In a surprising move, Apple unveiled the new iPad Air 4 with an unexpected processor: la puce A14 Bionic. The reason behind this is that it has been a very long time since an iPad had a processor before than the iPhone. Specifically, the original iPad and iPad 2 did this with the A4 and A5, respectively.
That was ten years ago and it is happening again now. Clearly, due to the global coronavirus pandemic and the delays caused in the completion of the manufacturing phase of the expected iPhone 12. Whatever the reason, the truth is that the team that designs silicon for Apple products took over from Intel in pushing Moore’s Law into the future. And there is no doubt about it.
Doubling (almost) the number of transistors in Apple’s latest “Axx” chips
The evolution of the iPhone CPU and GPU has been dizzying throughout its history. One aspect we left out at the time was the number of transistors that make up these processors. Over the past seven years, its increase has followed this path:
- 2013 A7: 28 nm and 1 billion transistors.
- 2014 A8: 20 nm and 2 billion transistors.
- A9 from 2015: 14 nm and 2 billion transistors.
- 2016 A10 Fusion: 14 nm and 3.3 billion transistors.
- A11 Bionic from 2017: 10 nm and 4.3 billion transistors.
- 2018 Bionic A12: 7 nm and 6.9 billion transistors.
- 2019 A13 Bionic: 7 nm and 8.5 billion transistors.
- 2020 A14 Bionic: 5 nm and 11.8 billion transistors.
The number of transistors is the number that measures the capacity of a processor. Intel co-founder Gordon Moore made a prediction that has come true so far. Known as Moore’s Law (though it strictly isn’t), in 1965 he claimed that the number of transistors on a chip would double every two years. Years later, Moore’s or Rock’s Second Law was drafted, which states that the cost of making a chip doubles every four years.
What’s interesting about this law is that it applied to Intel for decades. It’s well known TIC Tac, through the one that combined a reduction in the manufacturing process with a change in microarchitecture every other year, collapsed in 2016. There, the magic that drove Intel to double the number of transistors every two years stopped.
If we go back to the top list, we will see that it almost coincides with when Apple starts doubling transistors of its own chips every two years. From 2015 to 2017, they went from 2,000 million to 4,300 million transistors; from 2016 to 2018, from 3.3 billion to 6.9 billion; from 2017 to 2019, they go from 4.3 billion to 8.5 billion; and from 2018 to 2020, from 6.9 billion to 11.8 billion transistors.
The dizzying evolution of the iPhone CPU and GPU over the years on the doorstep of the iPhone 12
Apple has been designing its own chips for a decade. But without a doubt, I couldn’t have reached this milestone without the existence of TSMC as a manufacturing partner. In view of this, it must be said that it is not all about the number of transistors.
Customization on demand for custom devices
Our aim is to manufacture chips with cutting edge performance, powerful custom technologies and extremely efficient use of energy so that each of our products is best in class.
Tim Millet, vice president of technology at Apple, opened the section of the keynote dedicated to the A14 chip of the new fourth generation iPad Air. The truth is it wraps up in one sentence (long) Apple’s strategy for its developed processors at home. This is not something new, but it is advisable to collect it to understand Apple’s strategy with its chips.
While other companies must produce chips that appeal to as many customers as possible, Apple only needs to meet its own needs.
Unlike Intel, Qualcomm, AMD or Nvidia, Apple does not sell chips to third parties. You don’t need to sit down with them every year to see what needs they will have in the future. Where do they want to go? Nor how to combine them all in a balance that satisfies as many customers as possible. Maximize profits, reduce costs, etc. You don’t need it, because you are your own customer.
The company puts in these chips what it can and aspires to create in a roadmap development that she chose herself. It doesn’t depend on others be told where to go. This ballast it doesn’t exist, so you can choose your own path. And that’s what Apple has done with its “A” series chips.
Just take a look at the screenshot above to see what kind of tech we find in these processors. It has sections of the chip dedicated exclusively to computational photography, professional video encoding and decoding, encryption acceleration, machine learning accelerators and 16-core neural engine, among others.
Intel’s straightjacket and the journey to Apple Silicon
Both approaches are valid and have advantages and disadvantages. The one applied by Apple allows you to stand out from the crowd and create a series of products that aspire to be the best in their class. This personalization becomes an insurmountable advantage in the market of clothing, where energy efficiency is more important than anywhere else. One only has to look at the Apple Watch Series 6 and its S6 chip to realize just how much of an advantage it has over its (distant) competitors.
Now that we are at the gates of Apple Silicon, let’s see what happens with Macs, of course their performance, efficiency and custom features are promising.