Cambrian released the first AI cloud chip and the third generation terminal processor.

category:Internet
 Cambrian released the first AI cloud chip and the third generation terminal processor.


The release of the third generation terminal AI chip Cambrian 1M At the Cambrian era, the third generation machine learning terminal processor 1M was released. Its performance is 10 times higher than that of the previous Cambrian 1A. In terms of configuration, Cambrian 1M is produced by TSMC 7Nm process, and its 8 bit operation efficiency is 5Tops/watt (5 trillion times per watt). Cambrian provides a processor core of three sizes, 2Tops, 4Tops, and 8Tops, to meet the needs of different levels of intelligent processing in different scenarios. Cambrian is known to be able to further improve processing efficiency through multi-core interconnections. The Cambrian 1M processor supports a variety of deep learning models, such as CNN, RNN, SOM, and so on, as well as the last two generations of chips. It also increases the acceleration of classical learning algorithms such as SVM, k-NN, k-Means, decision tree and so on. Cambrian founder Chen Tianshi said Cambrian 1M is the first intelligent processor in the world to support local machine learning and training. It can provide an efficient computing platform for visual, voice, Natural Language Processing and other tasks. It can also be used in different fields, such as the Yu Zhineng phone, the smart sound box, the camera head, and the autopilot. Launch the first cloud AI chip MLU100 In addition to the layout of terminal chips, Cambrian also released CambriconMLU100 cloud AI chip and cloud intelligent computing card (PCIe interface). It is reported that MLU100 uses the latest Cambrian MLUv01 architecture and TCP 16nm process, which can work in two different modes of balance mode (main frequency 1Ghz) and high performance mode (main frequency 1.3GHz). The equivalent theoretical peak velocity can reach 128 trillion fixed-point operations and 166 trillion and 400 billion fixed-point operations respectively, and the power consumption is 80W and 1. 10W. The Cambrian indicates that MLU100 cloud chip is also highly versatile and supports all kinds of deep learning and commonly used machine learning algorithms. Chen Tianshi claims that MLU100s power consumption is only a fraction of its equivalent, and will set a new benchmark for the industry. At the press conference, some partners also demonstrated some applications based on Cambrian chips. Lenovo launched the server ThinkSystemSR650 based on CambriconMLU100. Zhongke Shuguang also launched CambriconMLU100 based server product line PHANERON. At the press conference, HKUST also disclosed a deep cooperation research and development project with Cambrian. Cambrian has made up the key defects of Chinese chips It is pointed out that the Cambrian plays a very important role in the field of AI computing in China. It is completely owned by the Cambrian, which can be used as a common platform for AI computing, which fits the core development needs of China. In addition, in the conference, Cambrian also put forward the idea of end cloud collaboration, that is, the MLU100 cloud chip can be adapted to the Cambrian 1A/1H/1M series terminal processor and collaborate with the complex intelligent processing task. Pay attention to NetEase intelligent public address (smartman163), to interpret the big events, new ideas and new applications of large companies in AI field. Source: NetEase intelligent responsible editor: Wang Chao _NT4133