Meta unveils a new AI supercomputer

Meta (formerly known as Facebook) has unveiled a new AI supercomputer. The company claims it will be the fastest and world's largest AI supercomputer it's fully ready by the end of 2022.

Meta calls the new AI supercomputer, AI Research SuperCluster (RSC). The company says it is already being used by Meta researchers to train large models of Natural Language Processing (NLP) and computer vision models for research. "AI can currently perform tasks like translating text between languages and helping identify potentially harmful content, but developing the next generation of AI will require powerful supercomputers capable of quintillions of operations per second," says Meta.

RSC will help Meta build better AI models that will be trained by trillions of examples, work across hundreds of different languages, seamlessly analyze text, images, and video together, be used for AR, and much more.

"We hope RSC will help us build entirely new AI systems that can, for example, power real-time voice translations to large groups of people, each speaking a different language, so they can seamlessly collaborate on a research project or play an AR game together" — Kevin Lee and software engineer Shubho Sengupta, Meta

meta ai supercomputer

Source: Meta

The current generation of RSC supercomputers has 760 NVIDIA DGX A100 systems with a total of 6,080 GPUs. Meta believes that the current generation is already among the fastest supercomputers in the world. Based on the early benchmarks, Meta says that the current RSC can run computer vision models up to 20 times faster and nine times faster than the NVIDIA Collective Communication Library.

RSC is up and running today, but its development is ongoing. Once we complete phase two of building out RSC, we believe it will be the fastest AI supercomputer in the world, performing at nearly 5 exaflops of mixed-precision compute.

Throughout 2022, Meta will work to increase the number of GPUs from 6,080 to 16,000. The company says that it will increase its throughput by 2.5 times.

Source: Meta | Via: The Verge, CNN