Tesla’s D1 processor is scheduled to start work in 2022, helping to train artificial intelligence.
Although Elon Musk has repeatedly stressed that he sees the future darkly in artificial intelligence- Tesla needs the technology to make advanced self-driving cars. Therefore, the company has now developed a processor that serves this purpose.
Tom’s Hardware reports that the D1 chip was modelled on a supercomputer called Dojo, which Tesla uses to teach AI models within the company. The unit was created by TSMC, one of the world’s largest semiconductor manufacturers, with 7nm production technology. More than 50 billion transistors have been placed on the insert, with a total surface area of 645 square millimetres.
According to Tesla, the D1 has 362 TeraFLOP computing capacity – although the company has optimized the processor only to perform a certain type of task. At the same time, it still managed to defeat Nvidia’s Ampere graphics processor, which has “only” 312 TFLOP power.
— Dennis Hong (@DennisHongRobot) August 3, 2021
The chips will help train AI to detect different objects based on the information tesla’s cameras see. This can make self-driving vehicles more precise and contribute to their spread. The first servers with D1 processors will be ready to go in 2022.