It’s been more than a month since teardown shops like iFixit and Chipworks took apart and dissected Apple’s iPhone 7. The biggest surprise remains Apple's decision to dual source its LTE modems from both Qualcomm and Intel.
But there was one little chip that has gone mostly unnoticed. Inside the iPhone 7 is a field-programmable gate array, or FPGA, made by Lattice Semiconductor, according to Chipworks. An FPGA is a type of processor that can be reconfigured after it's been manufactured and installed in a device. Increasingly, these chips are are used for accelerating machine learning applications in data centers. This is the first time an FPGA has appeared in an iPhone.
"It is very unusual and intriguing," said Kevin Krewell, a principal analyst at Tirias Research. "Apple is good at committing to final pieces of silicon. A programmable chip is going to add extra costs to the build of materials. Not a lot of phones use FPGAs."
Apple is likely thinking ahead. Krewell thinks the FPGA inside the latest iPhone model could be used for running machine learning algorithms for, say, some advanced health monitoring feature that Apple hasn't yet introduced. It could do additional image processing for unreleased virtual (or augmented) reality feature. But the FPGA could also be a temporary stopgap solution that Apple eventually wants to address by adding a dedicated chip in future iPhones.
It's tough to figure out what function the FPGA is serving right now or whether the iPhone 7 is currently utilizing the chip. Apple certainly didn't spend any time talking about it at the iPhone 7's September launch event. Because of the reprogrammable nature of the chip, Apple may update the firmware on the iPhone 7 to change or engage the chip's function.
The specific Lattice chip Apple is using is the ICE5LP4K, which is designed for low-power devices like phones. The two big FPGA makers, Altera (now owned by Intel) and Xilinx, make more intensive FPGAs better suited for data centers. Such data centers are where FPGAs are starting take off, because of their ability to accelerate machine learning software. That's a big reason why Intel paid $16.7 billion for Altera last year. Intel is eager to maintain its dominance in data center processors, so it's starting to pair its server processors with Altera FPGAs. Microsoft has been investing heavily in making its own custom FPGAs to boost the artificial intelligence capabilities in its data centers.
"There's been explosive interest in FPGAs," said Krewell.
Apple isn't the only company to put FPGAs in phones. Samsung's Galaxy S5 phone from 2014, for example, had a Lattice FPGA. Samsung never revealed how the chip was being used, and dropped from the Galaxy S6 phone the following year. So FPGAs inside mobile phones remain a rarity.
With each new iPhone, Apple is embedding more and more artificial intelligence. The advanced camera capabilities in the iPhone 7, for example, come from computer-vision algorithms running on a new image signal processor, an in-house Apple design. The company is trying to differentiate from rivals like Google, which rely heavily on the cloud -- rather than the device -- for access to artificial intelligence. Apple's argument is that it's more secure and private to do some of the computing on the device instead having to route every piece of data to the cloud.
