Last year, Apple surprised the world with the A12 Bionic: the CPU of the XS and XR iPhones had performance levels basically unprecedented in the mobile world, and even surpassed the performance of the iMac Pro in some metrics. It was to be expected, therefore, that your successor would follow this lineage of power at all costs, right?
Well… yes and no. THE A13 Bionic (which equips the iPhones 11 and 11 Pro) does surpass its predecessor in absolute numbers of performance, as proved by benchmarks preliminaries of the new devices. Still, Apple’s focus on building the new CPUs wasn’t on performance – instead, the Apple’s engineering team’s focus was directed to another crucial aspect: energy efficiency.
In a long report by WIRED focused on the new chips, Apple’s senior vice president of global marketing, Phil Schiller, and the head of processor engineering, Anand Shimpi, told some details in the process of creating the A13 Bionic. According to Shimpi, the focus on energy expenditure was very clear from the beginning of the process – and it ends up being beneficial, too, for CPU performance.
Publicly, we talk a lot about performance, but in reality, our main focus is on performance by watt. We see the whole thing from the angle of energy efficiency, and if you build an efficient design, you end up producing a well performing design as well.
Apple’s obsession paid off: according to information brought by the report, the A13 Bionic is about 30% more efficient, energetically speaking, in relation to its predecessor. In addition, he has a 20% higher performance in relation to the A12 Bionic, in all areas – be it the CPU, GPU or Neural Engine, responsible for the machine learning part of the devices.
The new Apple chip, as already reported, is built on a 7 nanometer architecture (just like its predecessor) and has 8.5 billion transistors. There are six nuclei: four of them, nicknamed «Thunder» (thunder), are focused on energy saving and perform less demanding tasks; the other two, called «Lightning» (lightning), run at 2.66GHz and take care of the heaviest and most complex tasks, when necessary.
Apple’s secret to cheaper chips, however, goes beyond the division of labor between cores. The company’s engineers study, year after year, how users use their devices and how apps and resources interact with the CPU / GPU, optimizing these relationships in future designs. According to Schiller, machine learning is essential for these studies:
Machine learning is running all the time, whether managing your battery or optimizing performance. There was nothing like that running ten years ago. Now, he’s there all the time, doing things.
In other words: if the development of the chips continues at this rate (and machine learning continues to do its job properly) … well, the next few years may hold very pleasant surprises in the area of mobile processing. Great, isn’t it?