Tesla has issued a new picture of its upgraded supercomputer, which is now regarded as the seventh most powerful computer in the world by GPU count.
Elon Musk, CEO of Tesla, has been sending indications about the development of a new supercomputer in-house optimized for neural net video training for years until they finally released an image. In 2020, he hinted that Tesla’s Dojo might be capable of more than an exaflop, which is one quintillion (1018) floating-point operations per second, or 1,000 petaflops.
The automaker concluded it could be better to train its computer vision neural nets internally as it was dissatisfied with the existing hardware options. With a fleet of more than a million vehicles, Tesla manages a staggering quantity of video data to train its neural nets.
Tesla introduced its Dojo supercomputer during its AI Day last year, but the company was just stepping up its efforts at that time. However, Tim Zaman, an engineering manager at Tesla who oversees the AI Infrastructure and AI Platform, has now confirmed that the GPUs in the company’s supercomputer was recently upgraded and has shared an image on Twitter to prove it.
He said, “Tesla is sponsoring the @MLSysConf, come visit our booth for opportunities on the AI team and see our hardware. We have recently upgraded our GPU supercomputer (photo) to 7360 A-100(80GB) GPUs, making it Top-7 by gpu-count. Reach out to build #1” This declaration is part of an effort to recruit for the AI team as Tesla is sponsoring the MLSys Conference.
To support its expanding artificial intelligence efforts, Tesla aspires to have the most powerful supercomputer on the planet and contribute to “general artificial intelligence” through its humanoid robot program. For now, the company is anticipated to provide additional information about its supercomputer and all other AI-related projects next month during its Tesla AI Day 2.
