Forget the stunning iPhone X screen or the beautiful glass back of the iPhones 8/8 Plus. Of all the elements introduced in Apple’s latest keynote regarding its smartphones, none is as decisive for the future as the chip A11 Bionic that equips them. The new processor, in addition to enabling all of Apple’s new bets on augmented reality and facial recognition, also represents Apple’s almost complete independence from partners – almost all of its elements were designed by the company itself.
Now, a story from Mashable brought new information about the chip and clarified some interesting points about its development process. The report also incorporates an interview with Apple’s senior vice president of global marketing, Phil Schiller, and the senior vice president of hardware technology, Johny Srouji, which offers a closer look at the events and the engineering team that made the A11 Bionic real.
One of the most interesting parts of the report / interview talks about the process of developing the Apple chips. According to the executives, it is at least three years from the beginning of planning until the premiere of components in prime time – that is, when Apple was announcing the iPhones 6/6 Plus with A8 chips, in 2014, the engineering team in Cupertino, I was already starting to put the first pieces of the A11 Bionic in place.
Of course Apple, like the gigantic company it is, has very different teams developing the same product simultaneously – so while an engineering team is creating the A11 chip, another software team is working on augmented reality technologies and another hardware is designing the end-to-end OLED screen. All of these teams, of course, need to have a synergy so that everything is at one point at some point. As the article stated:
Teams like Schiller’s marketing group and the screen team come to Srouji with orders, essentially ideas of what they think they will need in three years (“how can we implement a Super Retina display?”).
“The process is susceptible to change,” said Srouji, who has been with Apple since the first iPhone. If a team arrives with an order that was not part of the original plan, “we need to make this happen. We don’t say ‘no, let me go back to my planning and in five years I will give you something’. ”
Schiller and Srouji did not enter specific orders, but Schiller admitted: “There have been some critical things in the last few years when we asked Johny’s team to do something on a different schedule, on a different plane than they had years ago, and they moved skies and land to do it, and that’s impressive to see. ”
Regarding the A11 Bionic specifically, the chip brings several punctual evolutions in relation to its predecessor, the A10 Fusion, but also some totally new and developed aspects from scratch. For example, the component can now perform asymmetric multiprocesses, that is, activate anything from just one to all six cores simultaneously – saving the device’s battery and improving its performance.
THE Neural Engine – part of the chip responsible for tasks related to artificial intelligence – and the new graphics engine are also totally new; this second, in fact, is Apple’s first attempt to build a graphics processor after the noisy break with Imagination Technologies. Srouji said that Apple’s attitude of increasingly designing its own solutions is extremely beneficial:
If you look at our SoC, CPU, ISP, screen, all the places where we believe we can differentiate ourselves and offer Apple optimized value, we go there and do this thing. We have done this consistently for 30 years.
In relation to Neural Engine, Schiller made it clear that Apple has a very specific list of uses for the coprocessor – the idea is to prevent developers who want to use the component from only using it for a very specific amount of tasks, so as not to overload it. It is possible, for example, to use the engine for facial and speech recognition technologies, but not much else.
Speaking of the past / future of chip development at Apple, Srouji said the evolution is stark: ten years ago, the company made processors with 100 million transistors in a 65 nanometer process, while today chips with 4.3 billion transistors in a 10 nanometer process. The executive stated that his team is “world-class” and, when asked about the next frontier – since silicon is beginning to show signs of physical limit and large companies are already looking for an alternative, such as quantum computing -, from that :
We are thinking ahead. I can tell you that, and I don’t think we’ll be limited. It’s getting hard.