In 2022, Tesla’s D1 processor will begin production, assisting in artificial intelligence training. Tesla requires artificial intelligence technology to develop advanced self-driving cars, despite Elon Musk’s frequent statements that he views the future as bleak. As a result, the company has created a processor that fulfills this need.
According to Tom’s Hardware, the D1 chip was inspired by Tesla’s Dojo supercomputer, used to train AI models within the company. With 7nm production technology, TSMC, one of the world’s largest semiconductor makers, built the unit. The insert has almost 50 billion transistors and has a total surface area of 645 square millimeters.
— Dennis Hong (@DennisHongRobot) August 3, 2021
According to Tesla, the D1 has a 362 TeraFLOP computational capacity, albeit the CPU has been tuned for a particular sort of activity. Despite this, it defeated Nvidia’s Ampere graphics processor, which has “only” 312 TFLOPs of processing power.
The chips will aid AI in learning to recognize various objects based on what Tesla’s cameras view. This could improve the precision of self-driving vehicles and help them spread. In 2022, the first servers using D1 processors will be available.