pretty much from scratch inside Apple, while Srouji and the other engineers took a massive gamble on the very wide approach to mobile CPU design based not on experience but just on seeing the likely trends of Si going forward? (ie transistors are getting cheaper and cheaper, but not much faster)