In this context, from the mouth of NVIDIA CEO suddenly sounded enough references about the processor Tegra future generations Xavier 7 ml Transistors. The head of the company called the processor "single-processor supercomputer with the support of artificial intelligence", reminding about the ability of this solution to provide a performance level comparable to a complex DRIVE PX2, which consists of two discrete graphics processor GP106 and two mobile processor Tegra generation Parker. Consume CPU Xavier should only a small part of the electricity consumed by "road supercomputer" DRIVE PX2.
In general, each time referring to the evolution of the autopilot in cars, NVIDIA founder never tired of repeating that the problem of improving such systems - it does not recognize obstacles and the creation of artificial intelligence system that allows the car to be in a constantly changing environment. A. Chapter NVIDIA claims that Tesla Motors is ahead of its competitors in the area of the creation of the autopilot for five years. In general, the head of the first companies admitted today that it is difficult to imagine a car "medium capacity", which in five years will be deprived of any autonomous control functions. Most likely, the majority of new production cars by that time will have a sufficiently high degree of autonomy in management.
Robocar need more NVIDIA processors. For example, a fully automated taxi, as suggested, the CEO, can contain up to eight or twelve NVIDIA processors. And the fact that Tesla Motors will not work with "lidar", does not mean that the NVIDIA hardware platform will not be able to cope with them in the autopilot system of any other brand.
"Autopilot" will continue to evolve and require impressive computing resources.