Original title: An in-depth look at Apple's six-core A13 Bionic processor, found in the new iPhones, which promises 30% more efficiency than last year's A12 chip
About 72 minutes after this year's iPhone launch, Phil, Apple's senior vice president of marketing,
When Sasanam's speech ended, all we could remember was the numbers. Apple's latest chip contains 8.5 billion transistors and six CPU cores: two high-performance cores (called Lightning) running at 2.66 GHz and four efficiency cores (called Thunder). A13 bionic chip has four-core graphics processor, LTE modem, image processor designed by Apple and eight-core nerve engine for machine intelligence function, which can perform more than 5 trillion operations per second.
The new chip is smarter, faster and more powerful, but consumes less power than the previous generation. The efficiency of the A12 chip is about 30 percent higher than last year's A12 chip, which is one of the important factors contributing to the five-hour increase in battery life of the new iPhone every day.
The launch of the iPhone 11 series only reiterates Apple's real advantage over its competitors, namely, its exclusive software, system hardware and chip design. You can see these benefits in many of the features of the iPhone, including Augmented Reality and Computational Photography, such as Deep Fusion and Night Mode.
Speaking about the A13 bionic chip and its functions, Schiller said:
Apple has come a long way since the first iPhone was launched in 2007. The company's first smartphone was slow enough to perform even the most basic tasks, such as copying and pasting text. The battery life of the original iPhone was bad and the shooting was not good. The original iPhone had few multitasking features, and its chip main frequency was only 412 MHz. The phone can be said to be the product of a patchwork, including the use of chip components in a Samsung DVD player. It's hard to imagine that such a device would subvert the entire mobile phone, computing and communications field.
Apple quickly realized that if it wanted to stay ahead of its competitors, especially those in the Android ecosystem, it needed to build the whole system, whether hardware or software. Apple's decision to design and build its own chips was made sometime in 2008. At that time, the company had only 40 engineers integrating chips from various suppliers.
Then, in April 2008, Apple bought a chip startup called P.A.Semi for $287 million. This brings the total number of chip engineers to about 150 and brings the most important expertise in mobile phones
Over the years, Apple chips have supported many amazing new features, such as smart assistant Siri, video calls, fingerprint and image recognition and camera capabilities, all of which are the result of Apple's progress in chip development. When iPhone X was released in 2017, a blog post said:
Apple's chip advantage has never been overlooked in the industry. Commercial chips have always lagged behind Apple, which has been working to enhance its chip advantages, as has been the case with every release of mobile phones and tablets. Huawei and Samsung quickly realized that the future of mobile technology would require custom chips, which would enable them to stay ahead of Android competitors and compete better with Apple.
Siri, vice president of chip business at Apple
These companies, as well as chip manufacturers such as Qualcomm, are engaged in a chip arms race and are shuffling their positions in the rankings. When the last generation of A12 bionic chips was released, Apple had a slight advantage over its competitors. This year, Apple used the launch of the iPhone 11 to strengthen its leading edge.
Speaking of the previous generation of A12 bionic, Gwennip pointed out that while Apple is in the lead in the single CPU competition, other companies are also competitive. He said:
So, since Apple launched A12 last year, have they stepped up their efforts to catch up? How on earth are the new six-core A13 bionic chips stacked compared with the latest chips of Apple's three major competitors? Let's look at the relevant figures:
Samsung's latest processor, Exynos 9825, has eight cores distributed in three clusters: two high-performance custom Mongoose kernels run at 2.73 GHz, the other two Cortex A75 cores run at 2.4 GHz, and the four efficiency focus cores Cortex A55 run at 1.9 GHz. It is equipped with Mali GPU and Samsung neural processing units, supporting LTE and memory functions.
Huawei's chip, Kirin 9905G, uses a similar method of three clusters and eight cores: two high-performance cores, Cortex A76, running at 2.86 GHz, and two A76 dual cores running at 2.35 GHz. Four efficiency-focused cores, Cortex A55, run at a slower 1.95 GHz. The complete chip is a 16-core GPU and a three-core Da Vinci nerve engine. Huawei's chip contains 10.3 billion transistors.
Qualcomm's new 555 Plus is very similar to Kirin 990 and Exynos. It uses custom Kryo 485 gold core. One of the powerful clusters runs at 2.96 GHz. The other three Kyro 485 gold core runs at 2.42 GHz. The four efficiency-oriented Kryo 485 silver core runs at 1.78 GHz. It includes Adreno GPU and Qualcomm Hexagon 690 AI engine.
These chips are equipped with faster components and a larger number, so you might think they perform better than Apple chips. But the reality is that we've hardly ever used the full capacity of chips in mobile devices, and one or two high-performance kernels are enough to meet most of our activity on mobile phones. Apple's six-core design may look backward compared with rival eight-core processors. But in fact, the two large processors on its chip easily outstrip competitors' designs.
Apple processors are more energy efficient, giving them a clear advantage over competitors. Samsung's Mongoose chips, for example, need to be used carefully to avoid overheating the devices that configure them. Even the newly designed custom efficiency kernel in A 13 is superior to its competitors. Earlier this year, Gwennip pointed out in the Microprocessor report:
Therefore, the conclusion here is that the specifications and benchmarks do not take into account Apple's real advantages, namely, close integration with devices, and the company's development strategy, which is to squeeze more running time out of batteries while improving key application performance.
So how can mobile phone companies demonstrate the benefits of these technologies in a way that resonates with customers? Words about chips don't matter. The most important thing is to have the best camera, the fastest cell phone and the biggest battery. The longer we use Instagram, Facebook or YouTube, the more willing we are to spend money on these high-end phones. How do Apple's iPhone 11 pro and iPhone 11 Pro Max extend their battery life by four hours and five hours respectively?
The answer to this question clearly shows that Apple has the inherent advantages of the system as a whole. In order to understand how vertical integration is reflected on A13 bionic chips, Schiller and Anand
The performance of the new A13 chip far outperformed last year's A12, with all its main components: six CPU cores, graphics processors and neural engines improved by 20 percent. For chips that are already high performance, seeing such a significant improvement makes people feel a little bit like looking at Usain
Both Simpi and Schiller emphasize this focus on energy efficiency and performance. For example, the CPU team will study how to use applications on iOS and then use data to optimize future CPU design. In this way, when the next version of the device comes out, it will better support what most people do on iPhone. Xinpi said:
This strategy is not only applicable to CPU. The same performance-power ratio rule applies to machine learning and graphics processing. For example, if a developer who develops an iPhone camera software sees a high GPU utilization, he can work with GPU architects to find a better way to work. This will bring more effective design for future graphics chips.
[化] synergistic effect
So what happens when the A13 bionic chip starts to work? Usually it involves allocation, delegation and transfer. For low-energy tasks such as opening and reading e-mail, the iPhone will use a more efficient kernel. But for more energy intensive tasks, such as loading complex web pages, this will be the responsibility of the high-performance kernel. For some routine and machine learning jobs that are already competent, the nerve engine can do it alone. CPU and its special machine learning accelerator can help to update and more sophisticated machine learning models.
Apple's secret, however, is that all these different parts of the chip can work together in a more energy-efficient way. In a typical smartphone chip, some parts of the chip are opened to perform specific tasks. You can think of it as turning on the power to the whole community, letting them have dinner, watching Game of Thrones, then turning off the power, and then turning on the power to another community that wants to play video games.
For the A13, when doing the same thing, it is based on precise control of a single family, for which less electricity is wasted. Schiller said:
Ultimately, the development of this technology depends on what we want from mobile phones, such as playing games that run smoothly on mobile phones, or taking beautiful and clean pictures in the dark. As we tap and slide mobile screens, Apple engineers are paying attention to and rethinking their designs and working on more advanced chips next year, which will tempt us to upgrade our devices. (small)<#comment>#comment><#comment>#comment>