Facebook Parent Meta built the world's fastest AI supercomputer

in Science & Technology2 years ago

Facebook's parent company Meta said it has created what it believes to be one of the fastest artificial intelligence supercomputers running today. They hope that this machine will help lay the groundwork for creating the Metaverse, a virtual reality construct that will replace the Internet as we know it today. Facebook says this AI supercomputer will be the world's fastest supercomputer when it's fully finished by the middle of this year.


“Meta has developed what we believe is the world’s fastest AI supercomputer,” said Meta CEO Mark Zuckerberg in a statement. “We’re calling it RSC for AI Research SuperCluster and it’ll be complete later this year.”

Supercomputer reflects large tech companies' large investments in AI research. Other tech companies like Microsoft and Nvidia have also announced their own "supercomputers with artificial intelligence". According to reports, it is slightly different from conventional supercomputers.

In a blog post, Meta engineers Kevin Lee and Shubho Sengupta stated that help create corporate AI models that can learn from “trillions of examples and work across hundreds of different languages. It will also seamlessly analyse text, images, and video together, develop new reality tools, and much more."

Work on RSC started a year and a half ago when Meta engineers designed the machine's various systems (cooling, power, networking, cables) from scratch. Phase one of RSC is already up and running and consists of 760 Nvidia GGX A100 systems containing 6,080 connected GPUs. Meta says it`s already providing up to 20 times improved performance on its standard machine vision research tasks. However, by the end of 2022, the second phase of the RSC will be completed.

At that time, it contained about 16,000 GPUs and could train AI systems "with over a trillion parameters on exabyte-sized data sets."

All these figures are very impressive, but they raise the question of what is an artificial intelligence supercomputer and its comparison to what we commonly call supercomputers?

Comment your thoughts about it