Wired
After the opening of the 5G gate and the rise of AI technology, the future of the smart world must be based on 5G+AI. Both the terminal and the cloud must accelerate towards a smart world with the Internet of Everything.
On the first day of the opening day of the 2021 World Artificial Intelligence Conference (WAIC), Meng Pu, chairman of Qualcomm China, delivered a keynote speech entitled “Continue to promote 5G+AI to work together to build a smart and interconnected future”, revealing Qualcomm’s 5G+AI technology At present, there are more than 130 mobile phones based on Snapdragon 888. Qualcomm’s AI technology has developed to the sixth generation, covering 1.5 billion terminals worldwide, and cloud AI technology is also developing rapidly.
Mobile phones become the most popular platform for AI: more than 130 mobile phones use Snapdragon 888
In the speech, Meng Pu mentioned that Qualcomm can provide a comprehensive and rich product portfolio that meets the needs of different terminal AI applications, as well as leading solutions for cloud AI reasoning.
At present, the most popular AI terminal platform is the smart phone. After years of development, Qualcomm’s AI technology has covered 1.5 billion terminals worldwide. The Snapdragon mobile platform can be said to be synonymous with an excellent AI experience.
Qualcomm’s top smartphone platform is the Snapdragon 8 series. This year’s representative is Snapdragon 888. Now more than 130 5G smartphones using Snapdragon 888 have been released or are under design, including Xiaomi, OPPO, and vivo. Leading domestic mobile phone manufacturers have launched a new generation of flagship models equipped with Snapdragon 888.
Snapdragon 888 upgrades the sixth generation of AI technology: more powerful performance and lower power consumption
Qualcomm began supporting AI technology on the Snapdragon platform in 2016. After years of development, the sixth-generation AI engine has been used on the Snapdragon 888, and it also supports the second-generation Qualcomm sensor hub.
Snapdragon 888 integrates the Hexagon 780 processor and is also the first AI fusion accelerator, including three parts, which can seamlessly share acceleration workloads:
The first is a scalar accelerator, which improves performance by 50%; the second is vector expansion, which supports more data types; the third is a tensor accelerator, which is used for matrix processing and doubles the calculation density.
Finally, under the integration of the three, Hexagon 780 achieved 26TOPS (26 trillion times) performance, energy efficiency increased by 3 times compared with the previous generation, dedicated shared memory increased by up to 16 times, and the improvement of data interaction timeliness in some use cases can be achieved. An astonishing 1000 times.
In addition to powerful performance, the AI engine of Snapdragon 888 also focuses on reducing power consumption and improving battery life. It adopts the second-generation Qualcomm sensor hub. Its integrated dedicated low-power AI processor can take advantage of situational awareness and combine with 5G New data sources such as, Wi-Fi and Bluetooth support use cases such as screen wake-up, hand-raising to turn on the screen, user activity recognition, voice event detection and other use cases to further enhance the performance of the platform.
Take the photography that is most concerned by current mobile phone users as an example. Qualcomm’s powerful AI performance will also directly participate in computational photography, including 3A processors for autofocus, auto exposure, and auto white balance, and cooperate with software vendors such as ArcSoft and Morpho. With in-depth cooperation and optimization, performance can be fully released.
Among the mobile phones using Snapdragon 888, Xiaomi 11 Ultra won DXOMark’s mobile phone camera king, up to 143 points, and the AI performance of Snapdragon 888 is indispensable.
More than mobile phones, Qualcomm AI is everywhere: the Internet of Things, cars, and the cloud
At present, smart phones are the most popular AI platform, but AI is not only applicable to mobile phones. In the future, AI coverage will be ubiquitous. The Internet of Things, automobiles, and the cloud will also require AI computing power.
Meng Pu mentioned that in the field of Internet of Things, Qualcomm provides Internet of Things solutions to nearly 13,000 customers around the world. Qualcomm’s AI empowers a wider range of Internet of Things scenarios for robotics, smart manufacturing, smart cities, smart retail and other industries. Bring new value.
In the automotive field, Qualcomm’s automotive solutions have empowered more than 150 million vehicles worldwide, and its business covers four major areas-vehicle connectivity and cellular vehicle networking (C-V2X), digital cockpits, advanced driver assistance systems (ADAS) and Autonomous driving, cloud-side terminal management.
In addition, Qualcomm’s Snapdragon Ride platform can provide different levels of computing power for different autonomous driving scenarios, including 10 TOPS for automotive windshield ADAS cameras with a power consumption of less than 5 watts, to L4/L5 autonomous driving scenarios Provides more than 700 TOPS of computing power.
There is also AI in the cloud. Qualcomm’s Cloud AI 100 can support the needs of many operating environments for AI solutions, including data centers, edge devices, infrastructure, and automobiles. In these scenarios, a large number of high-density servers with outstanding performance are required to provide the AI inference performance required for the continuous operation of the above-mentioned platforms.
A single Qualcomm Cloud AI 100 PCI-E card can support up to 400 TOPS of computing power. Not long ago, Qualcomm also used multiple AI Cloud 100 PCI-E cards and partner server solutions to achieve AI reasoning breakthroughs of 10 million per second. The mark of 100 million operations.
Finally, Meng Pu pointed out that we are rapidly moving towards a world where people and everything are intelligently interconnected. The integration of multiple technologies such as ultra-fast 5G connections, high-performance low-power computing, and terminal-side AI is driving this trend, driving the development of a new generation of intelligent edge terminals and cloud computing.
The combination of 5G, AI and cloud will strongly promote innovation and economic growth, empower new business models, new services and new sources of income, and accelerate the digital transformation of many industries.
Please indicate the source for reprinting: Fast Technology
.