“Science and Technology Innovation Board Daily” reported on the 25th that Tesla’s much-watched Dojo supercomputing instruction set structure details were disclosed for the first time in history. Dojo, a supercomputer developed by Tesla, can use massive video data for “unsupervised” annotation and training. It is reported that each Dojo ExaPod integrates 120 training modules, built-in 3000 D1 chips, has more than 1 million training nodes, and has a computing power of 1.1EFLOP* (petaflops per second). In terms of microarchitecture, each Dojo node has a core and is a full-fledged computer with CPU-specific memory and I/O interfaces.
media coverage
Titanium Media IT Home CNBeta Sina Technology
event tracking
- 2022-08-25Tesla Dojo project first public
- 2021-10-08Tesla headquarters “moved” to Texas, plans to sell 20 million cars by 2030
- 2021-08-20Tesla releases AI training Dojo D1 chip, Tesla robot
- 2021-07-28 Unaffected by negative news, Tesla’s sales in China account for half of the world’s
- 2021-06-22Tesla shows the prototype of self-driving supercomputing Dojo, ranking in the top five in the world
This article is reproduced from: https://readhub.cn/topic/8iLBSl7Zs7i
This site is for inclusion only, and the copyright belongs to the original author.