As Huawei CEO, Yu Chengdong announced this is the first built-in independent NPU (Neural network unit) AI mobile computing platform in the world that should bring the most intelligent mobile experience to users. He also said mobile devices want to be further developed, therefore AI should not only allow them to use their whole power but also use cloud technology for bringing more massive information and services. That’s why Huawei’s current strategic layout is to focus on collaboration AI with cloud computing. The development of end-side intelligence will enhance user’s overall experience.
In the era of AI, more intelligent mobile devices will become users’ personal assistant. And as every user has different habits when interacting with a phone, AI will learn them to provide more accurate information and services. The AI will know you, understand you and help you. In the most cases, it will take the initiative to perceive the user’s state and the surrounding environment as well. Therefore, it will be more useful when providing information or services to end-users.
To reach such a state the AI will use too much information. Thus it should collaborate with cloud technologies to store date there. As Yu Chengdong mentioned Mobile AI = On-Device AI + Cloud AI. During recent years cloud storages have been used widely. But there are still many issues preventing manufacturers to use it more efficiently. Those problems mainly refer to the real-time performance, stability, privacy, etc. The Kirin 970 and the AI it uses are going to showcase how cloud technologies can be used for better purposes.
AI is the core technology of the massive data processing. However, the current SoC architecture (CPU, GPU, and DSP) that is known as the core of the traditional computing architecture has been unable to adapt to the computing performance requirements of AI. But the mobile industry doesn’t stop. The Kirin 970 is the best example of that. Unlike the server-side AI design, the Kirin 970 chose a highly energy-efficient heterogeneous computing architecture to dramatically increase AI’s computing power to respond to the data center. Based on the achievements of past-gen chips, the Kirin 970 development team has integrated the HiU mobile computing architecture with the dedicated hardware processing unit named the NPU (Neural Processing Unit). As a result, they got an AI performance much better than that seen on the CPU and GPU. Compared to the quad-core Cortex-A73 architecture, the new heterogeneous computing architecture provides about 50 times more efficiency and 25 times better performance when dealing with the same AI application. Thus the Kirin 970 chip can do any high-energy AI calculation task. For example, it provides an image recognition speed of up to about 2000 / min, which is much higher than the industry level.
As for the Kirin 970 architecture, it comes with an octa-core design 4xA73 2.4GHz +4 xA53 clocked at 1.8GHz and paired with a Mali-G72 MP12 GPU. As the name suggests, there are 12 cores that provide 20% improved performance, 50% energy efficiency, and support a longer time for 3D large-scale game running. It has a built-in upgraded self-research dual-camera ISP supporting AI scene recognition, face recognition, sports scene detection, while night shooting performance has been improved as well. Plus, the video support reached to HDR10 for the first time, and it also supports support 4K @ 60fps video decoding, and 4K @ 30fps video encoding. All these improvements have become possible due to the 10nm process technology and 5.5 billion transistors (Snapdragon 835 has 3.1 billion transistors, while Apple’s A10 comes with 3.3 billion transistors).
At last, the Kirin 970 uses more advanced 4.5G LTE technology to support the world’s highest LTE Cat.18 connectivity and to achieve the industry’s highest 1.2Gbps download rate. There are also built-in TEE and inSE security engines with higher security options.
Well, we already got the Kirin 970 is going to beat all of its competitors and the AI it uses will help it in this deal. But what about the first smartphone with the Kirin 970 on board? What about the Huawei Mate 10? It’s said this device will feature a bezel-less screen at 6.2 inches (or 6 inches) with 90% screen-to-body ratio, 4000mAh battery, and 20+12MP dual-camera with Leica technology upgraded to the third generation and supporting laser focus. Moreover, it should run on EMUI 6.0 based on Android 8.0. Let’s cross fingers, but it’s rumored the Huawei Mate 10 will support face recognition like the upcoming iPhone 8. The price of the Mate 10 should start at 4000 yuan ($610).