Tesla reveals its new computer that will be responsible for running the Autopilot and self-driving AI in electric vehicles. The car manufacturer said that it will be the succeeding Dojo computer of the EV manufacturer.
Since we know that supercomputers are oftentimes used in large-scale computing, the California-based company will use its creation this time to provide support for automobiles in terms of their software capabilities.
Tesla SuperComputer to Run Self-Driving Software
Tesla's powerful supercomputer
According to Fox Business, Tesla will be testing the supercomputer to enable the FSD feature of the EVs. The automaker will be incorporating it through neural nets which would house a huge space for data for storage.
Earlier, the Dojo project of Tesla was hinted at by its CEO Elon Musk. The tycoon said that it would contain a supercomputer for a breakthrough. It has an exaFLOP or one quintillion (1018) floating-point operations per second (1,000 petaFLOPS). This makes it the fifth most extraordinary computer in the world.
Over the last few years, Musk has been finding ways to launch his plans in line with the Dojo project. Now, we are seeing that it is slowly coming to fruition this 2021.
As per Tesla's head of AI, Andrej Karpathy, there were previously made supercomputers intended for Dojo. During the 2021 Conference on Computer Vision and Pattern Recognition, the company introduced one of them.
Moreover, Karparthy said that the team behind the supercomputer has worked so hard for the project. The supercomputer later became one of the world's top supercomputers--an amazing feat for the EV creator.
Read Also: Tesla Invests Another $3.1 Million for the Development of EV Batteries in Canada--Here Are its Goals
What Makes Tesla's Supercomputer Special?
Having one of the best supercomputers in the world, Tesla is dedicated that its specs are beyond everyone's expectations. In addition, the company has revealed the full list of the specifications for the new cluster. Here are they:
-
720 nodes of 8x A100 80GB. (5760 GPUs total)
-
1.8 EFLOPS (720 nodes * 312 TFLOPS-FP16-A100 * 8 GPU/nodes)
-
10 PB of "hot tier" NVME storage @ 1.6 TBps
-
640 Tbps of total switching capacity
"We have a neural net architecture network and we have a data set, a 1.5 petabytes data set that requires a huge amount of computing. So I wanted to give a plug to this insane supercomputer that we are building and using now. For us, computer vision is the bread and butter of what we do and what enables Autopilot," Karpathy commented in a report by Electrek on Monday, June 21.
Karpathy added that they have invested a lot for this supercomputer through coming up with the build of 720 nodes of 8x A100 from the 80GB version. He claimed that it was the number five best supercomputer worldwide.
Musk said that his company would soon allow its supercomputers available to other firms so that they could test the neural nets for a similar purpose in terms of Autopilot and Self-Driving.
Related Article: Tesla Day 2021: Elon Musk Reveals New Event to Arrive Next Month, All About EVs to Recruit People
This article is owned by Tech Times
Written by Joseph Henry
