Edge AI applications are on the eve of large-scale landing. The huge IoT market and the violent interactions generated by revolutionary AI technologies will bring unprecedented application revolutions and business opportunities. So what are the bottlenecks in deploying AI applications on edge devices?
â—Some people have ready-made chips and application scenarios, but they are troubled by the lack of algorithms and platforms.
â—Some people have their own algorithms, but they are troubled by the lack of a useful embedded cross-platform framework.
â—Some people have their own algorithms and hardware platforms, but they are troubled by the limited computing power of the embedded platform.
OPEN AI LAB saw the pain points in the industry and launched an AI reasoning framework specially designed for embedded platforms-Tengine in response to market demand.
Neural network inference engine optimized for Arm embedded devices
OPEN AI LAB was established in December 2016 and was initiated by Arm China's joint industry partners. It is committed to promoting in-depth cooperation in the entire industrial chain of chips, hardware, algorithms, and software, and accelerating the deployment of artificial intelligence industrial applications and the expansion of application scenarios. Tengine is a lightweight modular high-performance neural network inference engine, optimized for Arm embedded devices. It perfectly supports the Arm platform, supports Arm Cortex CPU, Arm Mali GPU, Arm DLA, third-party DSP, and can be called the "all-rounder" compatible.
Developers can use Tengine to get all the mainstream framework models and embedded operating systems, as well as all kinds of hardware in the heterogeneous scheduling platform, to make full use of the hardware computing power. In addition, Tengine also provides common AI application algorithms, including image detection, face recognition, voice recognition, etc. It doesn't matter if you don't understand AI, you can run AI applications when you get started. Tengine also supports various common convolutional neural networks, including SqueezeNet, MobileNet, AlexNet, ResNet, etc., and supports optimization strategies such as layer fusion and 8-bit quantization. And by calling the HCL library optimized for different CPU micro-architectures, the performance of Arm CPU is fully explored.
Tengine's "engine"-HCL calculation library
Some people say that NCNN is a high-performance neural network forward computing framework optimized for mobile phones. From the beginning of the design, the deployment and use of mobile phones are deeply considered. In the past, NCNN was indeed like this, but there was no Tengine at that time.
Now, as a plug-in of Tengine, HCL calculation library is the fastest Arm CPU NN calculation library, and supports dynamic 8-bit quantization calculation, the accuracy is almost unchanged, compared with NCNN, it can bring 2-3 times performance improvement at the fastest , The memory usage is reduced to one third.
*Tengine performance data A72, A53 performance is based on RK3399 platform test, A17 performance is based on RK3288 platform test
NCNN data is September 7, 2018
HCL INT8 quantitative calculation accuracy change
*Using the ILSVRC2012_VAL data set to test 5000 pictures, the quantization accuracy test results are as follows
Tengine + HCL, tailor-made for different groups of people
As an AI application framework, Tengine not only provides reasoning API interfaces for upper-level applications, supports mainstream models and formats, but also defines hardware interfaces at lower layers to adapt to various computing modules of embedded platforms. In addition, it supports simultaneous operation of multiple AI application algorithms and calls. Calculation module calculation. Therefore, the launch of Tengine is a milestone for the AI ​​open source ecosystem. With Tengine, application and algorithm manufacturers can get rid of complex embedded platform adaptation work, and cooperate with HCL to directly tap the full potential of the hardware; with Tengine, chip manufacturers can stand on the shoulders of Arm's embedded AI ecosystem and achieve the fastest Speed ​​smoothly to bring the chip to the market and deploy it in edge AI applications. As the two major tools developed by OPEN AI LAB for the deployment of edge AI applications, Tengine + HCL will help the embedded artificial intelligence industry accelerate the advancement and contribute to the ultimate realization of the intelligence of all things.
Toolkits For Cutting Mahine and Cutting Materials
It is suitable for the blade of the Screen Protector Cutting Machine and the tools for installing the Screen Protection Film.
If you want to learn more about Accesseries For Cutter,Screen Protector Cleaning Tool, Cutting Blade, Cell Phone Scraper Tool, Tool Kit, Cutting Head Parts please click "Product Details" to view Accesseries For Cutter,Screen Protector Cleaning Tool, Cutting Blade, Cell Phone Scraper Tool, Tool Kit, Cutting Head Parts parameters, models, pictures, prices and other information .
Whether you are a group or an individual, we will try our best to provide you with accurate and comprehensive information about Accesseries For Cutter,Screen Protector Cleaning Tool, Cutting Blade, Cell Phone Scraper Tool, Tool Kit, Cutting Head Parts!Accesseries For Cutter,Screen Protector Cleaning Tool, Cutting Blade, Cell Phone Scraper Tool, Tool Kit, Cutting Head Parts
Shenzhen Jianjiantong Technology Co., Ltd. , https://www.hydrogelprotectivefilm.com