A new year means a new surprise, this week, the Graphics processor manufacturer Nvidia utilized CES 2016’s giant occasion to present its Nvidia Drive PX 2, charged by Jen-Hsun Huang as the world’s first in-car AI supercomputer for self-driving autos.
Additionally, Huang introduced another company, Volvo who will be the shareholder with Nvidia Drive PX 2 for the first time.
As you might have figured from that presentation, the Drive PX 2, redesigned from a year ago’s first model, is intended to be the onboard supercomputer that offers self-driving autos some assistance with navigating the streets — and shows signs of improvement after some time, utilizing a buzzy profound learning innovation.
More importantly, the function would be the “Brain” that conducts the self- driving autos, to self-drive. To do that it should be incredibly capable. What’s more, Huang says this new auto supercomputer is up to the test. The computational ability of the Drive PX 2 is generally the same as 150 Macbook Pros, according to Huang.
The Drive PX 2 has 12 CPU centers, fit for 8 teraflops of processing power and 24 profound learning TOPS, Huang says. Which is just industry language to say that it’s awesomely powerful. For sure, it needs liquid cooling so hopefully, it doesn’t overheat.
All of this will be inside a casing the size of a school lunchbox, according to Huang.It’s a crucial development, he says, on the grounds that “self-driving is hard.” Whether you take the Tesla methodology, of utilizing self-driving innovation as an assistive device, or the Google course of making the auto absolutely self-driving, you require a great deal of processing force in the background.
You need a considerable measure of information to securely ship individuals around, Huang says. You require uncommon handling energy to “read” the street and ensure that these self-driving autos show signs of improvement the more mileage they log.
The thought, proposed by Huang, is an “Internet of Cars,” as every Drive PX 2-prepared auto would contribute its information back to a mothership, that would help each other Nvidia-fueled auto gain from the aggregate experience of the crowd. Nvidia is calling that work of self-enhancing “brains” the “Drivenet.”
As the main supplier of graphical equipment for gamers and scientists alike, Nvidia has a great deal of ability in building frameworks that a machine can understand from video input.