Nvidia CEO Jen-Hsun Huang Demonstrates Jetson Xavier Chip at 2018 Taipei Computer Show (Claire Reilly / Cnet)
It can be used for driving, flying, swimming, booing, etc., or helping to pick strawberries, lettuce and apples. Or screw it in the lab -- just like Jarvis in Marvel.
The Jetson Xavier includes six processors that perform 30 trillion operations per second.In order to build it, Nvidia spent a total of five years - three years of design, two years of construction, a total of more than 8,000 people involved in the design andDevelopment.
This is the longest processor project we have ever done on our own. It has the same processing power as a workstation equipped with a $100,000 GPU, but it has a power of only 30 watts.
In short, it was Nvidia that successfully reduced the 1000-watt performance of the $100,000 workstation to the 30-watt ‘Baby’.Of course, this is not the first time that Nvidia has entered the field of artificial intelligence (AI) chips.
Currently, Nvidia's technology has been used by some companies for autonomous warehouse robots, machine picking, and agricultural robots. However, Huang Renxun said that the Jetson Xavier is 20 times more powerful than the previous AI chip.
In addition to this hot news, Huang also showed reporters two other AI processors:
Drive Pegasus focuses on deep learning applications, performance 280 T-Flops for pure self-driving taxis, and Drive Xavier for semi-autonomous driving cars.
It is reported that Jetson Xavier will ship to developers in August this year and sells for 1299 US dollars (about 8339 RMB).