[ad_1]
A few years There has already been – the company does not say exactly when – some Apple engineers began to think that the camera of the iPhone could be made smarter by using new machine learning algorithms known as neural networks. Before long, they were chatting with a skinny vice president named Tim Millet.
Millet leads a team of chip architects, who worked. When the iPhone X was unveiled last fall, Apple's camera team added a new, uncluttered portrait mode that could numerically adjust the lighting of the faces of the subjects and cleverly scramble the background. He took advantage of a new module added to the main chip of the iPhone, the neural engine, designed to execute machine learning code. The same new specialized silicon allowed the iPhone X's new facial recognition unlock system, the face ID. "We could not have done that properly without the neural engine," says Millet.
IPhone engineers could use internal chips to create an innovative feature, such as Face ID, which demonstrates the benefits of Apple's unconventional hardware strategy. Most computer and gadget manufacturers buy chips at the heart of their devices from semiconductor manufacturers such as Intel, Qualcomm or Samsung. In contrast, the iPhone 4 of all iPhone since 2010 is powered by a chip designed by Apple, made to order by an external company.
Millet says the approach stems in part from former CEO Steve Jobs, who believes that standard chips limit the dreams of his gadget designers. "It's about owning critical parts and not letting anything pass," says Millet. "The experiences we deliver over the phone depend primarily on the chip."
The rise of Apple as a chip designer was evident last week when the company unveiled three new iPhones at the Steve Jobs Theater, the underground auditorium of the company's new headquarters in Cupertino, California.
The three devices embark a new chip designed by the Millet team called the A12 Bionic. It is manufactured with more advanced chip technology than the equivalent chip of any mobile device on the market. The A12 transistors have features as small as 7 nanometers, especially smaller than the 10-nanometer transistors of last year's iPhone. This helped Apple to store 6.9 billion transistors, compared to 4.3 billion last year.
The luxury of a larger number of transistors has opened new creative avenues for the Millet team. The way they were deployed suggests Apple's priorities for the iPhone. The engineers made the graphics processor of the chip much more powerful. they also created a much larger neural engine, to enhance the functions of artificial intelligence. The neural engine last year was capable of 600 billion operations per second, the new design can achieve $ 5 trillion per second.
Millet says that these upgrades helped improve portrait mode, which allows users to adjust the depth of field after taking a photo, as well as more accurate and realistic augmented reality experiences. The neural engine is now open to external developers, along with recent software updates to inspire new applications based on artificial intelligence. "This is an emerging class of applications that is extremely important," explains Millet.
Google thinks so too. The company included a specialized image processor, designed to run neural networks in the Pixel 2 phones it launched last October. Google designed the component for its devices in conjunction with Intel, but bought a main chip ready to use from Qualcomm.
Apple's freedom to co-evolve its silicon with its software, hardware and corporate strategy is especially valuable now that iPhone sales have reached a plateau. The company needs new features to entice iPhone owners to upgrade and support price increases that have increased revenue. "They can offer a smoother experience and introduce new features sooner," says Patrick Moorhead, Semiconductor Analyst at Moor Insights & Strategy. "It helps them get a high price." Samsung manufactures both smartphones and smartphones processors, but these units are not closely related to those of Apple, and the South Korean company also sells its processors to other gadget manufacturers.
Apple will not say who makes its new A12 chips. The semiconductor industry says it's Taiwan's TSMC. At the TSMC event last October, Apple's chief operating officer said that TSMC was the only supplier of new chips for the iPhone and iPad, congratulating it on producing half a billion Apple chips in less than a year.
To ensure access to TSMC's 7-nanometer technology, Moorhead claims that Apple has made significant financial commitments. "They make capital expenditures to buy fab space and get first place," he says. Apple plans to spend $ 17 billion on capital expenditures in the year ending September 29. That's more than eight times the amount spent in 2010, when Jobs announced the first chip designed by Apple in the iPhone 4.
Apple is the most valuable company in the world, but its chip designers will soon be facing even greater power: the laws of physics. The semiconductor industry is reasonably confident of being able to deliver 5 nanometer transistors by 2020. Reducing their size is less obvious. Moore's Law, the name given to the trend of increasingly longer transistors for decades, has slowed down and could be over.
Apple's strategy can still work if, or when, the transistors stop shrinking. Chip design could become the primary way to get the most out of silicon chips, and Apple's end-to-end control of the iPhone should offer more flexibility.
Millet refuses to answer questions about his team's plans, although he notes that the team is already looking well beyond the A12 announced this week. "It takes us about two years to build a chip from start to finish," says Millet. Somewhere inside Apple's ring-shaped seat, the hardware that will power next year's new features is already on the horizon.
Biggest cable stories
Source link
